The numbers behind the work

The DOJ withdrew its 2022 web accessibility guidance in 2025.

Not supported. The live ada.gov page still serves the guidance, dated March 18, 2022. Its only banner says it does not reflect the April 2024 state and local government rule. A law-firm summary (Mondaq) of DOJ's March 19, 2025 withdrawal of 11 Title III guidance documents lists COVID Q&As, retail, lodging, gas station and customer-service guides. The web guidance is not among them. I did not open the DOJ's own withdrawal notice, so this rests on a secondary source plus the live page.

UsableNet counted 4,605 digital accessibility lawsuits in 2023: 3,086 federal and 1,519 state.

Seen only in search summaries of secondary sites. UsableNet's own 2023 blog says only 'over 4,500', and the report itself is gated.

In 2024 UsableNet counted 2,400 federal and 1,600 state digital ADA suits, and small businesses were the biggest target.

Seen only in a search snippet attributed to the 2024 PDF, which I could not read. UsableNet's 2024 blog gives 60%/40% of 4,000+ and has no small-business or revenue figure.

Share of 2023 lawsuits against companies under $25M revenue.

Not on any UsableNet page I opened for 2023.

eCommerce was 69% or 70% of 2025 lawsuits (exact report table value).

The blog says 'nearly 70 percent'. A search summary said 70% in the report's table, but I could not read the PDF.

Share of BEC sent from free webmail in Q2 2026 (the latest quarter)

I read the whole Q2 2026 report (13 pages) and it gives no webmail-versus-own-domain split. The newest published figure is Q1 2026 (72%), which is listed as a fact.

Median wire transfer amount requested in BEC, Q1 or Q2 2026

Both 2026 reports give only the average ($42,663 in Q1, $61,732 in Q2). Neither has a median.

Q3 2026 APWG Phishing Activity Trends Report

docs.apwg.org/reports/apwg_trends_report_q3_2026.pdf returned 404 on 2026-10-17. Q2 2026 (published 28 August 2026) is the latest edition.

Google completed the switch of all sites to mobile-first indexing in September 2020 (or March 2021).

The March 5, 2020 post says Google will switch all websites 'starting September 2020'. The later delay to March 2021 appeared only in secondary coverage (Ahrefs). I didn't open a Google primary source for that delay. The confirmed completion date is the October 31, 2023 post.

Googlebot reads only the first 15 MB of a page.

The current Googlebot page (updated 2026-02-03) says 2MB for supported file types and 64MB for PDFs. It does not say 15MB. Don't quote 15MB.

Romance scam median loss is $6,099 and employment scam median loss is $1,500 (as quoted in some press coverage of the 2025 report)

In the 2025 report these are the 2024 figures (Table 1, 2024 column). The 2025 figures are romance/friendship $4,500 and employment $1,000. Articles that give the old numbers are mixing up the years.

Business impersonation is among the riskiest scam types

The report has no separate 'business impersonation' category for consumers. The closest is 'retail business imposter', which is outside the top 10 by overall risk (index 9.4) and ranks No. 10 by exposure and No. 10 by susceptibility. For scams reported by businesses, business email compromise ranks No. 3 riskiest.

BLS says 20% of new businesses fail in year one, 50% by year five and 65% by year ten (the usual summary of these tables).

BLS's own Table 7 gives cohort-specific rates (77.9% / 51.4% / 34.7% for the latest cohorts), and they roughly match this summary. But I found no sentence on bls.gov stating the rounded 20/50/65 wording. The BDM entrepreneurship page (bls.gov/bdm/entrepreneurship/entrepreneurship.htm) is dated April 28, 2016, uses data through March 2015, and quotes no 1/5/10-year figures. Use the table figures, not the rounded wording.

64% of patients will book (or booked) appointments online by 2019 / self-scheduling will be 38–40% of appointments (Accenture)

This is a 2014 Accenture projection that blogs (Phreesia, eWEEK) now repeat as if it had happened. They don't agree with each other, and I never opened the original 2014 report.

68% of patients want to schedule, change or cancel appointments online (Accenture 2019)

Only seen in a RingCentral blog. The survey PDF would not open (see blocked). The U.S. press release I did open gives 40% of millennials vs 19% of boomers instead.

Online bookings have 1.8% no-shows vs 5.9% for phone bookings; self-scheduling cuts no-shows 29% (Kyruus)

Only on vendor marketing blogs (NextPatient, Simbo AI) with no primary source or method shown.

Online self-scheduling reduces no-shows (as a general rule)

The newer, controlled Mayo Clinic and orthopedic studies found no significant drop, and the mammogram study found slightly more no-shows. The big drops (Tricare, Murry Hill, ZocDoc, Dartmouth) are older single-site reports gathered in Zhao 2017, not controlled trials.

82% of consumers have read a fake review in the last year (92% of 18-34-year-olds)

This appears only on BrightLocal's undated 'why are fake reviews a problem' guide, which links to the survey's #fake-reviews anchor. Neither the 2025 nor the 2026 survey page contains it, so it probably comes from an older edition (around 2019-2020). It is not current.

24% of consumers were confident they spotted a fake review in 2025 (vs 19% in 2024)

This is on the 2025 page, but the sentence says 'less consumers are confident' while the numbers show a rise. The page contradicts itself, so it is not quotable.

29% of consumers wrote a review in 2025 (vs 27% in 2024)

This is on the 2025 page, but the sentence says 'slightly fewer' while the numbers rise. It also clashes with the 2026 page's 69%, so it is not quotable.

Share of people who think they have seen a fake review (2026)

The 2026 page has no figure on how many people have encountered fake reviews. It only gives views on punishment (97%) and platform responsibility (63%).

Wix, Squarespace or GoDaddy charge a fee (or state 'no fee') to transfer a domain away.

None of the five transfer-out, auth-code or lock pages from the three companies mentions any fee for leaving, for either side. Squarespace's Domain transfers FAQ talks about a fee only for transfers into Squarespace. The gaining registrar's fee (usually a year's renewal) is not on these pages. Saying 'free to leave' would be an inference, not a quote.

Wix refunds unused domain time when you transfer away.

Wix's domain refund page says domain subscriptions are non-refundable and does not mention transfers or unused time. No page says either way.

Squarespace's contact-change lock lasts 60 days.

The transfer-away page names the 'contact update lock' but, in what I fetched, does not give its length. Only the registration and transfer locks carry '60 day' in the text. The 60-day contact-change figure in the search summary did not show up in my fetch.

ICANN's 60-day lock after new registration (from the ICANN FAQ).

The ICANN FAQ text I fetched quotes the 60-day rule after a registrant change, and says registrars 'have the option of denying' a transfer within 60 days of a previous transfer. It did not show an exact sentence on new registrations, so that is left out here. Wix's and GoDaddy's own pages cover it.

GoDaddy Website Builder currently costs $9.99 (Basic), $14.99 (Premium) and $20.99 (Commerce) a month on annual billing

This only appears in third-party search snippets (a playcode.io guide, review sites). The playcode.io page I opened does not mention GoDaddy at all. GoDaddy's own page returns 403. No primary source.

GoDaddy Basic renews at $12.99/mo, or at $16.99/mo, with Premium at $29.99 and Commerce at $34.99

These are conflicting figures from review and aggregator sites (websitebuilderexpert, recurdash, whtop), seen only as snippets. One tracker's monthly renewal prices don't match its own yearly totals. None of them is a primary source.

GoDaddy Website Builder price in 2017 and 2025

I found the capture timestamps through archive.org's availability service, but could not open the captures (listed under blocked).

Squarespace Personal and Business plan prices in 2016, 2019, 2022 and 2025

The Wayback Machine could not be fetched, and the help center's legacy plans article lists no prices for Personal or Business. Prices from memory or blogs were not used.

In 2026 a legacy Business plan renewal went from $276 to $348 (or $331) a year

Found only in Squarespace community forum posts that contradict each other. The official legacy plans article confirms that price increases happen but gives no amounts.

Wix's cheapest ad-free plan and mid-tier plan prices in about 2016, 2019, 2022 and 2025, from archived pricing pages

I could not open any Wayback capture (tool blocked, API returned 429). Without the captures there is no primary record of past prices, so no price history can be built. Jacob should open the four Wayback URLs under blocked by hand and note each capture timestamp.

Current live Wix prices taken directly from wix.com/plans

The live page loaded, but the price text was garbled and served with regional tax labels. I did not record those figures. The only clean current prices come from the Wix Blog article, which has no year on its dates.

The 2024 CDC press release gives percentages for cognitive, mobility, vision and hearing disability.

I opened the July 16, 2024 press release. It gives only the overall 'more than 1 in 4' figure plus figures by age, race and Long COVID. The percentages by type come from the CDC infographic page and the DHDS dataset.

DHDS has newer state data than 2022.

Sorting the data.cdc.gov dataset by year, newest first, showed 2022 as the latest year for Indiana.

68% of businesses used AI in 2021-2022 (Census 2023 ABS)

The Census story says 'Businesses used AI at a higher rate (68.0%) than other technologies from 2021 to 2022' and then lists cloud 40.0% etc. for 2016-2020. The 2023 ABS has a 'Timing of adoption of Technologies' table, so these numbers are almost certainly the share of firms already using a technology that adopted it in that period, not the share of all firms. The page does not state the denominator. Do not quote these as adoption rates.

Share of firms using cloud, AI or digital tools by firm size, and the share reporting no use, 2022 reference year (2023 ABS)

The table exists on census.gov (TECHADOPT by Employment Size of Firm) but data.census.gov and the API were both blocked. No 'no use' share appears in any story or brief I opened.

Fewer than 6% of firms used AI-related technologies in 2017 (2018 ABS working paper)

Seen only in a search snippet. The paper's PDFs could not be read, and the census.gov abstract page has no percentages.

Digital marketing use share from the ABS

None of the pages I opened reports a digital-marketing measure. The 2023 ABS tracks AI, specialized software, robotics, cloud-based technology and specialized equipment.

21.7% of U.S. businesses used AI in July 2026, up from 17.7% in January (WeWork analysis of Census data)

Seen only in a search summary of a secondary analysis. The Census data table that would show it could not be read, and the Sept 10, 2026 BTOS release page contains no AI figures.

The latest national BTOS round shows 22.4% AI use (reference weeks May 18 to Aug 9, 2026; Keyhole Software)

Secondary blog seen only in a search summary; no Census page confirming it could be opened

About 18% of firms had adopted AI as of year-end 2025 (Federal Reserve FEDS Note, April 3, 2026)

The fetched Fed note does say 'about 18 percent of firms have adopted AI as of year-end 2025', but it is a secondary summary of BTOS. Left out in favor of the Census originals, which give the same figure.

Experts are more likely than the public to think AI would do a better job at customer service (42% vs. 19%).

Seen only in a Substack summary. Pew's HTML chapter confirms that experts are more likely to think so, but its text gives only the 'not sure' shares (16% of adults, 4% of experts). The better/worse figures are in a chart I could not read.

Customers who use live chat are 2.8 times more likely to buy (often credited to Forrester, sometimes to Baymard).

Appears only on vendor blogs. No Baymard page says it, and I found no primary source.

44% of online shoppers say live chat answers during a purchase are one of the most important features a site can have.

Vendor blogs give it with no primary source; I could not trace one.

Baymard data on how many checkouts offer live chat, or how often shoppers use it in checkout.

Baymard's own pages give only qualitative guidance (keep site-started chat off checkout, use static links to chat). I found no Baymard percentage for live chat.

Exact number of visitors a site needs to appear in Chrome UX Report data

The methodology page says 'An exact number is not disclosed'. Any specific figure from a third-party site isn't official.

Share of Chrome users who meet the requirements to be counted

The methodology page says Google doesn't publish what proportion of users meet the criteria.

Number of websites in the CrUX dataset

Neither the BigQuery docs nor the methodology page gives a count.

CISA recommends the 3-2-1-1-0 rule (one immutable/offline copy, zero errors).

only vendor blogs say so; CISA's Back Up Business Data page lists plain 3-2-1 with an off-site copy and mentions offline copies separately as a protection; the #StopRansomware Guide does not mention 3-2-1 at all

NIST says to back up through a cloud provider or a removable drive and keep it away from the office in case of fire.

seen only as a quote from author Pat Toth in 2016 MeriTalk/phys.org news coverage; the NISTIR 7621r1 PDF could not be read

NIST's Small Business Cybersecurity Corner ransomware page gives backup steps.

opened the page (updated June 16, 2026): it has no backup sentences and no statistics

Mandiant: average time-to-exploit was -1 day in 2024, across 112 vulnerabilities disclosed in 2024.

Seen only on vendor blogs (Hadrian, Upwind and others). No primary Mandiant/GTIG page found, the guessed blog URL returned 404, and the M-Trends 2026 PDF could not be read.

Under BOD 26-04, federal agencies must fix internet-exposed KEV vulnerabilities within 3 calendar days.

Comes from secondary summaries (infusionpoints.com, insideprivacy.com). The CISA directive page shows its timelines as an image, and the text only says 'three days' for entries that also need forensic triage.

Mandiant 2021–2022: 246 vulnerabilities, 62% (153) zero-days, average TTE 32 days.

Actually confirmed verbatim at https://cloud.google.com/blog/topics/threat-intelligence/time-to-exploit-trends-2021-2022/ (Sept 28, 2023). It is left out of facts only because the 2023 edition supersedes it. Exact quote: 'Mandiant Intelligence analyzed 246 vulnerabilities that were exploited between 2021 and 2022.'

Only one-third of businesses have implemented DMARC in any form (FTC 2017 staff report).

This figure appears only in a search summary of the staff-perspective PDF, which I could not read. The FTC press release I fetched gives the 'fewer than 10 percent' figure, which counts only DMARC set to reject.

73% (8 of 11) of small-business web hosts integrated SSL/TLS into site setup; 9% implement SPF and 18% DKIM by default.

From a search summary of the 2018 FTC PDF, which I could not open. The fetched press release says only 'the majority' for SSL/TLS and 'only two ... implement SPF or DKIM by default', so the search-summary breakdown is not confirmed.

Turning on MFA makes you 99% less likely to be hacked (often attributed to CISA).

Neither CISA's 'More than a Password' page (cisa.gov/MFA, now marked as archived) nor its Secure Our World MFA page shows a percentage. It says only 'significantly less likely to get hacked.'

Statistic in CISA's Secure Our World MFA page from the National Cybersecurity Alliance 'Oh, Behave!' survey.

The page cites the survey, but the fetch returned only a paraphrase of that sentence ('most regular MFA users secure banking...' but far fewer secure work or social accounts), with no exact figure. It's also a secondary citation of a nonprofit survey.

Bots now make up about 57% of web traffic and have overtaken humans (2026, Cloudflare Radar).

Only seen in news and aggregator write-ups. The Radar dashboard that would show it returned 403. The figure is a share of web page (HTML) requests, not of all traffic.

About 37% of bot traffic is malicious (attributed to Cloudflare).

Found only in a secondary summary. No Cloudflare primary source was found.

Bots were 53% of global internet traffic in 2025.

This comes from the Thales/Imperva Bad Bot Report, not Cloudflare, and was seen only through secondary coverage. Not opened.

Bots are 34.94% of all web traffic (July 2026).

This is a third-party analysis (technologychecker.io) of Radar data, not a Cloudflare publication.

A Cloudflare application security report for 2025 or 2026 with a newer overall bot share.

The search found no such post on blog.cloudflare.com. The newest one opened is the July 2024 update.

More than 5% of email analyzed by Cloudflare Email Security in 2025 was malicious, up 16% year over year.

This sentence is on Cloudflare's theNet page (https://www.cloudflare.com/the-net/year-in-review-25/, dated 2026-01-16). But the year-in-review blogs give 4.3% for 2024 and 5.6% for 2025, which is about a 30% rise, not 16%. The 2025 blog also says the share jumped from October, likely because of an improved classification system. The 'up 16%' figure is left out because Cloudflare's own sources disagree on it.

2025 or 2026 SPF/DKIM/DMARC pass rates (e.g. 'DMARC pass 88.99%, SPF 80.24%, DKIM 90.90%')

The only source found was third-party (technologychecker.io) claiming to cite Radar. The Radar dashboard returned 403, and neither year-in-review blog (2024 or 2025) mentions SPF, DKIM or DMARC. The newest primary pass rates found are from February 2024.

Share of email that is spam (overall)

The fetched pages give spam shares only for individual top-level domains (e.g. .christmas 99.8% spam or malicious, .best 69% spam in 2025). The fetch tool returned those in a paraphrased summary, so they weren't recorded. No overall spam share was found on any fetched page.

67% / 75% of consumers prefer texting with businesses about appointments and reminders (Zipwhip, EZ Texting, Avochato, Flowroute surveys)

Only seen in search snippets of texting-vendor marketing pages and trade press. I did not open the original reports or check their methods, and they are not primary sources under the rules.

Pew says adults send about 10 texts per day as a current figure

Pew's newest texting-volume data is from 2011. I found no newer Pew figure for texts per day, and the 2025 Mobile Fact Sheet has no texting statistics.

More than 259M 5G devices, which some blogs cite alongside the texting figure

The figure is in the CTIA PDF but is off-topic, so I did not include it.

79% of users scan web pages (as a figure with a stated sample size)

The 1997 'How Users Read on the Web' article gives 79% and 16% but no participant count. The companion Morkes & Nielsen page reports 15 of 19 Study 2 participants always scanned, which works out to 78.9%. That is very likely where the 79% comes from, but neither page says so, so don't quote a sample size next to '79%'.

The 2017 F-pattern update had an overall sample size

The 2017 article gives only per-heatmap counts (47 people, more than 45 people) and no total for the study, the mobile replays or the right-to-left study.

Americans exchanged nearly 2.2 trillion text and picture messages in 2024 (CTIA 2025 Annual Survey).

Seen only in secondary sources (search snippets, voxbooster and usecarly blogs, an NCSL summary). Benton's summary of the CTIA survey does not mention texts, and CTIA's own Highlights PDF could not be read. An Oregon testimony document cites the 2024 survey for "more than 2.1 trillion," which is the prior year's figure.

CTIA Principles: businesses need express written consent for marketing texts, must accept opt-outs by any reasonable means, and must send a single opt-out confirmation.

Only provider help pages and blogs (Dialpad, Telebroad, Bird, Telnyx, Flowroute) were seen. The CTIA document itself could not be read, so none of its wording is quoted. The FCC rules above cover the opt-out side from a government source.

10DLC brand vetting gives a trust score that sets how many texts per second you can send.

Seen only in carrier and provider docs in search results (Wavix, Telnyx and others), not on a fetched TCR page

Chrome UX Report docs state the 75th-percentile pass rule

Opened https://developer.chrome.com/docs/crux/methodology/tools (last updated September 9, 2025). That page has no 75th-percentile or threshold sentence, so the rule is cited from web.dev instead.

Google Search Central uses the 75th-percentile rule

The Search Central Core Web Vitals page (updated 2025-12-10) does not mention the 75th percentile. The rule only appears on the web.dev pages.

Only 2.5% of 73 million domains enforce p=reject; about 84–85% have no DMARC record (Red Sift, Dec 2025)

Vendor figure seen only in search snippets and secondary blogs; original report not opened, so its methodology was not checked

74% of US federal domains (851 of 1,144) were at p=reject by the BOD 18-01 deadline (Oct 2018); 52% in July 2018

Found only in trade-press coverage of Agari (vendor) monitoring; no CISA/DHS primary compliance report found

49% of Dutch government domains had DMARC reject (2019)

Seen on SIDN's news page, which cites the Forum Standaardisatie 2019 measurement; SIDN gives no sample, and the data is superseded by the 2025 Forum report recorded above

UK public sector domains with quarantine/reject rose from 192 to 572 in 2018

Seen only in Computer Weekly coverage of the NCSC ACD report; NCSC report not opened

dmarc.org updated DMARC statistics (Feb 2026)

The post and the stats page were opened, but neither gives counts or policy shares in text; it notes DomainTools data that 'does not cover the entire Internet'

Palisade (57.0% of top 100k publish DMARC, 63.4% enforce), PowerDMARC (US 95.8% / 49.0% reject), ShowMyIP, Guardian360 (22.4% reject), EasyDMARC

Vendor research seen only in search results and not opened; EasyDMARC skipped as instructed

State and local governments must comply by April 2026 (50,000+) and April 2027 (smaller).

Out of date. Those were the 2024 rule's original dates. A DOJ interim final rule effective April 20, 2026 (91 FR 20902) moved them to April 26, 2027 and April 26, 2028. ada.gov and the current eCFR now show only the new dates. Do not quote the 2026/2027 dates as current.

Full verbatim sentences for each exception.

The fetch tool caps quotes at about 125 characters, so only the headings and opening fragments were captured. Bracketed text in the quotes is the tool's paraphrase, not the source's wording. Jacob should copy the full sentences by hand from ada.gov or eCFR § 35.201 before quoting them as exact text.

The typical redemption/restore fee is about $X.

ICANN publishes no amount. Its policy and advisory only require registrars to post their fees. The only amount I saw on a fetched page was Network Solutions' own $99.00. Search snippets mentioned $80 (a GoDaddy UAE page) and $40 (a small registrar), but I did not open either, so they are not confirmed. No industry-wide 'typical' figure exists in the primary sources.

The IETF standard sets a 30-day redemption period and a 5-day pending-delete period.

I opened RFC 3915. It says 'The duration of each grace period is a matter of registry operational policy that is not addressed in this document.' The 30-day and 5-day figures come from ICANN's FAQ, not the RFC.

The ICANN ERRP itself sets the 5-day pending-delete period before public release.

The ERRP page covers reminders, DNS interruption, the 30-day RGP and fee posting. The 5-day PendingDelete figure is on ICANN's registrant FAQ (the fact above), not in the ERRP text. The ERRP registrant FAQ ('5 Things') does not cover pending delete.

In 2026, 78% of employees trust 'my employer', versus 64% for business, 58% for NGOs, 54% for media and 53% for government.

These numbers appeared only in search snippets and other people's summaries. Every Edelman report PDF and page that would contain them returned 403. This set of figures matches the 2025 edition closely, so the numbers may be mixed up between years. Do not use until Jacob opens the U.S. or global PDF himself.

Business is the only institution seen as both competent and ethical (2026).

Seen only in news coverage (Spotmedia) and search snippets. The Edelman press release returned 403.

58% say their employer does well at building trust; 39% say their government does well at healing divides.

Seen only in a Statista chart write-up and a search summary, not in a fetched Edelman document.

The 2026 global Trust Index is 57; the US is 47.

Seen in a search snippet from the UK report PDF, which returned 403 when fetched.

Fieldwork ran October 25 to November 16, 2025, with a margin of error of ±0.7 points.

Seen only in search snippets from the regional PDFs, which returned 403. The Edelman Smithfield page I did open gives October 23 to November 18, 2025 instead.

The FTC estimates X% of online reviews are fake.

I opened the FTC final rule on govinfo.gov (Federal Register, 22 Aug 2024; first 100,000 of about 400,000 characters). Its section on how common these practices are gives no share of fake reviews. It only says the FTC relied on statistics from platforms, researchers and surveys. The one number there is about review hiding, not fake reviews: more than 4,500 merchants were automatically publishing only 4- or 5-star reviews. I did not read the rest of the document.

The UK CMA found 11–15% of reviews are fake.

The 11–15% study was published on gov.uk by the Department for Business and Trade (25 April 2023), not by the CMA. Credit it to the UK government or the Department for Business and Trade. I did not open a CMA report that measures how many reviews are fake.

One in five (20%) Yelp reviews is fake.

This comes from press coverage (Benzinga) and, per that coverage, a Yelp spokesperson. The paper's own figure is roughly 16% of restaurant reviews filtered.

Up to 50% of online reviews are fake.

A search result tied this to a tech firm called Truth Engine, not a regulator or a peer-reviewed study. I found no primary source with a method.

90% of consumers are uncomfortable answering unidentified calls, and 78% missed an important call in the last month because they didn't answer one.

Printed in FCC 25-76 paragraph 26, but the FCC is citing a First Orion 2021 press release and Brand Impact Report. First Orion sells call-branding services. I did not open the original report or its method, so don't present this as an FCC number.

92% of consumers assume unidentified calls are fraudulent; 56% sometimes answer anyway out of fear of missing something.

Printed in FCC 25-76 paragraph 26, citing Hiya's State of the Call 2024. Hiya is a call-screening vendor and I did not open its report.

As many as 88% of business calls go unanswered.

Printed in FCC 25-76 paragraph 26, citing a May 2025 ex parte letter from TransUnion, a call-branding vendor. The method was not seen.

70% of Americans stop answering calls from numbers they don't recognize (Consumer Reports).

Seen only in a search summary. Not opened, and not an FCC or FTC source.

Unwanted calls are the FCC's top consumer complaint.

This sentence is on the FCC consumer guide, but every fetch of that page and its PDF returned 403. I saw it only in a search snippet.

The FTC received about 133,000 robocall complaints a month in FY2025, up from about 92,000 in FY2024.

This appeared in a search summary attributed to the FTC biennial report to Congress. The press release I opened gives no monthly figures, and I did not open the report PDF (https://www.ftc.gov/system/files/ftc_gov/pdf/P034305-2025-DNC-Report-to-Congress.pdf).

60% (or 63%) of financial institutions reported check fraud in the past year (Fed 2026 Risk Officer Survey).

Seen only in search snippets; the press release body and the report PDF could not be read. The two figures also disagree (60% vs 63%), so Jacob should check the PDF by hand before using either.

The share of returned checks handled by the Reserve Banks that were potentially fraudulent rose from 10.2% in 2018 to 15% in 2021.

Seen only as a quote inside a public comment letter on federalreserve.gov, which attributes it to the Fed's Dec 2025 check-services request for information. That notice itself was not opened. The Atlanta Fed figure for 2021 (15%) is confirmed above; the 2018 figure (10.2%) is not.

Check fraud has risen 385% since the COVID-19 pandemic.

It is printed in the joint Fed/OCC/FDIC notice (footnote 5), but only as a citation of a Feb 28, 2024 Treasury press release, which I did not open. It is the same kind of cited figure as the FTC and FinCEN ones above, so Jacob can add it with credit to Treasury via the joint notice if he wants.

Banks filed about 680,000 check-fraud reports in 2022, nearly double 2021.

Seen only in a search snippet, with no source named; the FinCEN source was not opened. The Atlanta Fed post gives a different, older figure: more than 350,000 check-fraud SARs in 2021, up 23% (also FinCEN-sourced, not checked).

63% of organizations had attempted or actual check fraud in 2024; 80% were targeted by payments fraud in 2023 (AFP Payments Fraud and Control Survey).

The AFP is a trade association, not federalreserve.gov or frbservices.org, and its report was not opened. It is the usual source for business-side fraud numbers, but it is outside this topic's source rule.

Federal Reserve figures on business-to-business payment fraud.

No Fed page I opened gives a fraud rate for payments between businesses. The FraudClassifier and ScamClassifier pages (fedpaymentsimprovement.org) only describe how to sort fraud into types, not how much there is. The Payments Study's 2025 release (July 2026) has no fraud data yet; it says more detail will follow.

X% of small employer firms reported difficulty reaching customers and growing sales (2025 SBCS, national)

The report web page gives the ranking only, with no number. The national figure is in the PDF chart, which could not be read. The search-snippet values for revenue-size subgroups were not seen on a fetched page.

X% of small employer firms had trouble hiring or retaining qualified staff (2025 SBCS)

The report page and the fedsmallbusiness.org Workforce topic page say only that it is among the most common challenges, with no percentage

X% of small employer firms faced at least one operational challenge (2025 SBCS)

Not on the web report page. The 91% figure from secondary summaries refers to the 2023 survey (2024 report) and was not opened or verified here.

Share of small businesses selling online / using digital tools or e-commerce (SBCS)

The web version of the 2026 report has no online-sales or digital-tools figures. Technology coverage this year is the AI module only.

Manufacturing and construction were 20% of BEC transactions in 2017 and 25% in 2018.

Seen only in search summaries and secondary coverage (CFO.com, NAFCU, Trend Micro). The FinCEN release page names the sector but gives no percentages, and the PDF could not be read.

FinCEN's 2020-2021 real estate BEC analysis found about $10.8 billion across 31,695 BEC filings, and individual homebuyers were hit disproportionately.

The figures appear only in a search snippet. The FinCEN news release (March 30, 2023) confirms only the 2020-2021 period and that nearly 88% of incidents first sent money to accounts at US banks. The PDF could not be read.

Since 2016, FinCEN received over 32,000 BEC reports involving almost $9 billion in attempted theft.

Seen only in a search snippet of the 2019 advisory PDF, which could not be read.

How often a compromised vs. a spoofed email account was used in BEC.

None of the FinCEN pages I could read gives a share. The 2016 advisory only describes both methods (a hijacked real account, or a fake look-alike address), with no percentage.

Cutting a form from 11 fields to 4 increases conversions by 120% or 160% (Imaginary Landscape)

The original is a web firm's 2008 case study on its own site. Its PDF could not be read, so the figure was seen only in a search summary. It is also a single before/after comparison over two months with no control group. Not confirmed.

Each extra form field costs X% of conversions (for example, '4% per field' credited to HubSpot)

Search results trace this only to blogs that repeat one another. No primary study or vendor report with a method was found.

Forms with 3 or fewer fields complete at 25% vs 15% for 6 or more fields (Formstack, 650,000+ submissions)

Seen only as a quote in a search result summary. The vendor report itself was not opened, and the figures are correlational.

Dropping one optional field lifts completions 26%, and 4-field forms complete at 58.4% vs 28.7% for 7 fields (Visionary Marketing 2026 'Form CRO Benchmark')

Marketing-agency benchmark seen only in search results, with no verifiable method. Not a primary source and not opened.

A test found a 15-field form beat an 11-field form by 109%, and Marketo's label test gave a 19.21% lift (MarketingExperiments / Michael Aagaard via VentureHarbour)

Seen only through a blog that cites these tests. The baseline is unclear and no primary report was opened.

web.dev / Google says fewer fields = X% more conversions

Opened web.dev 'Sign-up form best practices' and 'Payment and address form best practices' (both last updated 2020-12-09). Neither gives any measured figure on field count; they give advice only and link out to Baymard, Comscore and Zuko. The only measured Google data found is the Chrome autofill study listed under facts.

Galesic & Bosnjak 2009: 75% started the 10-minute survey vs 65% (20 min) and 62% (30 min)

These percentages appear only in secondary summaries. The OUP page shows the abstract only (full text paywalled) and the abstract gives the direction, not the numbers.

Median loss for all 2025 fraud reports (FTC Data Book)

None of the FTC pages I fetched states a 2025 median. The 2025 Data Book was not found at the expected URLs, and the testimony PDF could not be read.

Dollar losses and median losses for email, text and phone as the contact method in 2025

The April 2026 spotlight gives only each method's share of loss reports. The dollar values sit in a chart image and an unreadable PDF. A secondary search summary said phone calls were about $1.1B and texts had a 7% share, but I did not see the dollar figure on an FTC page.

2024 contact-method losses and medians, from the 2024 Data Book as the latest full edition

The 2024 Data Book PDF could not be read as text. The March 2025 press release ranks the contact methods but gives no dollar figures for them.

Imposter scams were the most-reported fraud category in 2025

Search snippets say so, but the press-release text I fetched gave only the counts (more than 1 million; nearly 1 in 3), not the words 'most reported'

Fake-review violations can cost up to $53,088 per violation.

A search summary attributes this to a December 2025 FTC warning letter. I did not open that letter or any page that shows the figure.

Full 2023 definition of 'clear and conspicuous' (16 CFR 255.0(f)), including 'unavoidable' for online disclosures.

govinfo returned only a fragment ('difficult to miss (i.e., easily noticeable) and easily understandable by ordinary consumers'), and eCFR was blocked. I did not see the full definition.

The full verbatim text of 255.2(b) on testimonials whose results are not typical.

The govinfo fetch gave only a preamble summary, not the regulatory sentence. The FTC FAQ quote about 'Results not typical' is in the facts as partial.

An FTC statistic on how many consumers read or rely on online reviews

I found no consumer-share figure on ftc.gov. I checked the 2024 final-rule press release, the 2023 proposed-rule press release, the December 2025 warning press release, the rule Q&A, the 2022 platform guide, the 2022 consumer alert 'Should we trust online reviews?', and nearly all of the Federal Register text of the final rule (the last ~3,500 characters were not read). The FTC only speaks qualitatively: the Q&A says fake reviews 'harm the many consumers relying on them to pick products and providers', and the 2025 release quotes the consumer-protection bureau director: 'As consumers increasingly depend on online reviews'. The final rule's cost analysis does contain 'the Commission assumes that consumers rely on reviews for twenty-five percent of the total revenue'. That number is a modeling assumption built on Toast restaurant-revenue figures, not a survey of consumers, so do not quote it as one.

82% of adults consult online reviews (attributed to the FTC record)

This number shows up only in a Consumer Reports comment filed with the FTC, and it is Pew 2016 data, which is already covered as local-17. It is not an FTC finding.

About 129,000 business-imposter reports and $306 million in losses in Q3 2025 (396,227 reports year-to-date)

Seen only in an OmniWatch search summary, which is secondary. The FTC dashboard and the 2025 Data Book could not be opened, so I have no 2025 business-impersonation report count from the FTC.

Fraud could have cost consumers up to $195.9 billion in 2024 once underreporting is counted

Seen only in search summaries of the FTC testimony. The testimony PDF could not be read and the press page I fetched did not show it.

Bank impersonators caused the highest losses among business impersonators in 2025

Appears in a search snippet of the June 2026 release, and a fetch summary mentioned it, but no exact fetched text was returned for that clause

Small business is the only institution with majority confidence among Democrats, Republicans and independents (2026).

This wording comes from Ground News's AI summary and a Forbes headline slug. Gallup's article, as fetched, does not give a separate partisan small-business figure. The topline crosstab does show 79% R, 62% I and 65% D, but it does not show that small business is the only institution above 50% for all three parties, so the 'only' claim is not confirmed in Gallup's words.

2025 party split for small business: 79% Republicans, 67% independents, 68% Democrats.

This appeared only in a search summary of a state trade group page (sbam.org). It is not a Gallup primary source and I did not open it.

Businesses that add photos to their Business Profiles receive 42% more requests for directions on Google Maps and 35% more clicks through to their websites (presented as a current Google help-center statement)

I opened the current help pages 6123536, 3038169 (read all of it using offsets), 7091, 6103862, 6335804, 3039617 and 7213077. None has 42%, 35% or any photo-effect sentence. The only Google source I found is the July 2016 blog post (recorded as a fact), which cites 2014 Google/Oxera data. A Vendasta article says Google removed the figures from its help pages. The 'Google and Ipsos study' version (DAC Group, Publer) has no primary source.

Google: 'Businesses with photos are more likely to receive requests for driving directions to their location, as well as clicks through to their websites, than businesses that don't have photos.'

BrightLocal quotes this, but it isn't on any current Google help page I opened. It is probably older help-page wording; the Wayback capture that might confirm it was blocked.

Listings with over 100 photos get 520% more calls / 20 times more website clicks (BrightLocal)

Out of scope: this is vendor research, not Google. It appeared only in search snippets and secondary pages, and I did not open the original study.

Businesses with photos on their Google profile get 42% more direction requests and 35% more website clicks than those without.

The current Google help page for profile photos (support.google.com/business/answer/6123536) has no such figure, only photo size rules and 'add at least three' tips. Every search result was a third party (Publer, Uberall, DAC Group, BrightLocal) crediting Google or a 'Google and Ipsos study' without a link to a Google page. No primary source was found; the archived help page may hold it (see blocked).

Businesses with complete profiles get more customer actions (photos or complete info lift directions and clicks).

The help page 'Tips to improve your local ranking' (support.google.com/business/answer/7091) has no numbers. It only says 'Businesses with complete and accurate info are more likely to show up in local search results.' That is a claim about showing up in results, not about customer actions, and the page is undated. The figure Google does print about complete profiles is already local-15.

Searches for local places without 'near me' grew 150% over two years (Think with Google, 2017).

Only search snippets showed it. Every Google URL tried redirected, returned 404, or returned unreadable PDF data (see blocked), so the figure was not seen on a fetched page.

Over 40% of leads come through the Google Business Profile.

It is on business.google.com/us/business-profile/, but it is one customer's quote (Donna Evans, She Moves), and the page itself says 'Performance statistics are specific to this advertiser and individual results may vary.' It is not a general statistic.

How many people search for businesses on Google each day or month (a current figure).

No current Google page giving a daily or monthly count of business searches was found. The closest Google figures are the 2018 (9 billion monthly connections) and 2020 (4 billion monthly connections from Search) blog posts recorded as facts. Both measure connections, not searches.

GoDaddy has a help article that answers whether a Websites + Marketing site can be exported or moved to another host

Absence only. Five web searches turned up no such article. GoDaddy's Websites + Marketing help pages that came up cover only store data: exporting products to a spreadsheet file (article 41670), downloading orders (40577) and moving a store into Websites + Marketing (41238). The other moving articles are for cPanel hosting (9201), WordPress (8997) and GoDaddy Pro (26997). None covers moving a builder site out. The Wayback Machine could not be checked (see blocked).

GoDaddy's 'View and download my website backup' article lets you download a builder site

The page (https://www.godaddy.com/help/view-and-download-my-website-backup-40585) applies to 'Website Security and Backups', not Websites + Marketing. It says 'You can select Download all, Download files, or Download database.' It never says it works for builder sites, so it does not support a builder-export claim either way.

GoDaddy's terms say the first-term price is promotional and renews at the then-current price (Websites + Marketing and domains).

The wording turned up only in search-result snippets from GoDaddy legal pages (Email Offer Terms, Free Trial Terms, Domain Name Registration Agreement). Every legal page returned 403 and the Wayback Machine cannot be fetched, so no exact quote was confirmed. The help pages that did open (Check my domain renewal price 26950, Renew my domains 19070, Turn on auto-renew 20007, Is my product set to auto-renew 31929, Manually renew 725, Compare plans 24509) only say the renewal price shows in your account or cart. None says how it compares to the first-term price.

Websites + Marketing renews at a higher price than the first term (for example Basic $9.99/mo rising to $12.99/mo).

Only third-party reviews and trackers give these figures, and they disagree with each other. GoDaddy's pricing and builder pages returned 403, and the Compare plans help page (24509) lists no prices.

GoDaddy .com renewal is $21.99.

GoDaddy's own 'Best Domain Registrars of 2026' page (data from 11/20/2025) prints $21.99 in its GoDaddy section and $22.99 in its quick comparison table. The 'GoDaddy vs Namecheap' page has the same split. Only the newer $22.99 figure (May 20, 2026) was recorded as a fact.

Google treats 410 (Gone) exactly the same as 404 (Not found).

This is printed only in the 2011 blog post: 'Currently Google treats 410 (Gone) the same as 404 (Not found).' The current HTTP status page (updated 2026-02-04) groups all 4xx errors except 429 together, but my fetch did not show a separate 404-vs-410 sentence there. Don't quote this as Google's current wording.

Sending every broken link to the home page is fine, or is good for search.

The 2011 post calls redirecting unknown URLs to the home page a soft 404 pattern that 'can have negative effects on our understanding and indexing of your site'. That is older guidance, and the current soft 404 section I fetched doesn't address home-page redirects directly, so I'm not recording it as a current fact.

Signals fully transfer to new URLs within a fixed time (e.g. 'within a year' or 'a few weeks').

Google gives no fixed transfer time. It says keeping redirects for at least 1 year 'allows Google to transfer all signals', a few weeks for most pages on small to medium sites, and 'There are no fixed crawl frequencies'. Quote those exact words rather than a promised deadline.

Searches for local businesses per month on Google (a monthly local-search volume figure)

None of the Google pages I opened gives a monthly local-search volume: the business.google.com Business Profile landing page, the business.google.com/us home page, and the help articles 10515606, 7091 and 3474122. The closest Google-stated figures are the 4 billion connections a month (2020) and 2.5 billion direct connections a month (2025) above.

Over 40% of leads come through Google Business Profile

On business.google.com/us/business-profile/ this is a quote from one customer (Donna Evans, She Moves), shown next to the disclaimer 'Performance statistics are specific to this advertiser and individual results may vary.' It is a testimonial, not a statistic.

How Google counts the 2.5 billion connections and 19.5 million businesses

The economicimpact.google methodology page does not explain how either figure was calculated. The method may be in the linked PDF report, which I did not open.

Customers send more than twice as many messages to businesses from their Business Profiles (2020)

The Google blog post 'New ways to connect with and understand your customers' (Dec 2, 2020) has this line, but the fetch only returned part of the sentence and the comparison period was unclear. I left it out.

87% of consumers use Google to find or evaluate local businesses; 93% use Google Maps; 4–7% of profile views become website clicks

These only turned up on third-party blogs (WiserNotify, Metricool, Search Endurance, Statista or BrightLocal citations). None appears on a Google page I opened, so they are not Google-stated numbers.

Google rewards first-hand experience (in those exact words, on the helpful-content page)

The current helpful-content page says "first-hand expertise", not "first-hand experience". The phrase "first-hand, life experience" was found only as a fragment in the December 2022 Search Central blog post, so that fact is marked partial.

There's nothing wrong with pages that drop after a core update

That line appeared in older versions of the core-updates page. The current page (last updated 2025-12-10) has no sentence containing 'nothing wrong', 'not a penalty' or 'reward'. It makes the point with a restaurant-list analogy instead, which we did not get as a full verbatim quote.

Trust is the most important part of E-E-A-T

The fetch tool showed the fragment "Of these aspects, trust is most important." on the helpful-content page, but it would not give the full paragraph verbatim, and the rater-guidelines PDF could not be read. Confirm by hand before quoting.

March 2024 Search Central blog post on core updates and spam policies (Chris Nelson, March 5, 2024)

Opened, but only short fragments came back: "There's no longer one signal or system used to do this" and "many pages are generated for the primary purpose of manipulating Search rankings". Not enough for a full verbatim quote.

Current (2025/2026) share of Chrome page loads over HTTPS from the Google Transparency Report (often quoted as 'about 95%' or more)

The Transparency Report charts would not render in the fetch tool. A search summary said desktop users load 'more than half' of pages over HTTPS, but that is old text and not a fetched page. No current figure was seen.

Google keyword blog, Oct 20, 2017: 64% of Chrome traffic on Android protected, 75% on ChromeOS/Mac, 71 of top 100 sites on HTTPS

Not a failure: I opened and confirmed this page (https://blog.google/topics/safety-security/say-yes-https-chrome-secures-web-one-site-time). I left it out of facts because the Feb 2018 Chromium post above gives newer figures from the same series. Use it only if a 2017 → 2018 trend is wanted.

Google has a list of 'common mobile mistakes' (the old 'Avoid common mistakes' page: blocked files, unplayable content, faulty redirects, tiny fonts, tap targets too close)

The current Google Search Central docs have no such page. The current page has a best-practices section and a troubleshooting error list (recorded above) instead. I did not open a Wayback copy of the old page.

Google retired the Mobile-Friendly Test and Mobile Usability report on December 1 or 4, 2023

Google's April 19, 2023 post says 'starting December 1, 2023' for the Mobile Usability report, but the fetch only paraphrased the part about the Mobile-Friendly Test. The actual December 4 shutdown is only in trade press (Search Engine Land), not a Google source I opened.

Businesses with photos get 42% more requests for directions and 35% more website clicks (attributed to Google).

None of the current Google help pages I opened (6103862, 6123536, 7213077, topic 6130059, 6335804) show these figures. Search results trace them to a Google/Ipsos study and say Google dropped the numbers from its help docs. I could not open the Wayback copy.

Google says businesses with photos are more likely to get direction requests and website clicks (no number).

Only secondhand sources (BrightLocal, Whitespark) and a search summary show this. It was not visible on any Google page I fetched.

Google help pages tell owners they 'should' reply to every review.

The reply help page (3474050) encourages replies ('you show them that you value their feedback') but never says owners should or must reply.

Google Safe Browsing finds about 42,000 malware sites and 26,000 phishing sites each week.

Only press coverage gives these numbers (PCWorld, Jun 25, 2013: 'nearly 42,000 malware sites per week' as of June 16, and 'roughly 26,000' phishing). I could not open the Transparency Report chart itself. The numbers are also from 2013, so they are too old to quote in any case.

Current weekly counts of unsafe sites from the Transparency Report, split into malware and phishing, with the latest date

Google's own help page (support.google.com/transparencyreport/answer/7381518) says the 'Unsafe websites detected per week' chart 'stopped being updated and were archived' in February 2021. The malware charts stopped in April 2020. The 'Warnings displayed per week' chart has shown data only through December 2022 since March 2023. So there is no current weekly figure to quote, and I could not render the archived chart to read its last data point.

GOV.UK says to write for a 9 year old reading age, because that is the average reading age in the UK.

The page the question names (gov.uk/guidance/content-design/writing-for-gov-uk) no longer exists. GOV.UK's content API returns document_type 'redirect', first_published_at 2016-02-18 and public_updated_at 2026-06-05, and it now redirects to guidance.publishing.service.gov.uk. None of the replacement pages I opened (tone-of-voice, writing-guidelines, clear-language, meet-user-needs, clear-structure) mention a reading age or age 9. The only copy I saw was a third-party change-tracking site (govdiff.njk.onl, 'Brexit guidance change explorer'). Its 2022-05-05 version reads 'This is why we tell people to write on GOV.UK for a 9 year old reading age.' and 'By age 9, you're building up your 'common words' vocabulary.' That copy is not an official source, and even it does not say 'the average reading age in the UK is 9'. That wording comes from secondary sources such as a Stormid agency blog. Archive captures were blocked.

Research shows readers understand sentences of around 25 words (old GOV.UK wording).

The retired page, as shown on the third-party govdiff mirror, said 'By using common words we can help all users understand sentences of around 25 words.' The current official page only says 'Try to split up sentences that are over 25 words long.' and cites no research for the 25-word figure. The old wording could not be checked on an official or archived copy.

The page's last updated date.

The current official guidance pages show no published, updated or reviewed date anywhere, including the footer. The old page's dates come only from the GOV.UK content API (first published 18 Feb 2016; turned into a redirect 5 Jun 2026) and from the third-party mirror (published 25 February 2016, updated 5 May 2022 in that capture). Neither is a date printed on a live official page.

The 2025 median mobile page is 2,362 KB.

This sentence is in the Page Weight chapter text: 'As of July 2025, the same median page is now 2,362 KB.' But the chapter's own Figures 14.1 and 14.7 give 2,559 KB / 2.56 MB for the same month. And 845 KB plus the 202.8% growth the chapter reports works out to about 2,559 KB. So 2,362 KB looks like a typo in the source. Don't quote it.

The median mobile page makes 72 requests (2025).

The text says 72, which is recorded as a fact above. But the fetch's summary of the Figure 14.33 chart description said the median mobile page makes 68 requests in total. Check the figure before publishing. If they still disagree, 'about 70 files' is the safe wording.

Indiana ranked 15th with 20,777 complaints in the 2025 IC3 report.

This came only from a search-engine summary of a garbled extract of the 2025 IC3 PDF. I could not open the PDF's text to confirm it. It would also mean Indiana fell from 8th (23,659 complaints, 2024 report, per local news) to 15th, so check it by hand before using it.

In 2025, Indiana saw more than $125 million in losses (FOX59).

$125.1 million is the 2024 Indiana total reported in local news coverage of the 2024 report. The FOX59 segment appears to give it the wrong year. The 2025 total sits in the unreadable PDF state table.

Indiana ranks 2nd nationally in cybercrime complaints per capita (342 per 100,000).

This is a third-party (Click Intelligence/Click Insights) analysis of 2024 IC3 data, reported by a radio station. It is not an IC3 publication, and it covers 2024, not 2025.

Americans over 60 reported about $7.7 billion in losses to the FBI in 2025 (about 201,266 complaints, up 37% or 59-60% from 2024).

This appears only in secondary coverage (search snippets, AARP, Medium and blog summaries), and those sources disagree on the growth rate. The primary PDF could not be read and fbi.gov returned 403.

Tech-support scams: 47,794 complaints and $2.1 billion in losses nationally in 2025.

Secondary summaries only; not seen in the fetched primary.

Investment fraud was the costliest category at about $8.6 billion in 2025.

Secondary summaries only; not seen in the fetched primary.

The 2025 IC3 report makes a statement about spoofed or lookalike domains.

No such text turned up in any fetched page or search result, and the PDF text could not be read. Unknown.

Indiana ranked 2nd nationally with 342 complaints per 100,000 people and over $125M lost.

This comes from a 'Click Intelligence' marketing study surfaced in search, not IC3, and it is based on 2024 data. Do not use.

People 60 and older filed 201,266 complaints and lost $7.748 billion in 2025, up 59% (complaints up 37%).

Seen only in secondary coverage (HousingWire, AARP, ELACA, others). The primary 2025 IC3 PDF could not be read as text, and fbi.gov pages returned 403.

Tech and customer support scams cost people 60+ about $1.04 billion ($1,040,730,043) in 2025, second only to investment fraud ($3.52 billion).

Seen only in news summaries. The primary PDF was unreadable.

In 2025 the average loss for people 60+ was $38,500, and 12,444 of them lost over $100,000.

Seen only in secondary summaries. The primary PDF was unreadable.

Tech support scam losses across all ages were $2,134,675,818 in 2025.

Seen only in a search-result excerpt of a mirror copy. Not seen on a fetched page.

Phishing/spoofing was the most-reported crime among people 60+ in 2025 (48,064 complaints).

Secondary sources only.

A separate '2025 IC3 Elder Fraud Report' exists.

The IC3 annual reports index lists no 2025 elder fraud report; the 2025 elder data appears to sit inside the 2025 Annual Report. The 2025_IC3ElderFraudReport.pdf URL redirected to the index page.

A .com domain gets a 5-day pending delete after the 30-day redemption period, and a 0-45 day grace period after expiry.

RFC 3915 (IETF) says the length of each grace period is 'a matter of registry operational policy'. Verisign's page would not open. The only source giving these figures was a third-party guide (domaindetails.com), which is not primary.

The auth code (TAC) now expires after 14 days, or 30 days, under the new transfer policy.

These come only from working-group proposals, public comments and a registrar blog (NameSilo). The adopted Transfer Policy pages I fetched (2016 and 21 Feb 2024 versions) still use the 'AuthInfo' code and 5-calendar-day wording. Any newer final policy from the Transfer Policy Review was not found on icann.org.

ICANN's 'Registrant Rights and Responsibilities' covers transfer codes and expiry reminders.

The current ICANN page is titled 'Registrants' Benefits and Responsibilities' and, per the fetch, does not mention auth codes or expiry reminders. Those rules sit in the Transfer Policy and the expired-domain policy (ERRP).

Harvard study: more than 70% of URLs in law journals are broken (full sentence)

The figure is on the fetched Harvard Law Review Forum page, but the fetch tool only returned the fragment 'more than 70% of the URLs within the above mentioned journals, and 50% of the URLs within U.S. Supreme Court opinions', not the whole sentence. It is folded into the Supreme Court fact's figure field and not given a separate line.

Liebler and Liebert (Yale) 2013 study on link rot in Supreme Court opinions

It came up only in search-result coverage. I did not open the paper.

Chrome HTTPS share by platform from the live Transparency Report (e.g. ChromeOS/Mac ~99%, Android ~97%, Windows ~95%, Linux ~85%)

Live chart values were not readable. Only Google's Oct 2025 blog figures could be confirmed: overall 95-99%, Linux 84%, Windows 95%, and the public-site-only numbers. Older numbers in press coverage (2016: Windows 53%, Android 42%; 2018: 68%/80%) are out of date and come from secondary sources.

Firefox share of page loads over HTTPS (global, e.g. ~80%)

No current figure was printed on any fetched page. The Let's Encrypt 10-year post gives 'around 80%' globally, without a date, and that is the closest confirmed statement. Note it measures all Firefox page loads, not just public sites.

Let's Encrypt protects one billion websites

The Dec 2025 post says 'We’re closing in on protecting one billion web sites' and 'on track to reach a billion active sites'. That is a goal, not a reached figure. The Oct 2026 data shows 730.6 million fully-qualified domains active.

Google does not use Lighthouse or PageSpeed Insights scores directly in ranking; only real-visitor (field) Core Web Vitals data counts.

None of the Google pages I opened says this outright: Search Central page experience, Search Central Core Web Vitals, Search Console Core Web Vitals help, PSI About, PSI FAQ, web.dev lab-vs-field, Lighthouse overview and Lighthouse scoring. The wording appears only in SEO blogs and vendor articles (Vercel, Search Engine Land, Core dna), which are not primary sources. The closest Google wording is the page-experience FAQ: perfect scores aren't worth chasing for SEO, good tool results don't guarantee rankings, and there is no single signal.

Core Web Vitals are only a tiebreaker in ranking (often attributed to Martin Splitt).

Seen only in a secondary search summary. Not found in any Google documentation I opened.

DMARC.org end-of-2025 policy mix (share of domains at none, quarantine or reject)

dmarc.org/stats/dmarc/ opened, but the figures are only in chart images ('DMARC policy mix at the end of 2025'). The text gives no numbers. The page warns that the DomainTools dataset 'does not cover the entire Internet' and that hundreds of thousands of policies published in late 2023 later disappeared from DNS.

2.32 million organizations adopted DMARC between February and December 2024 (Red Sift)

Seen only in a search snippet of a Red Sift vendor blog. I did not open it or check its method, and it is not a primary source for this topic.

Only 9% of domains with DMARC records actually enforce protection (DMARC Report, 2026)

Seen only in search-result metadata from a vendor blog. Not opened, no method seen. It also sits close to the do_not_use DMARC-adoption figures.

M3AAWG's training says a 100% reject policy is 'very aggressive' and recommends starting with monitor-only

Seen only in a search summary of 2012 M3AAWG training slides (Segment 3 PDF). Not opened, and 2012 guidance is outdated.

A current M3AAWG public page giving DMARC adoption statistics

No M3AAWG page I found or opened gives adoption numbers. Its 2013 release and the training-series blog post explain why DMARC matters but give no figures.

Microsoft sees X thousand password attacks per second / per day (2025 report)

Not on the 2025 launch blog post or the report web page, which now shows the 2026 edition. The PDF could not be read. Secondary coverage gives no per-day password figure for 2025.

Share of attacks that target small and medium businesses (MDDR 2025)

Neither Microsoft web page gives a percentage for SMBs. The blog only mentions Russian actors 'targeting small businesses in countries supporting Ukraine', with no number. The PDF could not be read.

Token theft is 2.4% and adversary-in-the-middle attacks are 0.003% of identity compromise attempts

Seen only in third-party summaries (flare.io and others). Not on any Microsoft page I could open.

Phishing was the entry point for 23% of investigated intrusions, up from 7% (MDDR 2026)

Seen only in a secondary summary. Not on the MDDR 2026 web page I fetched.

Only 7% of companies respond to a web lead within five minutes, and 55% take more than five business days (Drift, 433 companies).

Only third-party blogs repeat it. Drift's own report could not be found on drift.com, and the Wayback Machine is unreachable from this tool.

More than half of companies ignore lead forms (Drift 2018 Lead Response Report).

Seen only as a CEO quote in search results about a Drift press release; the report and its sample were not opened.

Calling within one minute raises conversion by 391% / 400% (Velocify).

Confirmed and listed under facts. Noted here because blogs often misattribute it to Harvard or MIT.

28% of business calls go unanswered (CallRail).

Found only on aggregator blogs citing a 2026 CallRail product announcement, with no sample or method. Not found on any CallRail page opened.

Up to 85% of callers whose calls go unanswered will not call back.

It appears in CallRail's press release with no source or method. A fact-check site says no primary source exists.

78% of consumers abandon a business after an unanswered call, and 42% leave a voicemail (CallRail 2025 consumer survey).

Seen only on secondary sites; the survey itself was not found or opened.

69% of callers don't leave a voicemail; 33% of micro businesses fail to answer their calls (Moneypenny, 2016).

The Moneypenny news page was opened and shows these figures. But it is a vendor marketing page: a commissioned survey of 300 UK micro businesses plus '10,000 businesses' call data', with no description of how calls were measured. Also UK-only.

38% of calls to small businesses go straight to voicemail, and only 82 of 200 answered (Time Etc, Birmingham).

Seen only in UK trade-press write-ups, undated; no original Time Etc report was found.

Dreamforce 2015 attendees took 37 hours 25 minutes on average to respond to a web lead (XANT/InsideSales).

Figure is on insidesales.com/dreamforce-lead-response-times, but the page gives no sample size or method beyond naming the ResponseAudit, so it is not recorded as a fact.

88% of car-dealer web inquiries got a reply within 24 hours; 69% got a phone call (Pied Piper 2017, 18,277 dealerships).

Opened only a news article reposted on Pied Piper's site, not the study itself. Car-dealer-specific.

The NCSC Small Business Guide says there are 5.5 million small organisations in the UK with 0–49 employees.

The sentence is on the NCSC page (published 9 April 2026, last reviewed 21 July 2026), but it cites no source. The government's own Business Population Estimates 2025 (DBT, 2 October 2025) give 5.64 million small businesses, which rounds to 5.6 million, not 5.5 million. Use the DBT figure if a count is needed.

NCSC Small Business Guide: '1 in 2 small businesses suffer a cyber incident every year' (50%).

It's on the NCSC page with no source given. It matches the already-verified DSIT figure (websec-15), so it adds nothing new. Don't cite it separately.

Share of UK small businesses (10–49 staff) requiring two-factor authentication, or with a formal incident response plan.

The report text gives these only for all businesses, micro, medium and large. Small-business values are probably in the ODS data tables, which weren't opened.

Exact questionnaire wording for the insurance, 2FA, training and incident-plan questions.

The technical report fetch didn't include the questionnaire text (Appendix A). It refers to these questions only by code (RULES20, TRAINED, INCIDCONTENT1).

Share of US establishments (not firms) with fewer than 10 employees, from County Business Patterns

The 2023 CBP press release (census.gov/newsroom/press-releases/2025/2023-county-business-patterns.html) gives no size breakdown. The CBP tables are on data.census.gov or in CSV downloads, which a fetch can't read. The only establishment-based figure confirmed is the Census story's 'under five employees = 55.7% (2022)'.

About 10% of US workers are self-employed (including incorporated owners)

Seen only on secondary sites (Bloomberg Tax column, Walnut Invest, Oberlo). BLS's own table A-8 shows only unincorporated self-employed, and the BLS article gives 5.7% unincorporated. The incorporated series was not confirmed on a fetched BLS page.

"Open now near me" searches grew 200% year over year (2021).

PYMNTS cites Think with Google's August 2021 Global Insights PDF, but the PDF's text could not be read. Only a news article confirms it.

"Open now" searches tripled since 2015.

Search snippets trace it to Create with Google: Search 2021, an image-only PDF. The text could not be extracted.

"Near me" mobile searches with "can I buy" or "to buy" grew over 500% in two years.

Third-party blogs repeat it. The Think with Google page now redirects to the home page and no archived copy could be fetched.

90% of shoppers search online before visiting a store / 6 in 10 check stock before visiting.

The Think with Google data pages are retired and redirect to the home page. The newer business.google.com URL returned 404.

"Near me" searches doubled, 80% from mobile (2015).

Search Engine Land returned 403 and the Google AdWords blog post body did not render.

88% of mobile "near me" searchers are on mobile / mobile "near me" searches grew 146% YoY.

Only a Search Engine Journal summary of a Lisa Gevelber piece. The original Google article was not found.

41% of small business owners call the cost of health insurance critical (down 10 points from 2020).

Seen only on a trade-press page (hcinnovationgroup.com), which attributes it to the NFIB report. The NFIB release reposted on WisBusiness has no percentage for health insurance, and the PDF could not be read. It needs checking in the PDF.

Rank and % critical for technology, cyber-attacks, online sales/internet competition, poor sales, or finding customers in the 2024 NFIB report.

None of the press-release copies, SmallBizTrends, or search results mention these items. They exist only in the full 75-item table in the PDF, which I could not extract. I also could not confirm which of these items are on the 2024 list.

Credit card payment processing costs: a new item in the 2024 survey, with a rank.

A search summary says it was added in 2024, but no fetched page gave its wording or rank.

Security Recommendation 4-11: sending domain owners who deploy SPF and/or DKIM are recommended to publish a DMARC record (NIST SP 800-177)

This wording appears only in vendor blogs (Valimail, dmarcian, PowerDMARC) quoting NIST, and those may be quoting the 2016 original rather than Rev. 1. The NIST PDF could not be text-extracted, so the wording and numbering are not confirmed against the primary document.

CISA-published adoption statistics for DMARC or p=reject among federal domains (for example 74% at reject by October 2018)

The CISA BOD 18-01 page and the cisagov/cyber.dhs.gov 18-01 guide publish no adoption figures. The only statistic there is unrelated (7 of the 10 most common federal vulnerabilities would be fixed by the directive's web actions). The 18%, 52%, 74% and 81% figures in circulation come from Agari/Fortra and Proofpoint vendor tracking, reported by TechTarget and MeriTalk. They are not CISA data and were not opened.

NIST SP 800-40 Rev. 4 cites statistics on exploitation or patching delays

The abstract and NIST's two announcement pages (csrc.nist.gov News 2022, nist.gov/node/1700231) contain no statistics. The full PDF could not be read, so I can't say whether the body cites any

CISA's 'Understanding Patches and Software Updates' gives statistics on why patching matters

Fetched: the page has no statistics, only qualitative advice

NIST Executive Summary: 'Software used for computing technologies must be maintained because there are many in the world who continuously search for and exploit flaws in software.'

Seen only in a search-result summary. The PDF could not be read, so this is not confirmed

In NN/g's 2016 trust study, 4 of 5 participants chose BoxGreen.co for monthly snack boxes

The fetch returned the sentence cut off, so the reason given for the choice wasn't visible. The overall sample size isn't stated either. Too thin to quote as a figure.

A dedicated NN/g pricing-page or plan-tier study with measured figures

Two site-restricted searches found only the 2006 and 2013 B2B pricing articles plus phone-plan pieces ("Get Started" Stops Users, iPhone plan), which have no measured figures in the search results. No dedicated pricing-page study with numbers turned up.

NN/g Contact Us research with percentages

The 2019 Contact Us study is qualitative (20 participants), and the fetched page gives no percentages.

Nielsen Norman Group recommends 50–75 characters per line (or some other ideal line length).

The 2015 Legibility article does not cover line length. An nngroup.com search found only general statements (shorter lines and enough line spacing make text easier to read) with no character count. No NN/g figure was found.

NN/g says body text should be at least 16px.

No NN/g page I opened gives a pixel minimum. NN/g's figures are 10 point (2002), 12 point for seniors (2002, 2013) and 'at least an 8-pt font' (2015, low-contrast article). The 16px figure in search results came from non-NN/g sites.

NN/g gives a minimum contrast ratio (e.g. 4.5:1).

The 2015 Legibility article and 'Low-Contrast Text Is Not the Answer' (Katie Sherwin, June 7, 2015) only say to use high contrast. The second one points to WCAG 1.4.3 but states no ratio itself.

NN/g measured an absolute reading speed for its legibility guidance (e.g. users read X% slower on low-legibility designs).

The 2015 Legibility article's '20% slower' figure is a made-up example ('say, 20% slower') showing how to measure, not a finding. Not recorded as a fact.

Mobile content is twice as hard to understand (reading on a phone halves comprehension).

NN/g's 2011 article does say this, but it rests on one 50-person lab study of privacy policies on a simulated iPhone-size screen. That article now carries an update pointing to NN/g's own 2016 study, which found no practical difference for typical web content. Don't quote it as current fact.

Per-device comprehension percentages from the 2016 NN/g study (for example, mobile X% vs desktop Y%).

The article text gives only the 3-point difference. The fetch tool reported that the per-device scores appear only in a chart image, which it could not read.

In the 2023 smartphone study, phone readers also took longer to read.

A secondary summary (aacademica.org) said so, but the paper itself says 'we decided to focus our analyses on the comprehension scores rather than on time.' The time data was used only to exclude outliers.

Text reminders cut no-shows by about 42.8% (JMIR 2024, Sides and Kbaier)

Opened at https://www.jmir.org/2024//e43894. The figure is on the page, but the paper is an uncontrolled before/after case study at one GP practice in Wrexham, UK, not a meta-analysis or a controlled trial. The authors themselves warn against generalizing it, so it is left out as too weak to quote.

Voice reminders work better than text reminders

Robotham 2016's abstract says voice notifications 'appeared more effective', but the wording is tentative and no effect size is given in the abstract. Not recorded as a quotable fact.

No-show rates by type of clinic or specialty (e.g. primary care vs specialist), from Dantas et al.

Only the abstract was readable, and it gives a breakdown by continent only. The full text on ScienceDirect returned 403. A podiatry.com PDF copy turned up in search, but it is a third-party host, so I did not use it.

Telehealth appointments have lower no-show rates than in-person ones (Greenup & Best 2025, BMC Health Services Research, PMC12063363)

Opened. It reports a pooled odds ratio (OR = 0.61, 45 studies) and no no-show percentages. It is about telehealth, not booking tools, so I left it out. It could be added if wanted.

Text-message reminders make people about 1.48 times more likely to attend (Guy et al. 2012, Health Services Research)

Opened the abstract. It does say 'The summary effect from the RCTs was 1.48 (95% CI: 1.23-1.72)', but it doesn't say which effect measure that is, so I couldn't turn it into a plain-English line without guessing.

NTIA: X% of Americans use the internet to find information about local businesses

I read the 2023 sections of NTIA's full CSV that cover online activities, devices, locations and non-use reasons. None of those variables is about finding local business information. The nearest are eCommerceUser (shopping or other consumer services) and requestingServicesUser (requesting services from other people), and both are reported above. The NTIA survey does not seem to ask this question, so no figure is given.

NTIA: X% of Americans shop online (shopping alone)

In the 2023 data, NTIA groups shopping with travel bookings and other consumer services in one variable ('Shops, Makes Travel Reservations, or Uses Other Consumer Services Online', 75.8% of internet users 15+). No 2023 figure for shopping alone appeared in the rows I read.

80% of consumers want self-service online appointment scheduling (Kyruus 2023 Care Access Benchmark Report, Wakefield Research, 1,000 consumers)

Seen only in a TechTarget news story and a PR Newswire release it links to. The report itself was not opened, and the article gives no field dates, mode or margin of error. It is a vendor-commissioned survey.

72% of patients want to book, change or cancel appointments online (Accenture Connected Health Pulse Survey, 2012, 1,110 US patients)

Seen only in search summaries of press coverage. The Accenture report was not found or opened.

Accenture: 64% of patients would book online by the end of 2019, 38% of appointments would be self-scheduled, and online booking takes under a minute versus 8.1 minutes by phone

These are 2014 forecasts, not measurements. They were seen only in trade-press summaries (HealthExec, MedCity News, eWEEK). The original report and its methodology were not found, and the 64% and 38% figures don't reconcile with each other.

MaineHealth: 68% of members would be interested in using online scheduling for primary care

Seen only in a search snippet of an annual-report story about an internal health-system survey of about 1,500 members. Not opened, and the methodology is unclear.

The OWASP 2025 data covers security testing from 2020 to 2024.

This came from a search-result summary. The fetch of the 2025 Introduction said "2020" does not appear on the page, so the data years are unconfirmed.

Broken Access Control average incidence rate 3.73%

Secondary sources say 3.73%. The fetched A01 score table shows 3.74% twice, on the live page and on the GitHub source, so use 3.74% if this figure is used at all.

53% of mobile visits are abandoned if a page takes longer than 3 seconds (2018 Think with Google page).

The 2018 benchmarks article I fetched does not show the 53% sentence; it only cites Google/SOASTA 2017 for a chart. The Think with Google stat page now redirects to a generic hub. The 2018 Chrome blog's 'more than half' wording is recorded above. The 53% figure is already local-11.

web.dev 'Why does speed matter?' repeats the 53% / 3 s and 15 s figures.

Fetched web.dev/articles/why-speed-matters (updated 2023-11-01): neither figure is on the page.

Pew 2025 internet use and home broadband by education level

The fetch returned the 2025 internet-use education values (93%, 98%, 99%) without column labels, so they can't be assigned. Broadband by education (62% / 81% / 92%) came back with uncertain headers. Not recorded.

Census ACS household broadband rate for householders 65+ and by income (2021)

The press release gives no age or income broadband rates, only that smartphone-only households were more likely to be 65+ or to earn $25,000 or less. The full report (census.gov/library/publications/2024/acs/acs-56.html) was not opened in time.

13 million more Americans used the internet in 2023 than in 2021 (NTIA); 4.2% more internet users aged 65+

Seen only in search snippets and an advocacy summary; the NTIA primary page was not opened (404 on the guessed URL).

94% of adults have some form of broadband (home or smartphone)

Seen only in a secondhand write-up in search results; not seen on a Pew page I fetched.

Recent Pew data (2019, 2024 or 2025) on where Americans get information about local restaurants or businesses

Pew's newer local news studies do not ask about this. The 2024 'Americans' Changing Relationship With Local News' report and its 'Local news topics' chapter cover weather, crime, traffic, schools, sports, the economy, government and arts. Both pages say they do not mention businesses, restaurants or events, and the 2025-data fact sheet has no business figures. The newest Pew business-information data I found is from 2011.

Pew: search engines and the internet are the top source for local business information among all adults

I only confirmed the figures among the people who look up this information (the 47%/36% and 51%/38% facts). The all-adults ranking sits in the 2011 overview, which returned 403.

NBC News (about 2012): 22% relied on word of mouth and 8% on local TV for local business information

The figures match Pew's own page, which I fetched and used instead. A press article is not a primary source.

A 2024 or 2025 Pew survey on how many Americans research products online, read reviews before buying, or compare prices on their phone in a store.

Three searches of pewresearch.org and the open web found no such Pew report. The newest Pew shopping survey found is the July 2022 short read (devices, in-person preference, influencers), and it asks nothing about reviews, research before buying, or phone use in stores. The 2023 short read is Census sales data only. 2024 articles quoting 'Pew' (e.g. a May 2024 PYMNTS piece) reuse the 2015/2016 figures.

About 4 in 10 social media users say influencers affect what they buy (Pew 2022).

This is on the fetched 2022 page ('About four-in-ten social media users (39%) say influencers or content creators impact their decisions about what to purchase at least a little, but only 3% say these groups have a *lot* of impact'). It is left out of facts only because it doesn't fit the topic. Add it if wanted.

WordPress ships three major releases every year.

That's the handbook's target and the 2026 plan, not a fixed fact. The project said in April 2025 it would ship one major release a year, and 2025 actually had two. In 2026, 7.0 came out May 20 (planned April 9) and 7.1 on August 19. 7.2 is planned for December 8 and hasn't shipped yet.

WordPress minor releases are where the security fixes go (from the Version Numbering handbook page).

The Version Numbering page I fetched never mentions security, and the sentence I got back was cut off. The April 2025 cadence post and the release notes do confirm it, so I used those instead.

W3Techs share of PHP 8 sites still on 8.0 or 8.1.

Only 8.3 (34.8% of PHP 8 sites) is written as a sentence on w3techs.com/technologies/details/pl-php/8. The rest (8.2 27.9%, 8.1 15.3%, 8.0 5.8%, 8.4 12.0%, 8.5 4.2%) appear only as chart labels, so I didn't record them as facts.

A VA letter rewrite cut calls by 90 percent.

This appears only in a 2008 op-ed by Annetta Cheek, chair of the advocacy group Center for Plain Language, hosted at governor.wa.gov/sites/default/files/2023-01/Commentary.pdf. It is a secondary source with no office or year. It probably describes the same Jackson MS letter (94 to 16 is an 83% drop, not 90%). Use the plainlanguage.gov table instead.

Arizona Department of Revenue saved $51,014 a year from avoided calls; Weights and Measures collected an extra $144,000 a year.

These figures appear only in the same Cheek op-ed (advocacy, secondary). I found no Arizona state primary source in the time available.

A Washington State Department of Revenue letter rewrite produced 95% fewer hotline calls and $5 million in extra revenue.

This appears only on a secondary site (prism.kpwashingtonresearch.org). A search turned up a 2006 Marketplace news story giving different numbers: use-tax payers tripled, about $800,000 extra over two years, and a separate 32% call drop at the Department of Licensing. The WA executive order (2008) gives no figures. No primary source was found.

VBA's annual contact letter went from a 35% to a 55% response and saved about $8 million a year.

This appears only in a secondary summary (KP Washington research page). It conflicts with plainlanguage.gov's 43% to 65% and $4.4 million. I did not record it.

The Plain Writing Act text contains measured results.

It does not. The Act sets requirements and definitions only, so it is recorded as a requirement, not as evidence.

NIH recommends writing public health information at a 6th to 8th grade (or 7th–8th grade) reading level.

The nih.gov Clear & Simple and Plain Language pages returned 403. The only NIH grade-level wording I saw is the IRB's Grades 6–8 target for consent forms, which is a narrower context. The 7th–8th grade wording in search results came from a HEALing Communities tip sheet (RTI), which is not an NIH page.

A CDC Clear Communication Index score of 90 or higher is passing.

Only search snippets said this. The index and tool pages I opened don't state 90, and the user guide PDF couldn't be read.

plainlanguage.gov sets a specific grade-level target.

plainlanguage.gov wouldn't load (connection reset). Secondary pages (Chicago, Connecticut, Oregon) give targets from 5th to 10th grade, which suggests there is no single federal number.

34% of U.S. adults score at the lowest numeracy level (as an exact verbatim sentence)

The figure is confirmed on the NCES press release and results page, but the fetch tool only returned it paraphrased, not as the full sentence.

About one-third of hotels raised their rounded TripAdvisor rating by half a star or more within six months of first replying.

Seen only in secondary write-ups (Tourism Nova Scotia blog, search summaries) attributed to the HBR article. The HBR article is paywalled and the paper's PDF returned 403.

After hotels start replying they get fewer negative reviews, but the negative ones are longer.

Appears in a search summary of the OpenBU abstract. The OpenBU page returned 403, so I did not see it on a fetched page.

Replying to reviews stimulates more reviewing, especially negative reviews (Chevalier, Dover & Mayzlin 2018).

The NBER abstract I fetched states this direction but gives no figure. The full text with effect sizes could not be read.

Replying to negative reviews improves later ratings, while replying to positive reviews hurts them (Wang & Chaudhry 2018, Journal of Marketing Research).

Seen only in search-result catalog summaries. I did not open a primary page for it, and no effect size was seen.

Google says replying to reviews improves your ranking or visibility.

I opened Google Business Profile Help pages 3474050 ('Manage customer reviews') and 3474122 ('Tips to get more reviews'). They give only qualitative advice ('Helpful and positive replies to reviews can show that you're responsive to your customers.') and no numbers or ranking claims.

Indiana has 591,671 small businesses, 99.4% of all Indiana businesses (2025 SBA Advocacy profile).

Seen only in search snippets and on a third-party listicle (boostsuite.com). I could not open the SBA PDF itself (403 / connection reset).

Small businesses employ about 1.2 million Hoosiers, 43.2% of Indiana employees (2025 profile).

Seen only in search snippets. The primary PDF was blocked.

Number of self-employed people in Indiana (2025 profile).

No figure turned up, even in snippets. One snippet showed a count of firms with no employees, 480,921, but its column label was cut off. The PDF was blocked.

Indiana is home to more than 569,500 small businesses that support more than 1.2 million Hoosier jobs.

This is verbatim on an Indiana Economic Development Corporation (state government) page dated May 5, 2025 (https://iedc.in.gov/events/news/details/2025/05/05/gov.-braun-proclaims-small-business-week-in-indiana-citing-economic-community-impact-of-entrepreneurs). It names no source and gives no data year, and it repeats what appears to be the older 2024 SBA count, so it is secondary. It is not a substitute for the profile.

Small businesses make up 43.5% of U.S. GDP.

Every SBA Advocacy copy returned 403. Search snippets say this comes from the 2018 Kobe and Schwinn report and covers 2014 data, and that the 2026 FAQ repeats it without a newer year. If used, label it a 2014 figure.

There are 36.2 million small businesses (36,207,130) employing 62.3 million people, 45.9% of private-sector workers.

Seen only in search snippets of the SBA FAQ 2026. The PDF and landing page returned 403.

Small businesses created about 9 in 10 (88.9%) of net new jobs, a net 1.2 million, from March 2023 to March 2024.

Seen only in search snippets of the SBA 2025 U.S. profile, which returned 403. The underlying BLS firm-size tables were not opened in the time available.

1.28 million establishments opened and 1.13 million closed (March 2023 to March 2024); small businesses accounted for about 1.1 million openings and 982,940 closings.

Search snippet only; the SBA profile PDF returned 403. The BLS quarterly figures above are the confirmed replacement.

There are 36.2 million small businesses in the US, 99.9% of all businesses (SBA Advocacy FAQ 2026).

Seen only in search-result summaries. The SBA page and PDF returned 403, so it is not on a fetched page.

Small businesses employ 62.3 million people, 45.9% of private-sector workers.

Seen only in search summaries of the SBA FAQ 2026; the page was blocked (403).

82.3% of small businesses have no employees.

Third-party sites attribute it to the SBA FAQ 2026; the SBA page was blocked. Census confirms the 29,811,495 count for 2022, but SBA's 82.3% share is unverified.

Indiana has 569,851 small businesses (99.4% of Indiana businesses) employing 44.0% of the state's workforce.

Seen only in search summaries of the 2024 SBA Indiana profile (2021 data); the PDF returned 403. No 2025 Indiana profile URL was found.

Indiana had 486,290 nonemployer establishments in 2023 (506,144 in 2024).

Seen only in search summaries of Census QuickFacts; every QuickFacts URL returned 403.

Full normative text of 3.3.8 Accessible Authentication (Minimum), including its exceptions

The fetch tool only gave back the opening fragment ('A cognitive function test (such as remembering a password or solving a puzzle) is not required for any step...'). The fact above uses W3C's plain-language summary from the What's New page instead. Open https://www.w3.org/TR/WCAG22/#accessible-authentication-minimum for the full text with exceptions.

Full text of 2.5.8 Target Size, including its exceptions (spacing, inline links, essential, etc.)

The fetch tool returned only the first clause. That fact is marked 'partial'. The exceptions matter: smaller targets can pass if they have enough spacing around them. Check https://www.w3.org/TR/WCAG22/#target-size-minimum before quoting the rule as absolute.

Section 508 (US federal) has adopted WCAG 2.2

Not researched in the time available. As far as I know, the Section 508 rules still point to WCAG 2.0 AA, but I did not check this on a fetched page, so it is not reported as a fact.

Gmail's announcement blog states the 0.3% spam-rate threshold

The October 3, 2023 Keyword post only says 'a clear spam rate threshold that senders must stay under' and gives no number. The 0.3% and 0.10% figures come from the Google help pages cited in facts.

Outlook began outright rejecting non-compliant high-volume mail on a specific date (third-party sites say from September 2025 or early 2026)

No Microsoft page I fetched gives a rejection start date. The Postmaster page only says messages 'may be rejected' if issues remain unresolved.

Yahoo publishes a separate numeric complaint threshold in its FAQ

The Yahoo FAQ page gives no number. The 0.3% figure is on the Best Practices page only.

Shopify says its storefront theme and pages cannot be run outside Shopify.

None of the help.shopify.com pages I opened say this outright. These were Theme code, Downloading themes, Duplicating themes and Duplicate store. They say only that themes are mostly written in Liquid, 'Shopify's templating language', and that Theme Store themes are licensed to one store. A search of shopify.dev also found no explicit statement. Saying the theme can't run elsewhere is an inference from those pages, not a quote. One Shopify community answer said 'Shopify themes are unique to our platform', but a forum post is not a primary source, so I did not use it.

Shopify's pages and blog posts cannot be exported at all.

The Duplicate store page leaves pages and blog posts off its CSV export list and says to copy and paste them by hand. But the Rich text editor help page says: 'If you have many products, collections, blogs, or pages to update, then you can export them to CSV files'. So 'cannot be exported' is not safe to print. Use only the 'copy and paste by hand' line.

Shopify order export details (what is included and how it is delivered).

I opened https://help.shopify.com/en/manual/fulfillment/managing-orders/exporting-orders. It says transaction history covers 'only captured payments data' and that orders over 50 are emailed. These details are too technical for the page, so I did not make them into a fact. The Duplicate store page also says 'Orders can't be imported into a store through the Shopify admin', which Jacob could use for a migration line if wanted.

Newer (2021+) Pew figure for the share of adults who 'mostly' go online via smartphone

The 2021 Mobile Technology and Home Broadband report (fetched) has no 'mostly' figure. The 2024 URL from search (pewresearch.org/internet/2024/mobile-technology-and-home-broadband/) returned the 2019 report. 2019 (37%) is the newest one I confirmed.

Rural or lower-income breakdown of the share who 'mostly' go online by smartphone

The 2019 report says only 'Similar gaps are present by household income and race and ethnicity.' It prints no income or rural figure for this measure.

Pew share of adults who look up local businesses on their phone

No such figure on the Pew pages I fetched. U.S. Smartphone Use in 2015 covers health, real estate, jobs, banking and navigation, but not local business lookups.

77% of 18-29s and 17% of 65+ have bought something on their phone (Pew 2016)

This came from a secondary write-up (martech.org) in search results. Pew's own page did not show an age breakdown in my fetch, so it's not confirmed.

Small businesses lose $137–$427 per minute of downtime (Ponemon / 'industry benchmarks').

Seen only in vendor blogs (ManageEngine, transparent.ca) that cite Ponemon without a link to a primary report. No Ponemon document was opened.

Micro SMBs (under 25 employees) lose about $100,000 per hour, or $1,670 per minute (ITIC).

Seen only in secondary blogs. ITIC's own Part 1 and Part 2 pages give no such survey result. Part 2 says micro SMBs 'typically would not rack up' hundreds of thousands per hour, and its $10,000/$25,000 SMB figures are worked examples, not survey results.

57% of SMBs with 20–100 employees say an hour of downtime costs up to $100,000 (ITIC 2024).

Only in search summaries. The ITIC Part 2 web page that was opened does not show it, and the PDF was unreadable.

Downtime costs SMBs up to $50,000 per hour; 37% of SMBs have lost customers and 17% lost revenue due to downtime (Bitdefender, 2020).

Opened the Bitdefender Business Insights post (20 May 2020). It is a vendor blog that gives no survey sponsor, sample size, countries, firm sizes or dates for these figures, so it fails the methodology rule.

A small business might lose $5,000 or more per minute of downtime (N-able blog).

Search snippet only, and described there as a general estimate with no survey behind it. Not opened.

Small businesses can lose up to $10,000 per hour during critical outages (Corporate Technologies white paper, 2026).

Search snippet only. Methodology unknown and not opened.

X% of US small businesses have no website (Census or SBA figure)

I found no such figure from either agency. The Census 2018 Annual Business Survey digital technology story (2017 data, published Jan 2021) covers keeping records in digital form, not having a website. The 2019 ABS automation/technology module's 20 tables cover technology use, production and their effect on staff, with no website or online-presence measure. Searches of Business Trends and Outlook Survey releases turned up core questions and an AI supplement, but no website question. As far as I could find, neither agency publishes a share of small businesses with no website.

There are 36.2 million small businesses in the US, 99.9% of all businesses, employing 45.9% of private-sector workers (SBA Advocacy 2026 FAQ)

Seen only in search snippets. Every SBA advocacy.sba.gov page I tried returned 403 or a connection reset, so it is listed under blocked.

Small-business owners spend 22.4% of their time on admin, legal and compliance tasks (UENI, 837 businesses with 5 or fewer employees, May 2024).

The UENI page opened and does show the figure and sample, but it has no methodology section: no question wording, recruitment or weighting. UENI is a vendor that sells website services, and the figure lumps admin together with legal and compliance.

Entrepreneurs spend about 16 hours a week (36% of a 45.5-hour week) on admin (Time etc / Censuswide, 251 US entrepreneurs, Sept 2023).

Seen only on an aggregator (stealthagents.com). The 16.38-hour figure is a derived product of two averages, not a reported result. The original vendor report was not opened.

Small-business owners spend 23 hours a week on admin (attributed to a 'SCORE 2025' survey of 2,100 owners).

Appears only on SEO-style aggregator pages. No SCORE primary source was found.

24% of owners rate unreasonable government regulations as a critical problem (NFIB 2024).

Seen only on a third-party page. The NFIB PDF and press pages could not be read (see blocked).

29% of employer firms cite government regulations as an operational challenge (Fed SBCS 2024).

Seen only in a Coleman Report summary. The Fed PDF came back unreadable and the Fed press release does not state it.

Firms with fewer than 20 employees pay $10,585 per employee to comply with federal regulations (SBA Advocacy / Crain 2010).

The SBA pages failed to load. The figure is also from 2010 and was later disputed, so avoid it even if confirmed.

Text messages have a 98% open rate (often paired with '90% are read within 3 minutes').

No primary source found. Search results trace it to a 2010 Mobilesquared white paper ('Conversational Advertising'), and many sites wrongly credit Gartner. Secondary sources say Gartner's archived article only said 'various sources report' rates as high as 98% and did not measure it. I did not open the 2010 white paper or the Gartner page. Mobilesquared's own page ('A Guide To The SMS Market', byline 2025-06-07) says 'our data previously found that 98% of SMS messages were read' and that its 'latest data ... suggests that 55% of all SMSes are read'. The page names no year for the 98% figure and gives no method for either number. Mobilesquared is a vendor, so neither figure qualifies.

55% of text messages are read (Mobilesquared's newer figure).

Seen on Mobilesquared's own page (https://www.mobilesquared.co.uk/?p=13739), but it is a market-research vendor's bare number with no method shown. I did not open the linked 2022 article.

CTIA count of texts sent in the US in 2025 (2026 edition).

I opened CTIA's 2026 Annual Survey Highlights PDF (https://api.ctia.org/wp-content/uploads/2026/09/2026-Annual-Survey-Highlights.pdf). It reports data use, connections, investment, prices and cell sites but no text-message total, so 2024 (2025 edition) is the newest CTIA texting figure.

Current Pew figures on texting frequency by age.

Searches turned up nothing from Pew on texting volume or frequency newer than 2011-2012. Pew's recent mobile work does not appear to measure texting volume.

The old Personal plan was $16/month billed annually or $23 month-to-month, and Business was $23 annual or $33 monthly.

Only third-party reviews (NerdWallet, larryludwig.com, ecommerce-platforms.com) state this. Squarespace's current legacy-plans article lists no prices, and the Wayback captures that would show Squarespace's own pricing page could not be fetched.

A Squarespace help article titled 'Pricing changes' gives dated price increases.

No help article by that name turned up in searches limited to support.squarespace.com. The price-change wording sits inside 'Legacy website billing plans', 'Manage automatic subscription renewals' and 'Choose the right Squarespace plan', and none of them gives a dollar amount or the date of an increase.

Squarespace's Q1 2023 shareholder letter said 'pricing increases to our legacy customers contributed to our outperformance this quarter', and those increases began rolling out in Q3 2022.

Seen only as quotes in Domain Name Wire and earnings-call transcript sites. The original letter was not found, and the SEC 10-Q texts were cut off before the MD&A section.

In 2026 Squarespace raised annual prices: Basic $16→$19, Core $23→$29, Plus $39→$49, and the top plan $99→$139 (April 26, 2026).

Only DIYPhotography, CostBench and PricingSaaS report this, and they conflict on the top plan. One says Squarespace told customers by email rather than in a public announcement. No newsroom or help-center announcement was found. The current pricing page does show $19/$29/$49/$99 and $25/$39/$65/$139, but it gives no 'before' prices or date.

Barclays found Squarespace list prices rose $2–$7 a month (9%–11%) across all four tiers.

An analyst observation reported by TipRanks/The Fly, not a Squarespace disclosure. Not opened.

Real photos convert 35% better than stock photos.

The primary source (MarketingExperiments, 2011) is one test on one debt-counseling homepage. It compared a recognizable founder with a stock headset model, and the founder's fame or reputation may have driven the lift. The page gives no sample size, confidence level, test length or traffic. It does not support a general claim that real photos convert 35% better. The page also says the lift came on top of earlier gains from optimization.

The founder photo increased sign-ups by 34.7% at a 95% confidence level.

This appears only in a DigitalMarketer search snippet. The 34.7% figure and the 95% confidence level are not in the text of the MarketingExperiments post I opened, which says only '35 %'. The results chart is an image the fetch could not read.

A 2023 Marketing Experiments test found real photos beat stock by 45%.

Seen only in a search summary pointing to vendor blogs (metrifi.com, spiltmedia.com). No primary MarketingExperiments page was found or opened.

Staff news photos are judged more credible than stock photos (University of South Carolina, Tara Mortensen).

The university news post (sc.edu, Nov 19, 2020) says 'Credibility perceptions for staff photos were significantly higher than credibility perceptions for stock photos' but gives no numbers, sample size or journal. It is also about news sites, not business sites. The paper itself was not located.

Faces on websites increase trust (Cyr, Head, Larios and Pan, MIS Quarterly 2009).

Opened the Penn State repository abstract (MIS Quarterly 33(3):539-566). It reports that images with faces raised perceived appeal, warmth and social presence, but that 'human image conditions had no direct effect on trust'; the effect ran indirectly through appeal and social presence. It also gives no headline figure, and it does not compare stock with real photos.

Share of cleaned hacked sites running out-of-date software in 2024 (the newer version of 2023's 39.1%)

Sucuri has not published a 2024 or 2025 Hacked Website Threat Report built on its cleanup cases. The newest cleanup-based report on sucuri.net/reports is the 2023 one. The 2024 SiteCheck report scans sites from the outside, and its page has no outdated-software figure.

Share of hacked sites with backdoors in 2024 (the newer version of 2023's ~50%)

Neither 2024 report gives a backdoor percentage. The closest figure is 16,978 detections of web shells (tools attackers leave behind to control a hacked server), which is not a share of hacked sites. The GoDaddy summary describes backdoor plugins only in words.

Sample size of cleaned sites for 2024

There is no 2024 cleanup sample. The 2024 figures come from 70.8 million remote scans, which found 1,176,701 infected sites.

Any Sucuri 2025-data threat report

No 2025-data report appears on sucuri.net/reports or in searches. The only 2026 item is a customer survey with 51 responses, which has no hacked-site figures.

A current (2020s) government figure for the share of U.S. adults who book doctor's appointments online

The NHIS question on booking online stopped after 2018, according to IPUMS NHIS variable notes seen in search results only, not opened. NCHS Data Brief 482 (2022 data) does not ask about booking. ONC/ASTP briefs 77 (2024 data) and 69 (2022 data) measure being offered and using a portal, viewing test results and notes, and app use, but not booking. The newest government booking figure confirmed is 10.3% of adults in 2015.

35% of patients have booked an appointment through a website or patient portal (Jarrard survey, reported by Becker's)

Private survey seen only in a search snippet and a trade-press summary. Not a government source and not opened. Not used.

Cancer survivors aged 18–64 who booked appointments online rose from 4.9% (2011) to 20.5% (2018) (Preventing Chronic Disease, NHIS)

Seen only in a search summary. The CDC article (https://www.cdc.gov/PCD/ISSUES/2021/21_0163a.htm) was not opened in the time available. It also covers only cancer survivors, not all adults.

Let's Encrypt stats page: X% of web pages loaded (Firefox) use HTTPS / share loaded with Let's Encrypt certificates

Fetched https://letsencrypt.org/stats/ (last updated October 14, 2026). The charts are drawn by script, so the fetch returned the headings but no numbers. The page also says Firefox telemetry changed in version 156 and data after September 15, 2026 is paused. It gives no share of page loads using Let's Encrypt and does not mention 90-day or 6-day lifetimes. Jacob should open it in a browser.

Apple support page stating the 200/100/47-day schedule

Searches found only third-party coverage that names Apple as the ballot's proposer. Apple support 103769 (Nov 6, 2023) only gives the older 825-day limit for iOS 13/macOS 10.15.

Chrome Root Program page stating the 47-day schedule

The googlechrome.github.io/chromerootprogram/ landing page has no validity figures or SC-081 reference. I did not open the full policy text it links to.

Let's Encrypt went from serving 492 million websites to 762 million in the past year

The fetch tool returned this sentence from the 2025 end-of-year letter. It sits oddly next to 'more than 700 million websites' in the same letter, so I left it out until Jacob reads the paragraph by hand.

Combosquatting domains are far more common than typosquatting domains (often cited as ~100x more)

Not in the Kintis et al. abstract I fetched; the body text is in the PDF, which I could not read

Szurdi et al. found millions of typo domains / a specific total count

The abstract only says 'millions of registered domain names appear to be potential typos'; no exact count was visible on the fetched page

89% of consumers / UK adults use online reviews when researching a product or service.

Seen in the CMA's Google (Jan 2025) and March 2026 press releases, but the March 2026 release attributes it to Which? research. I did not open the Which? original, so it is second-hand.

Up to 50% of online reviews are fake; 87% of consumers think they have read fake reviews.

Listed in the Statistics section of the CMA's 27 March 2026 release, credited to Truth Engine and PowerReviews (2022). These are third-party vendor figures. I did not open the originals, so there is no primary source or method.

£23 billion breakdown by sector (e.g. £14.4bn in travel and hotels).

Only in secondary coverage of the 2015 report. I did not open the 2015 CMA report PDF. The 2015 press release does not carry the £23bn figure; the 2025 gov.uk releases do.

53% of operators had an outage in the past three years, per Uptime's 2025 outage report (as Data Center Knowledge puts it)

The figure is in the 2025 outage analysis summary (p.4), but it comes from the Global Data Center Survey 2024 (n=768), not 2025. The newer 2025 survey figure is 50% (n=754), which is fact 1. Do not label 53% as 2025 data.

Outage rates fell from 78% in 2020

The outage analysis summary (p.4) says '78% in 2020'. The Global Data Center Survey 2025 trend chart (Figure 10, p.18) shows 74% for 2020 (n=494). Uptime's own documents disagree, so don't quote the 2020 starting point.

Power causes 54% of outages

54% is the 2024 survey figure (n=97), shown in the outage analysis Figure 3. The 2025 survey gives 45% (n=96). The 54% does not describe 2025.

Q3 2026 DNIB figures

No Q3 2026 release found as of 2026-10-17. Last year's Q3 release came out October 23, 2025, so Q2 2026 (published July 23, 2026) is the latest available. Recheck in late October 2026.

The world passed 400 million registered domains for the first time in Q2 2026

Only trade press (DNJournal) says 'first time'. The Verisign pages I opened give 401.6 million but do not use the phrase 'first time', so the milestone wording is not confirmed.

Brute-force attacks rose from about 20% to 60% of basic web application attack incidents (2025 DBIR).

Seen only in vendor blog summaries (aembit, flare and others in search results). The SMB snapshot names brute forcing for this pattern but gives no percentage, and the full 2025 PDF could not be fetched.

Credential abuse was the initial access vector in 34% of breaches (2025 DBIR).

Appeared in a search-result summary only. The fetched Verizon snapshot and press release both say 22%, so the 34% figure looks wrong or out of context.

Basic Web Application Attacks were 12% of all 2025 DBIR breaches (up from 9%).

The 2025 SMB snapshot's inner cover chart (p. 2) shows 9% → 12%. But the 2026 Executive Summary's ring chart (p. 2) labels the 2025 share as 18% Basic Web App Attacks and 12% Miscellaneous Errors, the reverse of the 2025 document. The two Verizon documents conflict, so I am not reporting either figure.

Weebly's share of all websites and of the CMS market a year ago (October 2025)

Weebly is below the 1% cutoff, so it is left out of both W3Techs monthly history tables. Its detail page and the Wix/Squarespace/Weebly comparison page show the trend only as a chart image with no numbers, so no year-ago figure was seen.

Year-ago figures printed on the main CMS overview page

The overview page (content_management) shows only the current figures, with no year-ago column. All year-ago figures above come from W3Techs' separate monthly history pages (all-websites and market-share views).

In the US, people with disabilities have over $200 billion in annual discretionary spending (W3C business case page).

The sentence is verbatim on w3.org but has no footnote. AIR's 2018 study, the usual US source, gives about $21 billion discretionary ($490 billion disposable), and AIR's 2026 study gives $107 billion discretionary. The W3C figure does not match either primary source, so do not use it.

In the UK, people with disabilities and their families spend at least £249 billion a year (the 'Purple Pound').

Verbatim on w3.org, but no source is cited there. No primary source was opened.

The global disability market is nearly $7 trillion / 2.3 billion people control $6.9 trillion.

The $7 trillion has no source on the W3C page. The 2.3 billion / $6.9 trillion figure cites a 2016 Forbes contributor article, which is not a primary source. Neither was traced to original data.

15% of the world's population has a disability (W3C, citing the World Bank in 2018).

The figure is out of date. The World Bank page now says 16%, and the WHO's 2023 fact sheet says 1.3 billion / 16%. Use the WHO or World Bank fact instead.

People spend 11.5% of their lifespan with a disability in countries where life expectancy is over 70.

W3C cites Disabled World (2017), which is not a primary source. No original data was found.

2025 mobile Core Web Vitals pass rates for Squarespace, Joomla and Drupal as absolute percentages

The chapter text gives only their year-over-year gains (+8, +7, about +4). The absolute values are in the Figure 12.8 chart image and the spreadsheet, which could not be read.

Webflow Core Web Vitals pass rate (e.g. 'Webflow sites pass at 67.4%')

Webflow is not named in the 2025 CMS chapter text. The 67.4% figure appears only on a third-party blog (pravinkumar.co) that also misstates the overall rate. No primary source.

Shopify Core Web Vitals pass rate in the CMS chapter

The CMS chapter gives no Shopify figure. The 76% figure comes from the Ecommerce chapter (Figure 13.11) instead, and is recorded above from that chapter.

Median page weight for Wix, Duda, Drupal and Webflow, 2025

The chapter text names only Squarespace (3,974 KB), Joomla (2,900 KB) and WordPress (2,894 KB). The other values are only in the chart.

Rakuten 24: a good LCP brings up to a 61.13% higher conversion rate

It is on the web.dev page, but it is a correlational analysis of homepage field data, not a before/after test. The fetch summarizer gave it as 'An increase of up to 61.13% in conversion rate' and the chart caption differently. It was not recorded because the wording is unclear and it is not causal.

Yelp: a 12% conversion increase from cutting load time by one second

Seen only in a summarizer's account of the Yelp blog (an earlier backend experiment). It is not a Google-published case study, and the full sentence was not confirmed.

WebAIM Million average errors per home page for Weebly, GoDaddy Website Builder and Duda

Neither the 2026 nor the 2025 WebAIM Million page lists these platforms anywhere. WebAIM publishes no figure for them, and no other source was found.

Webflow's average errors per home page in the 2026 WebAIM Million

Webflow is missing from the 2026 table. The latest WebAIM figure is 28.4 from 2025 (recorded as a fact above). A search snippet tied 28.4 to 2026, but that figure is from the 2025 and 2024 reports.

Wix and Squarespace figures from the 2025 WebAIM Million

The 2025 CMS table does not list Wix or Squarespace. Their figures (33.3 and 33.0) are on the 2026 page only. AbilityNet's claim that Squarespace pages had 'roughly half the errors' was not traced to a WebAIM page.

Percentages of respondents naming CAPTCHA, unexpected screen changes or missing alt text as problems

The #10 page gives only a weighted rank order of the 12 items, with no per-item percentages. Do not attach a percentage to these items.

A 2026 (#11) ranking of the most problematic website items, including CAPTCHA

The #11 results page (updated Sep 29, 2026) has no Problematic Items section and no CAPTCHA section. The newest ranking is still #10 (2024).

Weebly's archive is a zip of your site's pages, uploaded images and uploaded files, emailed to you from Settings > General > Archive.

Only seen in third-party pages (a MemberClicks help article about a Weebly-based editor, an Ultra Web Hosting knowledge-base page, and an independent consultant's FAQ from 2011, updated 2015) and in search summaries. Weebly's own help page would not load.

Blog and store pages are not included in the Weebly archive.

Found in a MemberClicks help article, community forum posts (reportedly a Weebly community manager saying 'it does not include your blog posts') and search summaries. None of these is the Weebly help center, and the help article itself would not load.

Contact forms, slideshows and other built-in features stop working outside Weebly after export.

A search summary said the official article says this, but the article text never loaded. A search summary is not confirmation.

Weebly says the archive file can be used to recreate the site at another host.

Only came from a search summary of the Back Up Your Site article. Not seen on a fetched page.

An exported Weebly zip cannot be imported back into Weebly.

Only in third-party guides and community forum replies, not the Weebly help center

Weebly is shutting down in 67 countries and unpublishing sites on January 2, 2027.

Only in a third-party hosting company's knowledge-base page (Ultra Web Hosting, updated July 2026). There is no Weebly or Square source, and it is outside this topic. Do not use it without checking.

A Weebly site cannot be moved into Square Online; you have to build a new site.

Seen only in a Square Community forum post by moderator MayaP (18 Nov 2024): "Please note that it's not possible to migrate the Weebly editor over to Square Online." and "You'll need to create an entirely new site." (https://community.squareup.com/t5/Online-Store/Is-there-a-way-to-migrate-my-store-from-the-Weebly-editor-to-the/m-p/762246). This is a forum reply, not a help article, and the official FAQ it links to returned no body text.

Weebly's Archive zip cannot be imported back into Weebly; it is only a personal backup.

Only a user's report in a Square Community thread (Alberte1, 12 Mar 2025, quoting a Square Specialist) and third-party hosting blogs. The official Weebly backup page is script-rendered.

Weebly's export leaves out blog posts and store pages.

Seen only in search snippets of a private forum thread and on a MemberClicks help page about its own Weebly-based product. No Weebly or Square page could be opened.

Weebly is shutting down in 67 countries (sites unpublished 27 Sep 2026; Network Solutions says platforms end after 2 Jan 2027).

Seen on Wikipedia, Khaleej Times, Network Solutions and migration-vendor blogs, which disagree on the final date. The primary Weebly notice is script-rendered. It is also out of scope for a US audience, since the US is reportedly not among the affected countries.

Old Wix plan prices (e.g. Combo about $16/mo, Unlimited about $22/mo, VIP about $45/mo)

No Wix page I fetched names the old prices. The help articles name the old plans but give no prices; the pricing article points to a PDF instead. Only third-party review sites state these figures.

The legacy-to-new migration began with renewals from March 3, 2024

Seen only in search snippets of a Wix help article that now returns 404; not confirmed on a fetched page.

The new Light/Core/Business plans rolled out September 12, 2023

Comes from a forum.wixstudio.com post by user 'noahlovell' (Aug 16, 2023); the page does not show that the poster is Wix staff, so it is not a confirmed Wix source.

Wix raised Light from $17 to $19/mo in Aug 2026, and Business from $36 to $39 in Jul 2025

Only a third-party tracker (usagepricing.com) says this. It is not primary, and I could not check it against a clean capture of wix.com/plans.

Wordfence 2025 annual WordPress security report (published 2026) figures

Several searches, including one limited to wordfence.com, found no 2025 annual edition. Wordfence seems to have moved to quarterly reports, and those pages would not open. The 2024 report, published April 2025, is the newest annual edition I could confirm.

Share of attacks aimed at abandoned plugins

The 2024 report has no such figure. It only defines abandoned software as 'software that hasn't been updated in the last two years' and advises removing it (p.65). The closest figure is the ~35% of 2024 vulnerabilities still unpatched in 2025 (recorded as a fact).

A separate credential-stuffing volume

The report folds credential stuffing into the single 55-billion password-attack figure (p.50). No separate number is published.

Wordfence blocked about 54 billion malicious requests in 2024

Confirmed in the PDF (p.6, p.49), but it is the logged total, of which the already-used 48 billion were blocked. I left it out as a fact so the page doesn't show two conflicting attack counts.

Webflow lets you export CMS content as a CSV, but images come out only as links to Webflow-hosted files.

Only search-result summaries and third-party guides (clonepartner.com, digidop.com) say this; every Webflow-owned page I tried was blocked.

You can move a Shopify theme to any host.

Shopify's licensing page says the opposite: a Theme Store theme is licensed only to the store where it was bought, works on one store at a time, and can't be bought anywhere outside the Theme Store. I found no Shopify page saying a theme runs off Shopify.

About half of the WordPress.org plugin directory hasn't been updated in two years.

The only source found was WP Tavern (Nov 20, 2015), passing on an analysis by Luca Fracassi of Addendio: 'about 22K plugins have been updated in the last 24 months representing a little more than half the directory.' That is a news write-up of a third party's analysis, and the data is from 2015. No current primary figure for the whole directory was found.

Average time from disclosure to patch for WordPress plugins (e.g. 'site owners take 14 days to patch')

Neither Patchstack report measures how long fixes take to arrive (the 2026 report gives the share unpatched at disclosure and how fast holes get exploited). The 14-day figure turned up only in secondary blogs with no dataset behind it. WPScan's and Wordfence's pages could not be read.

About 35% of vulnerabilities disclosed in 2024 were still unpatched as of April 2025 (Wordfence)

Seen only in search snippets. The Wordfence PDF could not be read (see blocked).

Nearly 30% of critical WordPress plugin vulnerabilities are never patched (attributed to Bitdefender)

Seen only as a secondary citation in search results. No primary document was found.

More than 59% of plugins have not been updated in two or more years

Appears in a blog with no named primary source. Not traced.

42% of consumers have arrived at a business only to find it closed because of incorrect hours (BrightLocal, 2022).

This only turned up in a search snippet from a umb.edu subdomain that I couldn't open. None of the BrightLocal pages I opened (2018 Citations Trust, 2023 Discovery & Trust, Consumer Review Survey 2025 and 2026) contain it.

52% of local search users have visited a closed business; 55% of listings had wrong hours (Infogroup).

The only copy I could open is a Thryv blog post from 2013 quoting the survey with no sample size. Infogroup's original release is now a 404 on data-axle.com.

36% arrived too early or late because of incorrect hours (BrightLocal 2018).

This is on the page ('A further 36% arrived too early or late due to incorrect hours displayed online.'), but it contradicts the same page's Key Findings figure of 24%. 36% is also the page's figure for calling a wrong number, so it looks like a typo. Don't use it.

The BrightLocal Local Consumer Review Survey reports shares who turned up to a closed business or lost trust over wrong information.

I opened the 2025 edition (published 29 Jan 2025, 1,026 US adults via SurveyMonkey) and the 2026 edition (published 11 Feb 2026, 1,002 US adults via SurveyMonkey). Neither has a figure on wrong hours or closed businesses. The 2026 edition only advises businesses to 'monitor business listings for inaccuracies'.

About 53% of Yelp reviews are 5 stars and 18% are 1 star (with 16% 4-star, 8% 3-star, 6% 2-star).

Only found on third-party listicles (Wallstreetzen, WiserNotify, electroIQ, Sci-Tech Today), which disagree with each other: 5-star figures range from 52% to 55%, and electroIQ swaps the 4-star and 1-star figures. Yelp's own Fast Facts chart was opened but its values are graphics. No figure is recorded until someone reads the chart by hand.

Share of reviews that are 4-5 stars versus 1-2 stars on Yelp.

Same as above: the only primary source is the Fast Facts chart, and its values could not be read as text.

Yelp's recommendation software hides about 18% (or 19%) of reviews.

18% is from the 2024 report and about 19% is from the 2022 report, both seen only in search snippets. The newest edition (2025 data, opened on yelp-press.com) says 17%. Use the 17% figure.

trust.yelp.com gives an exact share of reviews that are recommended.

Opened trust.yelp.com/recommendation-software/. It only says 'Hundreds of millions of reviews are posted on Yelp' and gives no percentage. The trust.yelp.com 2025 report page showed '22 million cumulative reviews were contributed to Yelp in 2025' but no percentage split in its text.

Number of reviews Yelp removed for being incentivized or paid for (2025)

None of Yelp's own 2025 pages gives this count: not the report page, the 2025-report subpage, the press release or the blog post. They say only that the software was improved to filter reviews from people who 'received incentives in exchange for their review'. The nearest numbers are 128 Compensated Activity Alerts (these count businesses, not reviews) and 1,991 accounts closed for review-exchange rings (these count accounts). Search also turned up nothing.

Yelp issued 527 warnings to businesses for one-time incentivized reviews in 2025

A search summary attributed this to the report page, but the fetched report page and 2025-report subpage do not contain 527. Do not use it.

A raw count of reviews not recommended in 2025

Yelp gives only the share (17% of about 22 million). A count like '3.7 million' would be our own arithmetic, not a printed figure.