Corrections
45 mistakes in our own reviews that we found and fixed. Prices that a vendor changed are not mistakes and are not listed here; those are on the Price Watch. This page is for the times we had it wrong.
Each entry says what we got wrong and quotes the sentence that records the correction on the review itself, so you can check it there. The date is when that correction entered our records; entries marked “or earlier” predate our version history. Most of our research and writing is done by an AI assistant under fixed rules (who does what), and most of these mistakes were found by re-reading the vendor’s own page.
Found something wrong? Write to daniel@aibuildermarketplace.com. The same list is available as data: /api/v1/corrections.json.
We gave a Perplexity answer about SaneBox as the example of an answer without sources. In the 2026-10-01 edition that answer cites ten sources; the only unsourced answer is ChatGPT on Kit, which asked which product we meant. The example now comes from the edition itself, and two sentences that presented lost pricing conditions as a finding were rewritten as a risk this study does not measure.
11 of the 21 example quotes were cut at 160 characters, some mid-sentence, on the report pages and in the edition's stats.json. They now carry the full evidence sentence from the same edition's tools-flags.csv; no count changed.
The summary card said 11 tasks while the text said ten. It is ten main tasks plus one extra feature check (the AI Actions Library, locked on the trial); no result changed.
- Our YouTube videos
25 review videos said a product had a free plan because a price text contained the word 'free' (for example 'free SSL' or 'toll-free'). Each of those videos now opens its description with a correction, and the video script no longer claims a free plan unless the pricing page lists one.
We reported conflicting Pro prices; the card has a credit selector, and $29 and $49 buy different amounts of credit.
“Correction to our 16 September reading, which reported Pro at $49 on the card and $29 in the table”
We said the first Team seat costs more than the others. It does not: those figures came from the individual-plan column of the same table.
“A correction: earlier versions of this review said the first Team seat costs more than the rest.”
We budgeted a filing from a worked example; the vendor's own chart puts it about $137 higher.
“We previously budgeted from that example ($885.30, $1.48 a form)”
We said monthly billing renews at $6.89; it is $4.39.
“Starter $4.39 renewing monthly, not the $6.89 list our earlier text said”
The price field described a different company's product ($3 per user for employee recognition), not Assembly's client portal.
“Correction: until 28 September 2026 this field described a different company's employee-recognition plan”
We gave the plan as $99 a month. $99 is the Influencer Pages add-on; the plan was $199 when we first read the page.
“Correction: until 28 September 2026 this field put the plan at $99 a month”
We carried Pro at about $15 a month from third-party trackers; it was never confirmed on Consensus's own page.
“Correction: our earlier reading, which we could not confirm at the time, put Pro at about $15 a month”
We read the Studio Plus price as Pro's ($275 at 200 clients).
“our earlier '$275 at the 200-client end' read Studio Plus as Pro and is corrected”
We claimed that adding a 26th user lowers the bill; that claim is withdrawn.
“The structured data names the same bands without 'first' and 'next'”
We put unlimited workspaces and the $1,500 budget on the Business plan; they belong to Enterprise.
“Correction 23 Sep 2026: our 5 Sep reading put unlimited workspaces and the $1,500 budget on Business”
We said Navan is free up to fifteen users; the page says five.
“Our earlier reading put the free threshold at fifteen users; the page says five”
We gave Pro as 'from $55 a month'; the ladder starts at $29 for 500 credits.
“CORRECTION 23 Sep 2026: this dossier previously gave Pro as 'from $55/mo'”
We said Enterprise is annual-only with a ten-seat minimum; the pricing page states neither.
“CORRECTION 22 Sep 2026: Enterprise is NOT annual-only and carries no published seat minimum.”
We said there is no volume discount below three stores; two stores are already cheaper per store.
“The second correction is ours alone: it is not true that there is no volume break below three stores”
We presented the annual-term rates ($249 and $499) as the standing monthly prices; month to month they are $299 and $599.
“Our earlier text put Business at $249 a month and Agency at $499”
We missed the Enterprise line ($500 a month plus $5,000 on signup) entirely.
“Above them sits an Enterprise line our earlier readings missed entirely”
An automated read took 0.99% from hidden page markup as a live UK rate; it belongs to another market's card.
“a WebFetch read it as a live figure on 2026-08-20 and that was wrong then too”
We quoted an agency ladder of about $750 a month and a per-minute rate derived from it. Those figures appear nowhere on Vida's site.
“Correction to our own earlier dossier: an agency ladder of about $750 a month”
We cited the wrong company's domain (weavehq.com) as Weave's pricing page.
“note the domain: our earlier reading cited weavehq.com/pricing”
We quoted $79 a seat from third-party reviews because our server could not read the page. In a browser it renders, and $79 is not on it.
“Our earlier version said Salesflow's pricing page could not be read and quoted $79 per seat”
We quoted euro prices that were reviewers' recollections, not anything Tellent publishes; they have been removed.
“The euro figures this review used to quote were reviewers' recollections”
We 'corrected' the per-response overage from $0.15 to €0.10. Both were right: the overage is priced per storefront ($0.15, €0.10, £0.15), so that correction was withdrawn.
“We carried $0.15 per additional response, then”
We reported a 28% price rise from $89 to $114. There was no rise: the two figures were different plans and billing terms, and the real Growth floor is $56.
“We previously carried $89 as the entry price”
We described an eightfold per-lead gap between Essentials and Business; it does not exist.
“So Essentials and Business cost effectively the same per lead, and the eightfold gap we previously described”
We wrote that a Startup tier was not on the page and struck it from the review; it was there.
“On 17 August we wrote that a 'Startup at $85/$55' tier did not appear on the page”
We carried a tier ladder and a free-preview model that were never confirmed on the vendor's own page.
“Our earlier and never-confirmed reading had design and preview free with downloads paid”
We said the flat Unlimited tier was gone; about one visitor in eighteen is still served it.
“The flat Unlimited tier is not simply gone, and this is a correction to what we published before.”
We described a permanently free option that we could not source from the pricing page.
“Our earlier note here also described a permanently free option, which we could not source”
We attributed a sentence about hybrid pricing to Feedvisor that is not on its site, and the price figures we carried came from third parties.
“One correction to what we wrote before: the sentence we attributed to Feedvisor”
We gave price ranges; they were the monthly and annual rate side by side. Hello Bar publishes exact prices.
“Hello Bar publishes exact prices, so the ranges we carried before were wrong”
We carried $10 and $19 as the price; those are the annual rates, and month to month costs considerably more.
“CORRECTION to the previous dossier: $10 and $19 were carried as the price”
We said VPS licences are cheaper than dedicated ones; that is true only of Web Host.
“We wrote that VPS licences run cheaper than dedicated ones”
Four figures were wrong (Essentials, Team, the ACH fee and the international wire); our own July and August captures already had the right ones.
“Four figures in our earlier text were wrong”
We put extra Business seats at $59 per user. Our own captures show $24 (Pro) and $69 (Business).
“Our earlier text put extra Business seats at $59 per user per month.”
We said the listing showed no paid tiers; our reading had failed on a table that loads in the browser.
“Correction to our own earlier note: on 18 August we wrote that the listing showed no paid tiers”
Our summary contradicted the detailed figures beneath it; the summary now matches the table.
“The detailed figures were right and the summary was wrong”
We reported a price movement and a Plus figure of about $269 a month; neither appears in any of our readings.
“Worth stating plainly because we had this wrong ourselves: Atria's prices have not moved.”
Our break-even for Premium (16-17 orders a month) was wrong; the review now shows the arithmetic.
“Our earlier note put it at 16-17 orders a month; that was wrong on both counts”
We missed the $10-a-month charge per additional user, so team costs were too low.
“The detail our earlier note missed: every paid plan adds $10 a month per additional user”
We put Turbotic at $10-15 a seat, which contradicted our own page saying $25.
“Our earlier text put Turbotic at around $10-15 per seat per month.”
We described a 7-day free trial; each plan is $1 for 7 days, then the monthly rate.
“The earlier "7-day free trial" and "$1/month for 7 days" wording was wrong”