The FY2021 group accounts of the parent (MPN Marketplace Networks GmbH) exist on Bundesanzeiger — but Bundesanzeiger has no stable per-document URL at all, so there is no link a reader can follow, and we never read the figure ourselves (the document is behind a CAPTCHA). We said as much in the honesty note and still badged it FILED. It comes down. Separately: under DiRUG, accounts from FY2022 go to the Unternehmensregister, and MPN's consolidated FY2024 report was published in March 2026 — so €103.2M is three years stale.
Corrections
Every figure on this site carries its source. When a source turns out not to say what we said it said, the figure changes — and the change is recorded here, with the reason, dated. Nothing is quietly edited.
23 correctionsThe prospectus we cite does not contain the figure: its reporting period ends at H1 2021 (RMB 438.5M for the half). The RMB 843M full-year number is in a LATER Shenzhen filing — the third-round review inquiry reply of June 2022 — which we could only reach through a third-party mirror, not the regulator's own copy. The number appears sound; our citation does not reach it, so the badge does.
Idealista's FY2024 group accounts are deposited at Spain's Registro Mercantil, but Spain publishes no free document URL — the accounts must be bought, and there is no per-company permalink. So no reader can follow us to the filing, and under our own rule the badge comes down. Our €300M+ is the GROUP figure as reported by trade press. Worth knowing: the entity-level number a reader is most likely to find is €159.1M (Idealista S.A.U., Spain only).
The number was right and genuinely filed; the citation was not. A reader following our source could not reach our figure, which is the one thing we promise they can do. The MAYA filing is now linked and carries it exactly.
The card asserted a £15M profit for 2022 and badged it FILED while citing no filing — only Wikipedia and Trustpilot. Bark's audited accounts are public at Companies House (10614196) and are now linked, but they are a scanned PDF that cannot be machine-read. I could not verify £15M, and I will not swap in another figure I have not seen myself. The profit claim is withdrawn. Accounts to 31 Dec 2024 have since been filed, so even the year is stale.
The $245M is the acquisition price Shutterstock announced in a press release. A deal price in a PR is a company statement, not a filing. FILED is the strongest claim we make, and from today it requires a source URL a reader can actually follow to the filing.
The interview is dated February 2025, not mid-2025, and in its body the founder says 'around $150k+ per month' — about $1.8M/yr. The card now carries his words and his date. Separately, the header of that same Starter Story page now displays '$756K revenue/mo', five times what the founder says on the page beneath it. We do not know which is true; the honesty note now says so rather than quietly picking one. The '$2.4K ARR at purchase' metric rested on a tweet that no longer loads, and is removed.
The number is correct and genuinely filed — UserWay was TASE-listed. But the Wikipedia page we cite for it does not contain it. A source that does not carry the number it is attached to is a trust bug even when the number is right: a reader who follows our citation cannot land on our figure, which is the one thing we promise they can do.
userway.org/pricing now runs $49 / $119 / $249 a month, read in a browser. The $149 ceiling was true once (a 2024 archive capture shows the old tiers) and is not now.
The card disclosed the 2023 acquisition but described the business as if it were still running. It is not: beehiiv wound the platform down and migrated its users to the beehiiv Ad Network. 'Acquired' and 'gone' are different facts and a reader should not have to guess which one we mean.
Read on 52frames.com/become-patron: the tiers are $199 and $360 (currently discounted to $299). $130 and $260 appear nowhere. Separately, the $11K/mo headline is a May-2023 figure that carried no date on the card — a true number presented as present tense, which is its own kind of false.
The cited pricing page lists $19.90 for a single vehicle; $14.90 appears nowhere on it. Prices are the fastest-decaying field on any card. The ownership was also stale by two moves: the product left SPAN Enterprises in 2022 and now operates under Fleetworthy, so the 'SPAN group' revenue estimate described a company that no longer owns it.
FY2025 has been filed and the trend has reversed. Scoris, reading the Lithuanian registry: "2025 finansiniais metais bendrovė gavo 5,25 mln. Eur pajamų... Pajamos per metus sumažėjo 13,2 %." The FY2024 figure was correct when written. But a card that leads with a growth rate does not just go stale when growth stops — it begins asserting the opposite of the truth. This is the sharpest lesson in the audit: levels age, rates lie.
Ravelry does not take a percentage of pattern sales. Its own post on how it makes money describes fees running "from free all the way up to $20 a month depending on sales volume" — a capped monthly charge — plus ads, merchandise and affiliate commissions. The 3.5% appears on no page of Ravelry's that we could open, and it cuts against their own statement that 98.7% of a pattern sale goes to the designer. Same shape of error as Chrono24: we described the wrong mechanism, not merely the wrong number.
The 11M was a third-party stats blog, badged as though Ravelry had said it. Ravelry's own About page says "over 9 million yarn-loving friends". Our own metric row already labelled the same number an estimate — the card was contradicting itself on its own front.
The €868,555 net profit is from CULTS' accounts filed at the French registry (societe.com reproduces the filed "Résultat net" line for 2023). The card's own label says "2023 filing". Badged STATED, which is the wrong kind — the same hand-set-badge defect as pappers, in the same direction.
The €492,823 net profit comes from Pappers' own accounts filed at the French registry (RNE) — the card's label already said "(filed)" and the honesty note already said "from the balance-sheet filing". It was badged STATED, which says a founder told us. Nobody told us; a registry did. This is the badge error running in the conservative direction, but it is still the badge lying about where a number came from.
The figure is the CEO's LinkedIn post, relayed by Biometric Update. It is in no filing: Yoti's most recent full accounts at Companies House (filed 17 Nov 2025) cover the period ending 31 March 2025. The card badged it FILED and its honesty note claimed it was "filed at Companies House". FILED is the strongest claim this product makes and it was attached to a founder statement. Also: the source says US$39M; our $37M was our own conversion presented as the source's number, so the headline is now the pound figure that was actually said.
Both halves were false. On testimonial.to/pricing, "Remove Testimonial branding" is ticked only from the $50 Ultimate plan (Free and the $25 Starter both show a grey dash), and video is not a paid unlock at all — the free plan includes "2 video and 10 text testimonials". Worse than the error: our honesty note asserted that this loop was "confirmed on the live pricing page (fact)". It had not been. Claiming a verification we did not perform is the one thing this product cannot do.
The source list had been written into dies_counter, so paying members were shown a bibliography where the argument against our own verdict belongs. The argument itself was sitting one field away in dies_evidence. Both are back where they belong. Separately: the honesty note attributed "82M+ returns filed" to the About page; that figure is on the homepage, and no page anywhere says "original" member of the Free File Alliance.
We cited IPinfo's homepage for the customer logos. Microsoft appears nowhere in that page's HTML — checked in a real browser, not a text extract. Snowflake does. A customer name we cannot source is exactly the thing we criticise third-party estimators for.
The old line was backwards, not merely unsourced. Chrono24's own FAQ: "In most cases, Buyer Protection is a free service." We were taxing the one thing they give away. The real meter is a published 6.5% sales commission for private sellers, and for dealers a monthly listing subscription plus a fee Chrono24 deliberately does not publish ("dynamically calculated for each sale", behind a dealer login). The $215M figure came from one third-party estimator that now paywalls its 2024 breakdown, while another puts 2025 at $875M — a fourfold disagreement between strangers is not a figure.
The estimate rested on a single input: "about 8% of registered users pay." Its only source was a Subscription Insider case study, now a 404. The one surviving Wayback capture shows the article body was behind a members paywall — the visible excerpt is literal Lorem ipsum placeholder text. The 8% existed only in the headline; nobody has ever read what it was based on. It is also dated May 2014, and it measured a pay-what-you-want model CellarTracker abolished in March 2024 (they now charge $40-$500 a year, tiered by cellar size). The founder has since said the paid rate doubled. No current conversion rate is public, so there is nothing honest to rebuild an estimate from.