How we verify pricing, and every correction we have made
This site publishes prices for 13 competing products, several of which compete directly with ours. That only has value if the figures are right and if you can see what happens when they are not. The rules are below, followed by a dated log of every correction made so far.
This page covers how we verify what we publish about other companies. For how the product itself scores a website and what its evidence means, see how Robot Visible measures AI visibility.
13
Products with published pricing
19 July 2026
Oldest price check on file
10
Corrections logged
How a price gets published
These are the working rules, not aspirations. The last one is enforced by an automated check that fails the build.
Prices come from the vendor's own page
Every figure is copied from the company's published pricing page. Third-party round-ups, sales decks and review sites are not used as price sources, because they are frequently stale and rarely dated.
Every figure carries the date it was checked
Vendors are re-checked at different times, so each row shows its own date rather than borrowing a single page-level claim that would overstate how fresh the oldest entry is.
A price is never inferred
If a vendor advertises a discount but does not publish the resulting figure, that figure does not get calculated and printed as though it were verified. This rule exists because we broke it once.
Currencies are never converted
These vendors bill in dollars, pounds and euros. Converting between them produces a number that changes daily and that the vendor never quoted, so prices stay in the currency they are published in.
No published price means exactly that
When a company does not publish pricing, the page says so and explains what is available instead. No estimate is substituted, however widely one is repeated elsewhere.
Only comparable units are compared
Tracked prompts, metered credits and bulk analysis volumes are different products. Cost-per-prompt figures cover only vendors publishing a comparable allowance, and the excluded ones are named.
Competitors are described in their own terms
Each comparison states what a competitor is genuinely good at before explaining where this product differs. A comparison a reader cannot trust about the competitor cannot be trusted about us either.
No structured data that misrepresents another company
Competitor prices are published in readable tables, never as schema.org Offer markup. Offer markup asserts that this site sells that product at that price, which would be false for every vendor listed.
Stale data fails the build
An automated check fails if any vendor's pricing has gone unverified for more than 45 days, so the maintenance claim on these pages is enforced rather than intended.
Corrections log
Every correction is listed here permanently, including what caused it. Entries are never edited or removed to make the record look better. Of the 10 logged so far, 6 reached the published site and 4 were caught in review before publication.
Was publishedFactual error
FAQPage guidance after Google retired FAQ rich results
What it said
The guide called FAQPage "the closest thing to handing an assistant its response" and said sites that omitted structured data and site-level files left machines guessing. Its 10 July 2026 modification date made those claims appear current even though Google had stopped showing FAQ rich results on 7 May, removed the related Search Console and testing support in June, and stated that no special markup or AI text file is required for AI Overviews or AI Mode.
What is true
The guide now separates three claims. FAQPage no longer produces a Google rich result; Google AI features have no additional markup requirement; and accurate schema can still reduce ambiguity for standards-aware parsers without proving ranking, inclusion, or citation. It publishes the May, June, and August retirement timeline, explains when keeping FAQPage is worth its maintenance cost, and cites the primary Google and schema.org sources.
Why it happened
The original article collapsed a schema vocabulary, a Google Search display feature, and answer-engine retrieval into one benefit. It also carried a fresh modification date without a source check against Google's current feature documentation. The correction adds primary-source requirements to the guide and states the intended outcome of each machine-readable layer separately.
Five published claims contradicted the derived tables printed on the same page. An answer about entry tiers gave "mid-market runs $250 to $500 (Scrunch AI $250, AthenaHQ $295, Kelsey Agency $499)", but Kelsey's entry tier is Essential at $99/month and $499 is its top self-serve tier. Prompt-metered tools were described as charging "roughly $1.58 to $2.00 per tracked prompt", where $1.58 is Trakkr's £1.58 printed with a dollar sign and the real top of the dollar range is Semrush at $3.96. Profound, Scrunch AI, Peec AI, Evertune, Knowatoa, AthenaHQ and Rankscale were listed as pricing enterprise by conversation, two sentences before the same answer quoted Rankscale Enterprise at $780/month. An "as of July 2026" date was typed by hand on a page whose every other date is derived. AthenaHQ at $295/month and Rankscale at $99/month were compared as peer credit-metered tiers without saying that the first is an entry tier and the second is not.
What is true
Every figure in the guide's prose and FAQ answers is now generated from the same vendor records the tables are built from, so a reprice moves the sentence and the table together. The entry-tier answer lists all nine dollar-priced entry tiers in ascending order with their billing qualifiers. Cost per prompt is quoted as $1.93 to $3.96 within dollars, with Trakkr's £1.58 and Peec AI's €1.70 stated separately and explicitly not ranked against them. The quote-only list is derived and no longer includes Rankscale. Dates come from the recorded verification window. The credit-metered comparison quotes each vendor's own entry summary, which names the tier.
Why it happened
The tables on this page were moved to generated data some time ago; the paragraphs and FAQ answers around them were not, and nothing asserted that the two agreed. Six of the wrong figures sat inside FAQPage acceptedAnswer text, which is the format answer engines quote from, so the page was handing machines wrong competitor prices in its most machine-readable form. Two were currency errors on a page that states twice that it never converts between currencies, and one of those was the exact inversion a code comment already documented as fixed in the table sort. Deriving a table and leaving the prose beside it hand-typed produces a page that looks maintained and is not.
The cost guide's meta description and Dataset markup both claimed "published pricing for 13 AI visibility platforms", and the tools guide said "every named AI visibility product with public self-serve pricing is listed below". The table renders 14 rows, and one of them is BrightEdge, which publishes no price at any tier.
What is true
The counts are derived. The tables compare 14 products, 13 of which publish a price, and BrightEdge is named as the one that does not rather than being silently included in a count of published prices.
Why it happened
Two counts of the same set were typed by hand and neither was updated when Robot Visible's own row was added to the table. The claim also conflated "compared" with "publishes a price", which is precisely the distinction this comparison exists to make.
The AthenaHQ row of the quote-only pricing table read "Self-serve tiers are published from Free tier available; Starter $295/month".
What is true
That column now lists each published plan by name and price, so the AthenaHQ row reads "Essential Free, Starter $295/month". The prices were always right; the sentence around them was not English.
Why it happened
The cell was built by stripping a leading "From " off each vendor's one-line entry summary and splicing the remainder into a fixed sentence. That prefix is present on most summaries and absent from AthenaHQ's and Semrush's, so the template silently produced a broken sentence for the vendors that did not match its assumption. Deriving text from a string that was written for a different purpose is not the same as deriving it from data.
A draft comparison was about to state that Rankscale Essentials includes no brand dashboards and no page audits, and that page audits begin at Pro. The published entry was thinner but not wrong: "Basic monitoring with credits, unlimited search terms, all AI regions".
What is true
Rankscale Essentials includes 120 monthly credits without rollover, 2 brand dashboard slots, up to 480 AI answers, 10 page audits per month, unlimited search terms, 17+ engines, all regions and Scout recommendations. The draft claim was withdrawn before publication and the recorded allowances now match the vendor's rendered table.
Why it happened
The comparison table on Rankscale's pricing page renders zeros in every allowance cell until client-side data loads. Two separate automated reads returned that zero state, which looked like a vendor publishing an empty tier, and the second read appeared to confirm the first. Opening the page in a browser showed real figures. This is the third correction in this log caused by treating a partially rendered pricing page as complete, so allowance figures are now read from the rendered page rather than fetched HTML.
Peec AI publishes in euros, and its Starter plan is €85 per month. Both the figure and the currency were wrong. The full ladder is now published: Starter €85, Pro €205, Advanced €425, Enterprise custom. A claim that the Starter and Pro tiers carried country limits was also removed, because Peec publishes multi-country support only on Advanced.
Why it happened
The price was taken from a pricing page that renders its figures after load, so the tool used to read it returned partial content. The figure it produced was recorded and dated as verified instead of being treated as unreadable.
AthenaHQ was listed at $295 per month, or $245 per month billed annually, on a plan called Self-serve.
What is true
AthenaHQ does not publish an annual figure, and the plan is called Starter. The $245 was almost exactly 17% below the monthly price, matching the advertised annual discount, so it appears to have been calculated rather than read. The annual figure has been removed and the plan renamed. AthenaHQ's free Essential tier is now shown as well.
Why it happened
A discount percentage was applied by hand and the result published as a verified price. Plausible arithmetic is not verification, and the resulting number was never on the vendor's page.
Kelsey was recorded as publishing no pricing, offering only a free scan.
What is true
Kelsey publishes a full price list: Essential at $99 per month, Growth at $199 per month and Agency at $499 per month, with prompt allowances and content-brief volumes for each. The comparison was rewritten, because published pricing changed how the product compares.
Why it happened
Only the site's home page was checked. The pricing page was never requested, even though every other vendor in the set had its pricing page checked directly. Stating that a company publishes no pricing is a strong claim and it was made from weak evidence.
AI visibility was described as included across all Semrush paid plans, with suite prices listed from $117.33 to $455.67 per month.
What is true
AI visibility is not included in every paid plan. The $139.95 SEO tier explicitly excludes AI search tracking, AI sentiment, AI visibility reports and custom prompt monitoring. AI tracking begins at $199 per month, and Semrush separately sells a standalone AI Visibility plan at $99 per month per domain that had been missed entirely. Monthly suite prices are $139.95, $199, $299 and $549.
Why it happened
Another partially rendered pricing page was treated as complete. The error was worse than a wrong number because it was a wrong claim: a reader following it could have bought a plan that does not include the feature they wanted.
BrightEdge was described as having no public pricing page, cited to its home page.
What is true
BrightEdge does have a product pricing page. It is a contact form carrying no figures. The conclusion that BrightEdge publishes no prices was correct, but the wording and the citation were both imprecise and are now exact.
Why it happened
A guessed URL returned a 404 and that was treated as evidence the page did not exist. Absence of a page at one address is not absence of the page.
If a price, allowance or description of your product is out of date or unfair, tell us and it gets checked against your own published pages. Corrections are made whether or not the company asks, and they are logged here either way.