AI repeated pricing claims that conflicted with published information.
In one AI Brand Intelligence Report™ measurement, three of the four AI engines tested stated the same incorrect price for the company - across six separate answers - while a different, correct figure sat uncited on a page the company already publishes.
Buyers increasingly ask an AI assistant what a vendor costs before they open a pricing page or speak to sales. This measurement tested what four AI engines - ChatGPT, Gemini, Perplexity and Claude - actually say when asked about the company's price, both directly and inside comparison-style prompts against named competitors.
What SEOWEBI measured36 prompts were run against each of the four engines - 144 stored responses in total, spanning discovery, recommendation, direct comparison and deep-dive questions about the company. Every response was stored in full, with the prompt, engine and timestamp attached, so each figure below traces back to the answer that produced it.
Each square is one stored AI response (144 total). The six orange squares are the responses that repeated the same incorrect price, from three of the four engines tested.
Two supporting facts sit behind that finding. First, the company's own pricing page states a different figure than the one circulating - the two are not the same number stated two ways, they are two distinct claims, and only one of them is correct. Second, that correct figure has not yet been picked up by any of the four engines tested: the gap is not that the truth is unavailable, it is that the correct number has not reached the sources these engines already draw from.
Why it mattersObservation: six of the stored AI responses state a price the company does not charge, drawn from a competitor's published comparison content rather than the company's own pages.
Interpretation: because AI-generated answers can shape a buyer's price expectation before any sales conversation starts, a repeated incorrect figure is the kind of gap that plausibly matters commercially. The measurement does not, and cannot, establish that any specific deal was affected - that would go beyond what this evidence supports.
The priority the report recommendedThe report's own 90-day roadmap listed correcting the pricing record at its source as the immediate, first-priority action - the single "quick win" ahead of every other recommendation, on the basis that one unambiguous, machine-extractable price is the cheapest correction available. This reflects what the report recommended at the time of delivery. SEOWEBI has no record of any correction, republication or remeasurement having taken place since.
Measurement limitations and what remains unknownThis report's overall confidence label is Incomplete: a small number of sub-signals, worth a minority of the total 100-point score, could not be verified from available data and are scored conservatively as zero rather than estimated. The pricing finding itself sits inside the AI citation layer, which was fully and completely measured - all 144 prompt-engine pairs returned a stored response, with no gaps. What remains unknown is whether the pattern has since changed: this case study describes AI-generated answers recorded during a single measurement window in 2026, not current or ongoing AI behaviour.
Where this fitsThis finding is one expression of the pattern SEOWEBI calls Incumbent Squeeze - a brand crowded out of category-defining answers despite accurate description once it is named. All three case studies in this collection identify the same underlying pattern from a different angle.
This is what an AI Brand Intelligence Report™ is built to surface.
The same evidence standard, confidence labelling and traceability shown in this case study applies to every AI Brand Intelligence Report™ - run against your own company and named competitors.
More anonymised research: Description vs. recommendation · Comparison framing · All case studies