What the evidence looked like, from three research measurements.
These are real AI Brand Intelligence Report™ measurements, run during SEOWEBI's initial testing programme. The case studies below present what was measured, what the evidence showed, and why it matters - not a sales story.
Three angles on evidence, one measurement standard.
Each case study is drawn from a separate, complete AI Brand Intelligence Report™ measurement: 36 prompts, four AI engines, 144 stored responses. All three identified the same underlying pattern - Incumbent Squeeze, a strong and accurate description that does not translate into presence in the questions that shape a shortlist - approached from a different angle each time.
AI repeated pricing claims that conflicted with published information.
The same incorrect price appeared across six AI-generated answers from three of four engines tested - while a different figure sat, uncited, on a page the company already publishes.
Read the case study →Accurate AI descriptions did not translate into recommendations.
Named and described accurately in almost every direct answer, then absent from every one of the twelve responses to the three questions buyers use to decide who to trust.
Read the case study →The way a comparison was framed changed the answer.
Favoured in every response when a comparison was anchored to the company's own differentiator - and out-favoured by one engine in three of four open, unanchored comparisons.
Read the case study →Evidence before interpretation, in every case.
- What was measuredEvery material finding traces to a specific prompt, engine, timestamp and stored response. Nothing below is inferred from page content alone.
- Confidence, stated plainlyEvery measurement referenced here carries an overall confidence label of Incomplete - full collection of all 144 responses does not mean every dimension of the score was fully measured, and each case study says which parts were not.
- Historical, not currentEach case study describes AI-generated answers recorded during that measurement's original testing window. AI systems change; none of this describes what an engine would say today.
Want to see the Report format itself, rather than a finding drawn from it? See a fabricated sample excerpt (or download the sample as a PDF) - both built with an invented company so no real evidence is needed to show the structure.
Questions about these case studies.
Are these real measurements?
Yes. Every finding in these case studies is drawn from an actual AI Brand Intelligence Report™ measurement - the prompts, engines, responses and figures are real. (The separate sample report excerpt is the only fabricated content on this site, and is clearly labelled as such.) None of these case studies imply endorsement, implementation of any recommendation, or measured business improvement.
Can I see the full, unredacted report?
No. The underlying reports and the companies they describe are kept confidential; these case studies present anonymised findings only. Running your own AI Brand Intelligence Report™ applies the same evidence standard to your company.
Why are the companies anonymised?
SEOWEBI anonymises every case study it publishes, as a matter of policy. These measurements were run during SEOWEBI's initial testing programme; the methodology and evidence discipline are identical to what every AI Brand Intelligence Report™ delivers today.
How current is this evidence?
Each case study describes AI-generated answers recorded during a single measurement window in 2026. AI systems change over time; none of this describes current or ongoing AI behaviour.
See what your own measurement would show.
The AI Brand Intelligence Report™ runs the same evidence standard shown in these case studies against your own company and named competitors.