SJK LABS RESEARCH · MAY 2026
How AI described 50 businesses.
The B2B AI Legibility Report examined how businesses were represented across ChatGPT, Claude, Gemini and Perplexity.
What the study found
- 32 of 50 companies had a wrong or incomplete competitive frame in at least one platform’s answers.
- 30 of 50 had a narrative or differentiation issue recorded.
- 23 of 50 had an entity recognition, name collision or wrong-company issue recorded.
- 17 of 50 had a proof, credibility or evidence-surfacing issue recorded.
These are findings recorded within the study sample. A company may appear in more than one group.
Study context and limitations
The study was conducted on 14 May 2026. It covered 50 regulated and capital-intensive businesses, including financial services, fintech, insurance, wealth, payments and crypto.
Company descriptions were collected through four web-search-enabled AI platforms and assessed against available company websites and audit notes. The same research approach was applied across the sample. Website access limitations were recorded where encountered.
This was a selected sample, not a representative estimate of all businesses. The results describe AI-generated answers at a particular time. They do not measure human buyer behaviour, sales, conversion, search rankings or the commercial effect of any website gap. Answers can change between platforms and over time.
This public summary presents the findings and study context. It does not include the detailed assessment protocol or scoring materials.
What this means for a business
The findings can help identify questions worth investigating about how a company is understood. A business-specific assessment is needed before deciding what to change.
Explore the AI Legibility Audit ↗
Discuss the research
For questions about the findings or a briefing on their relevance to your organisation, speak to Sarah.