What we can give a reporter
- Pilot index: 35 B2B SaaS brands, 105 buyer-question answers (4 Nov 2026). 73% of answers named the brand; 29% cited the brand's own site. Per-brand pages: ai-visibility-index.pages.dev.
- Variance study: 15 buying questions × 3 runs = 45 answers. Brand-set Jaccard 0.843; identical brand sets in 6 of 15. Full method and raw CSV on the variance pilot page.
- Benchmark page: AI visibility index by category, with the question list and citation owners.
- Method note: how the measurement works, including why one run is a sample and three runs are the unit we sell.
Assets you can link to
- Sample white-label report — the exact format a client or an agency receives.
- AI visibility index by category — 32 brands, 96 answers, per-category tables.
- 45-answer variance pilot — why a single run is not a measurement.
- Method note and machine-readable summary (llms.txt).
Quotable findings
- Being named is table stakes: most brands in the pilot were named at least once.
- Ownership is the gap: many named brands were cited zero times; the answer quoted a third-party comparison page instead.
- Answers move between runs: the brand set was identical in only 6 of 15 repeated questions, which is why a single screenshot is not a measurement.
How to cite
Please attribute to VisibleDiagnostic and link to the exact page you use. Call the data a pilot (one logged-out run per question on 4 Nov 2026) unless the paid, dated diagnostic is what you are describing. We do not have named client results to share; the published data is ours.
Interview contact
Steve Duan, founder. Email hello@visiblediagnostic.com with your outlet, deadline and the angle; we answer within one business day. If you need a fast background call, say so in the subject line.