State of AI Search Q2 2026
Quarterly benchmarks of brand visibility inside ChatGPT, Claude, Gemini, Perplexity, and Grok - computed directly from real AI responses logged by the Livesov platform.
Track your own brand freeThe Q2 2026 edition is being compiled
This report publishes only when the trailing-90-day sample clears our minimum thresholds (so the numbers are statistically meaningful and no customer is identifiable). Leave your email below and we'll send the edition the moment it goes live.
Stay current
Get each edition in your inbox
This check is a snapshot - AI answers change weekly.
Get monthly AI visibility benchmarks and the GEO tactics that are working, free. Unsubscribe anytime.
Methodology
Livesov continuously sends category prompts ("best CRM for small business", "top accounting software", and so on) to ChatGPT, Claude, Gemini, Perplexity, and Grok on behalf of the brands it tracks, and stores every full response. This report aggregates those logged responses over a trailing 90-day window:
- Mention rate- share of a brand's tracked prompts where the platform named the brand in its answer.
- Recommendation rate - share where the platform actively recommended the brand rather than merely mentioning it.
- Sentiment - per-response classification (positive / neutral / negative), reported over labeled responses only.
- Citations - URLs the platform cited; aggregated by domain with tracked-brand domains removed.
- Accuracy issues- responses contradicting a brand's verified canonical facts, normalized per 1,000 responses.
Cached duplicate responses are excluded so repeated identical answers don't inflate counts. Platforms appear only when both sample-size and brand-diversity thresholds pass.
Republish freely - with attribution
Every figure on this page may be quoted or republished with attribution to "Livesov State of AI Search" and a link to livesov.com/research/state-of-ai-search. Tell us where you used it via the contact page and we'll link back to your work.
State of AI Search - FAQ
Where does this data come from?+
From the Livesov platform itself: every figure is computed from real, logged AI responses collected while tracking customer brands across ChatGPT, Claude, Gemini, Perplexity, and Grok. No survey data, no estimates - these are measured responses.
Is customer data exposed in this report?+
No. Only aggregates are published. A platform's numbers appear only when enough distinct brands contribute to the window that no single brand is identifiable, and domains owned by tracked brands are excluded from the citation table entirely.
Can I republish these numbers?+
Yes - every figure on this page is free to republish with attribution and a link to Livesov State of AI Search (livesov.com/research/state-of-ai-search). If you cite us in a study or article, tell us via the contact page and we'll link back to your work.
How often is this page updated?+
The figures are recomputed from the live database on a rolling basis and always describe the trailing 90 days, so the page is effectively a continuously-updated quarterly report.
Why do mention rates differ so much between platforms?+
Each platform retrieves and synthesizes differently: some ground answers in live web search (Perplexity), others lean on training data (Claude), and some blend both. Different retrieval means different brand sets surface for the same question - which is exactly why single-platform tracking is misleading.