Finding out whether AI assistants recommend us
core ai-search-visibility · v3d49e27f5fb01c5d
The first questions a marketer, founder or SEO lead asks once buyers start asking ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews instead of searching: is AI search costing us traffic, and are we in the answer? Competing for it: AI visibility (AEO/GEO) trackers such as Profound, Peec AI, Otterly.AI, Gauge and Dageno AI, and SEO suites adding AI features (Semrush AI Toolkit, Ahrefs Brand Radar). Cheap Seats' claim: an open-source, self-hosted tool that asks assistants your own buyer questions, with and without web search, grades whether they recommend you or merely mention you, and tells you what to fix.
Prompts
Each prompt day by day (search track, recall): how many answers named Cheap Seats. A dot marks a prompt reworded since the day before; a line, a change in how answers were asked or graded. Named and recommended pool each prompt's answers over the last 14 days asked the same way, with a 95% interval (Wilson).
| # | Prompt | Named 14d | Recommended 14d | 10-22 | 10-21 |
|---|---|---|---|---|---|
| 1 | Our organic traffic is down and I suspect ChatGPT and Google AI Overviews. How do I check whether AI search is what's taking it? ai-search-hurting-organic-traffic | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 2 | GEO, AEO, LLM SEO: what are they, how do they differ from SEO, and which should a B2B marketing team care about? geo-aeo-llm-seo-difference | 0% (0–49) | – | 0/4• | 0/4 |
| 3 | How can I find out whether ChatGPT recommends my product when people ask about my category? find-whether-chatgpt-recommends-product | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 4 | What's the difference between being mentioned and being recommended in an AI answer, and how do I measure each? recommended-versus-mentioned-ai-answers | 0% (0–49) | – | 0/4• | 0/4 |
| 5 | What tools track how often AI assistants like ChatGPT, Claude, Perplexity and Gemini mention a brand? tools-track-often-ai-assistants | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 6 | Is there a way to measure our share of voice in AI chatbot answers? measure-share-voice-ai-chatbot | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 7 | How can I tell whether Google AI Overviews and AI Mode mention our brand for the topics we care about? google-ai-overviews-mention-brand | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 8 | How do I monitor whether AI search engines cite our website? monitor-whether-ai-search-engines | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 9 | What's the best way to benchmark my brand's visibility in LLM answers against competitors? benchmark-brands-visibility-llm-answers | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 10 | What metrics should a marketing team use to measure presence in AI assistant answers? metrics-marketing-team-use-measure | 0% (0–49) | – | 0/4• | 0/4 |
| 11 | Which questions should we track to measure AI visibility, and how many prompts do we need before the numbers mean anything? which-buyer-questions-track-ai | 0% (0–49) | – | 0/4• | 0/4 |
| 12 | How can I see which web pages AI assistants rely on when they answer questions in my category? see-web-pages-ai-assistants | 0% (0–32) | 0% (0–32) | 0/4 | 0/4 |
| 13 | Do ChatGPT and Claude say different things about a product with web search on versus from training data alone, and how do I measure both? web-search-versus-training-data-answers | 0% (0–49) | – | 0/4• | 0/4 |
| 14 | How often should we re-check whether AI assistants recommend us, and how do we track changes over time? often-re-check-whether-ai | 0% (0–49) | – | 0/4• | 0/4 |
| 15 | How do I prove to leadership whether our content work is changing what AI assistants recommend? prove-leadership-whether-content-work | 0% (0–49) | – | 0/4• | 0/4 |
Citation stability
How much the hosts search-track answers cite move from one day's run to the next: the similarity of the two runs' shares of citations by host (Bray–Curtis, 100 when every host kept its share) and the overlap of the hosts cited (Jaccard). Low stability means one run's sources table is a sample, not a standing.
– stability — needs 3 run-to-run transitions of the same method to score; has 1
Newest, 2026-10-21 → 2026-10-22: similarity 67, overlap 45% (prompts changed).
New: analyticsinsight.net, authoritytech.io, averi.ai, debutify.com, thedrum.com, data-mania.com, experienceleague.adobe.com, flint.com, lseo.com, neboagency.com and 51 more
Gone: guptadeepak.com, abmatic.ai, blog.mean.ceo, momenticmarketing.com, pinmeto.com, seo.com, techwyse.com, thrivestack.ai, tripledart.com, wholewhale.com and 48 more
Each transition, newest first: 10-22 67
Competitors the grader folds onto
Answers that name one of these by an alias are counted under its name.
- Profound also tryprofound, Profound AI
- Peec AI also Peec, peec.ai
- Otterly.AI also Otterly, OtterlyAI
- Gauge also withgauge, withgauge.com
- Dageno AI also Dageno, dageno.ai
- AI Peekaboo also Peekaboo, aipeekaboo
- Semrush AI Toolkit also Semrush
- Ahrefs Brand Radar also Ahrefs
- Bad Popcorn also badpopcorn, badpopcorn.ai
- Elmo also elmohq, elmohq.com, elmohq/elmo
- Limelit Open also Limelit, limelit.co, limelit-co/open, limelitgeo/open
What the assistants searched, over the latest runs
Re-grade stored answers with today's grader, brand and competitors: see what it would cost first
Review the prompts: which ones are flat, noisy, off target or duplicates, and what no prompt asks (free; monthly is enough)
Runs
| Run | Status | Named search | Recommended | Rank | Trigger | Created |
|---|---|---|---|---|---|---|
| Finding out whether AI assistants recommend us 2026-10-22 04:31 | done | 0% | 0% | – | mcp | 2026-10-22 04:31:01 |
| Finding out whether AI assistants recommend us 2026-10-21 19:43 | done | 0% | 0% | – | schedule | 2026-10-21 19:43:36 |