Measurement and optimization for the AI answer engines — with the confidence intervals other dashboards quietly hide.
Click any row to see that domain's most recent report. Want to see your domain on this list? Run a check above.
In 2026, AI assistants are how people start searching — for products, articles, apps, services, experts, stories, code, news. Whatever you're trying to be found for, the engines diverge sharply in what they cite, the existing dashboards hide their measurement uncertainty behind a single number, and most "AI SEO agencies" rebranded from classic SEO in the last three months. We were not the first AEO tool. We are the first that publishes its methodology, names the conflicts other scanners hide, and gives you a shareable report a non-specialist can read without a glossary.
| Anchor | Source |
|---|---|
| Only 11% of domains are cited by both ChatGPT and Perplexity. One-size-fits-all AI optimization is broken. | Profound platform analysis of 680M citations |
| Cited in Google AI Overviews → +120% organic clicks per impression, +41% paid clicks. The lift is real and measurable. | Seer Interactive |
| Domain Authority correlates only r=0.18 with Google AI Overview selection. Classic SEO is not enough on its own. | wellows AI Overviews analysis (15,847 results) |
| 80% of URLs ChatGPT cites do not appear in Google's top 100 results for the same query. | averi.ai · B2B citation benchmarks |
| If you're B2B specifically: 89% of buyers use AI for vendor research; the top quartile of SaaS brands earns 8.4× more AI citations than the bottom. | Averi B2B Citation Benchmarks 2026 |
Every check returns a one-page diagnostic at a permanent URL. Static-scan today; live engine probes across ChatGPT, Claude, Gemini, and Perplexity arrive next. Five things we surface that other scanners don't:
GPTBot is OpenAI's training crawler. OAI-SearchBot is the search index ChatGPT actually reads at inference. When you've blocked one expecting the other — or taken down both via a wildcard rule — we name the conflict by its actual mechanism, not a generic warning.
Training crawlers vs search-index crawlers vs user-initiated crawlers. Most scanners check 3–4 bots and lump them together. We cover OpenAI (3 bots), Anthropic (4), Perplexity (2), Google, Meta, Apple, ByteDance, Common Crawl — with vendor docs cited.
Anthropic Claude Desktop and Claude.ai respect llms.txt. Google has explicitly confirmed it does NOT (Gary Illyes, Search Central Live July 2025). OpenAI is unconfirmed. Most scanners tell you to ship llms.txt and imply it moves ChatGPT visibility. We tell the truth.
The Relixir 2025 study measured 41% citation rate for pages with FAQPage schema vs 15% without — a 2.7× lift. Most scanners surface FAQPage as one checkbox among many. We surface it as the single most leveraged thing you can ship this week.
Other scanners give you a 50-item to-do ordered by how easy it is to flag. We order by what would actually move citation share if you fixed it — calibrated against 2026 empirical work on retrieval mechanics.
Every report lives at a permanent canaifind.com/r/{slug} URL with an Open Graph preview rendering the verdict block as a static image. Screenshot it. Drop it in your team Slack. Quote-tweet it. Send it to whoever's going to ask. The reports are how this tool finds the next person who needs it.
There's an entire ecosystem emerging around AI readiness and it's easy to get overwhelmed. Each tool does a different job. We do the diagnostic. Use what fits the rest of your stack — these don't replace each other.
| Tool | What it answers | When to run it |
|---|---|---|
| Can AI Findyou are here | Do ChatGPT / Claude / Perplexity cite this brand? A static-scan diagnostic with methodology and confidence intervals. | Free, no signup. Start here to see if the AEO fundamentals are in place. |
| Cloudflare IsItAgentReady | Can AI agents authenticate against this site's APIs? OAuth/OIDC discovery, MCP server cards, WebMCP, agent-skills, payment protocols. | Free. Run alongside us if the site publishes APIs for autonomous-agent integration. Different layer; both matter. |
| Profound / Otterly / Peec | How is your AI citation share changing over time? Continuous monitoring dashboards across multiple engines. | $700–2K/mo. Run when you have an ongoing AEO program and need a time-series with prompt-set tracking. |
| AthenaHQ | Measurement + AI-driven recommendations + content production in one place. | Enterprise pricing. Run when you want execution help, not just measurement. |
| Ahrefs / Semrush / etc. | Where do you rank in traditional Google results? The classic SEO toolkit. | Run for the discoverability foundation — still load-bearing, even though it doesn't tell you anything about AI citations. |
Every other free "AI SEO checker" we've seen is a paint job over a classic SEO scanner — same UX, recoloured. Five things we surface specifically because the answer-engine path is not the search-engine path:
Other scanners say 'AI bots blocked.' We name the specific pattern: GPTBot (training) explicitly disallowed plus OAI-SearchBot (live retrieval) disallowed by fall-through. That's the actual bug, not a generic warning.
Training vs. search-index vs. user-initiated vs. developer-tooling. OpenAI (3), Anthropic (4), Perplexity (2), Google, Meta, Apple, ByteDance, Common Crawl — with vendor docs cited and the GPTBot-vs-OAI-SearchBot distinction visible on every row.
Anthropic respects it. Google has confirmed it does NOT (Gary Illyes, July 2025). OpenAI is unconfirmed. Most scanners imply llms.txt moves ChatGPT visibility. We tell the truth.
Pages with FAQPage JSON-LD show 2.7× citation rate vs. without (Relixir 2025: 41% vs. 15%). Most scanners surface FAQPage as one checkbox among many. We surface it as the single most leveraged thing you can ship this week — with the empirical anchor cited.
Every report names the methodology version it was generated under (currently v0.1 preview). The full method cites Sielinski 2026, Schulte 2026, Sharma 2026, Aggarwal 2024. When the method changes, old reports stay anchored to the version they ran on.
The metrics align with peer-reviewed work on AI-visibility measurement. Every cycle, every cell, reported with bootstrap CIs and excluding unverifiable URLs.
| Metric | What it counts | Typical CI | Anchor |
|---|---|---|---|
| Inclusion Rate (IR) | % of runs where the model mentions your brand at all. | ±3–5pp | Pooled over intent cluster × model × region per §4.2 |
| Citation Coverage (CC) | % of runs that cite a verifiable URL on your domain. Hallucinated URLs excluded. | ±2–4pp | HEAD/GET verified per §4.3 |
| Share of Voice (SoV) | Your IR vs named competitors' IR on the same prompt set. | varies | Per-engine and blended |
| Position score | When mentioned, where in the answer (lead / body / footnote). | ±0.4 | Weighted, 0–5 scale |
| Sentiment | Per-mention classification (positive / neutral / negative / mixed). | ±5pp | With classifier-agreement CI |
| Citation diversity | How many distinct URLs on your domain are being cited. | varies | Deep links count more than homepage-only |
Single-run visibility metrics provide a misleadingly precise picture of domain performance in generative search.
That's Sielinski, 2026. Our entire methodology is built around that finding: every metric sampled, pooled across intent clusters, reported with a bootstrap confidence interval. The hallucinated URLs flagged and excluded. The model snapshots detected and re-baselined. The competitor SoV computed per-engine because the engines diverge so sharply.
That's how this tool finds the next person who needs it. Free forever for one domain a day. No signup; no card; no Cloudflare dot-grid; no purple gradient. Just the answer.