DataForAEOby Advance Labs

The registry of AI-search data

Data for AEO, measured at the source.

See which AI assistants read your pages because someone asked. Check any request against the vendor’s published IP ranges. Try it in the live panel: it calls the real API.

Verify a request

Live API, free

Try a request
VerifiedChatGPT-User · OpenAI · User fetch

The IP is in OpenAI’s published ranges. A person asked the assistant something just now. This is real demand for the page.

View response13 ms, captured example
{
  "ok": true,
  "request_id": "4a151b0c-2a9b-48ac-85d8-aac4d73c4fd4",
  "time_ms": 13,
  "cost": {
    "checks": 0
  },
  "result": {
    "ai_bot": true,
    "bot": "ChatGPT-User",
    "vendor": "openai",
    "kind": "user",
    "verification": "verified"
  }
}

Calls POST /api/v1/ai-crawlers/verify. No key, nothing stored.

AI agents classified
15
vendors IP-checked
5
IP addresses stored
0

Registry

The agents that fetch because someone asked

Every row is a user fetch: a person asked, so the page was read. 6 of the 15 agents we classify are user fetches. Each token links to what it does, how to verify it and what to put in robots.txt.

See all 15 agents
TokenVendorIP check
ChatGPT-UserOpenAIOpenAIPublished list
Claude-UserAnthropicAnthropicPublished list
Perplexity-UserPerplexityPerplexityPublished list
MistralAI-UserMistral AIMistral AIPublished list
DuckAssistBotDuckDuckGoDuckDuckGoNo list
Meta-ExternalFetcherMetaMetaNo list

“Published list” means the vendor lists its IP ranges, so a hit can be verified or unverified. “No list” means the hit stays unknown.

API

One call, one key

Every response has the same envelope and says what it cost. Both calls below need a key. The docs list every endpoint, its errors and its limits.

GET /demand · measured demand for your site

GET /v1/demand
curl 'https://dataforaeo.dev/api/v1/demand?domain=example.com&days=7' \
  -H "Authorization: Bearer $DFA_KEY"

POST /answers/live · ask the AI engines now

POST /v1/answers/live
curl -X POST https://dataforaeo.dev/api/v1/answers/live \
  -H "Authorization: Bearer $DFA_KEY" -H 'Content-Type: application/json' \
  -d '{"domain":"example.com","prompts":["best crm for small agencies"]}'

Method

A claim is not a count

Request, match the User-Agent to a known agent, check the IP against the vendor’s published ranges, then count by day, page, bot, kind and verification.01Request

A page is served to a client.

02Match the agent

The User-Agent is matched to a known AI agent.

03Check the IP

The IP is checked against the vendor’s published ranges.

04Count

A daily count by page, bot, kind and verification.

  1. 01RequestA page is served to a client.
  2. 02Match the agentThe User-Agent is matched to a known AI agent.
  3. 03Check the IPThe IP is checked against the vendor’s published ranges.
  4. 04CountA daily count by page, bot, kind and verification.

A User-Agent is a claim, so the IP behind it is checked against the published ranges of 5 vendors. Every count says verified, unverified or unknown, and a person’s question is never mixed with a search refresh or a training crawl. Read the details in the AI crawler list.

Collector

Free for your site

A few lines in a Next.js proxy or a Cloudflare Worker. It sends only AI-bot requests, after the response, so pages never wait. Your own pages’ AI demand shows up in AEO Toolkit, and the anonymous aggregate feeds the registry.

proxy.js (Next.js)
export function proxy(req, event) {
  const ua = req.headers.get('user-agent') ?? '';
  if (!AI_BOTS.test(ua)) return;
  const events = [{
    ua,
    path: req.nextUrl.pathname,
    ip: req.headers.get('x-real-ip'),
  }];
  event.waitUntil(fetch(COLLECT_URL, {
    method: 'POST', headers,
    body: JSON.stringify({ events }),
  }));
}

During the beta, API calls draw on your AEO Toolkit plan’s monthly checks. The crawler endpoints are free.

See pricing

Questions

Questions about data for AEO

What is data for AEO?

Data for AEO (answer engine optimization) is evidence about how AI assistants like ChatGPT, Claude, Perplexity and Gemini find, read and cite the web: how often they fetch a page because someone asked, what they answer, and whom they cite. DataForAEO collects it and serves it as an API.

Is AI search volume measured or modelled?

Both, and every number says which. Bot fetches are measured: each one is a request a site served. Keyword AI volume is modelled from Bing volume and is labelled an estimate. No AI company publishes query volume, so anyone selling it as measured is estimating too.

How do you know a hit is really ChatGPT?

A User-Agent can be faked, so each hit from OpenAI, Anthropic, Perplexity, Mistral or Amazon is checked against the IP ranges those 5 vendors publish and counted as verified, unverified or unknown. Bots whose vendors publish no list stay unknown.

Is DataForAEO affiliated with DataForSEO?

No. DataForAEO is built and run by Advance Labs Inc. in London, Ontario. It is not affiliated with, endorsed by or connected to DataForSEO OÜ.

What does it cost?

The crawler list and IP check are free and need no key. During the beta, the other API calls draw on your AEO Toolkit plan’s monthly checks, and the collector is free for your own site.

Know which AI assistants read your pages, and why.