How Can I Measure Whether My Business Appears in Meta AI Search?
To measure whether your business appears in Meta AI, ask it the questions your customers ask, on the surfaces they use. Run each question several times and log which businesses are named and which sources are cited. Meta AI is built into WhatsApp, Instagram, Messenger and Facebook, and is available in its own app and on the web. Meta offers no visibility dashboard for businesses, so manual or tool-based sampling is the main method. Add two technical checks. Look in your server logs for Meta-WebIndexer, the crawler Meta uses to index pages for Meta AI search. Make sure robots.txt and your CDN allow it, since Meta says allowing it helps Meta AI cite and link to your content.
Where does Meta AI get its answers?
| Source | What it contributes |
|---|---|
| Web search partners | Meta said in 2024 that Meta AI uses real-time search results from Google and Bing for current information |
| Meta's own web index | Built by the Meta-WebIndexer crawler for Meta AI search results and citations |
| User-triggered fetches | Meta-ExternalFetcher retrieves pages when a user's request needs them |
| Model knowledge | What Meta's models learned in training, including from pages crawled by Meta-ExternalAgent |
Because Meta AI draws on search partners as well as its own index, your visibility in Google and Bing still shapes what it can find. Meta has also been building its own search capabilities, so allowing its indexing crawler is increasingly important.
Which Meta crawlers should you allow?
| Crawler | Purpose | Effect of blocking |
|---|---|---|
| Meta-WebIndexer | Indexes pages for Meta AI search results; Meta says allowing it supports Meta AI citing and linking to your content | Reduces your chance of being cited in Meta AI answers |
| Meta-ExternalFetcher | Fetches pages in response to user requests | A firewall or CDN block stops Meta AI reading your page when a user asks about it; robots.txt may not, since Meta says this fetcher may bypass it |
| Meta-ExternalAgent | Collects content for AI uses such as training and indexing | Keeps content out of model training, with possible indexing effects |
| facebookexternalhit | Generates link previews when pages are shared | Broken link previews on Meta's apps |
Many sites that block AI training crawlers still allow indexing and user fetches. Blanket rules that block every Meta user agent can remove you from Meta AI answers and break link previews. Check your CDN and firewall as well as robots.txt, since bot protection can block these crawlers even when robots.txt allows them.
How do you run the checks?
- Write 15–30 buyer questions, including local ones with the town stated: "Best family dentist in [town]?", "Who does emergency roof repairs in [area]?", "Which accounting software is good for freelancers?"
- Choose surfaces: the Meta AI app or website, plus the in-app assistant in WhatsApp or Instagram if your customers use those most.
- Ask each question two or three times in fresh conversations.
- Record the businesses named, your position, what is said about you, and any sources or links shown.
- Ask brand questions: "What does [business] do?" and "Is [business] in [town] any good?", and check accuracy.
- Repeat monthly with the same questions.
Keep a simple log with one row per answer: date, surface, question, businesses named, whether you appear, sources and accuracy notes.
How do you check the technical side?
- Server or CDN logs: look for requests from Meta-WebIndexer and Meta-ExternalFetcher, which pages they fetch and whether they receive 200 responses.
- robots.txt: confirm you are not disallowing Meta-WebIndexer unintentionally, for example through a rule covering all AI bots. Meta-ExternalFetcher may ignore robots.txt because a user requested the fetch, so check your CDN and firewall rules for it instead.
- Rendering: make sure key content appears in the page HTML, since crawlers may not run scripts.
- Analytics: watch for referral visits from Meta AI, which may appear under Meta's domains or link redirects, and compare them with your check results.
Why do Meta AI results vary?
- Surface: answers in a chat app may be shorter than in the Meta AI app.
- Location and language: local questions depend on location, and availability differs by country.
- Search source: which partner or index supplied the results.
- Generated answers: as with other assistants, answers vary from run to run.
Treat each check as a sample, and look for patterns across runs and months.
What should you fix if Meta AI does not mention you?
- Allow Meta-WebIndexer and Meta-ExternalFetcher in robots.txt and at your CDN.
- Strengthen Google and Bing presence for your buyer questions, since Meta AI has drawn on both.
- Keep your Facebook Page and Instagram profile complete and consistent with your website: name, category, address, hours, services.
- Publish clear pages answering the questions customers ask.
- Ask every customer for an honest review on the cited platforms that allow businesses to ask; not Yelp, which forbids it.
The same question set can be run in other assistants too. A similar check for DuckDuckGo is covered in how to check whether your business appears in DuckDuckGo AI answers.
What do people ask about Meta AI visibility?
Does Meta offer a dashboard for AI visibility?
Not for general businesses. Measurement relies on sampling questions and checking your logs and analytics.
Does my Facebook Page affect Meta AI answers?
Keep it accurate. Consistent public information across your Page, website and listings reduces the chance of wrong answers.
Should I block Meta-ExternalAgent?
That is a policy choice about AI training. If you block it, still allow Meta-WebIndexer and Meta-ExternalFetcher if you want to appear in answers.
Is Meta AI available everywhere?
Availability and features vary by country and language. Test in the markets where your customers are.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources