How to Track Your Brand in Meta AI (2026)
Meta AI answers questions inside WhatsApp, Instagram, Facebook, Messenger and its own app. How to build a prompt set for it, what to record per answer, which Meta crawlers to check, and which tools actually cover Meta AI.
Most AI visibility programs started with ChatGPT, then added Perplexity and Google's AI Overviews. Meta AI usually came last, if at all. That made sense when Meta's assistant was mostly a chat box for writing captions. It makes less sense now.
Meta's help center says the Meta AI assistant can be used in the Meta AI app, on meta.ai, and on Facebook, Instagram, Messenger and WhatsApp, with availability and features varying by country (Meta Help Center). That puts it inside apps people already have open when they ask "which one should I buy". And Meta's crawler documentation now describes a crawler whose job is to help Meta AI cite and link to web pages. So for a brand, Meta AI is an answer engine with sources, and you can track it the same way you track the others.
This guide covers how to set that up, what to record, and which tools in our rankings cover Meta AI at all.
Step 1: decide whether Meta AI belongs in your engine set
Not every brand needs it on day one. Promptwatch's own recommended starting set is ChatGPT, Perplexity and Google AI Overviews, adding AI Mode, Copilot or Alexa depending on the market. Meta AI is not in that default list.
Add it when your buyers are consumers who spend their day in Meta's apps: retail, travel, food, local services, beauty, entertainment. A B2B infrastructure vendor can start without it. A restaurant group or a fashion label probably shouldn't.
The cost is responses, not a separate add-on, on the tool we rank first. More on that below.
Step 2: write prompts the way people ask Meta AI
Your ChatGPT prompt set is a fair starting point, but trim it. People reach Meta AI from a chat thread or a search bar inside a social app, so expect shorter questions, more "near me", more "is X worth it", and more gift and recommendation phrasing.
Split the set into three groups:
- Category prompts with no brand named ("best running shoes for flat feet")
- Comparison prompts that name you and a rival ("Brand A or Brand B for a first espresso machine")
- Brand prompts that check what Meta AI says about you directly ("is Brand A legit")
Keep the category group biggest. That is where you find out whether Meta AI recommends you unprompted, which is the question a visibility program exists to answer.
Step 3: record the whole answer
For each prompt and each day, you want five things:
- Whether your brand was named
- Where it appeared in the answer, first or fifth
- Which competitors were named instead
- Which URLs were cited, if any
- The tone of what was said about you
The URL column matters more than people expect. Meta's documentation for its Meta-WebIndexer crawler says it "navigates the web to improve Meta AI search result quality for users" and that allowing it "helps us cite and link to your content in Meta AI's responses" (Meta web crawlers). If Meta AI cites a listicle that ranks you seventh, that page is what you fix or replace, not your homepage.
Step 4: check the Meta crawlers on your own site
Meta documents several crawlers, and they do different jobs:
| Crawler | What Meta says it does | Respects robots.txt? |
|---|---|---|
| Meta-WebIndexer | Improves Meta AI search results; helps Meta AI cite and link to your content | Yes |
| Meta-ExternalAgent | Training foundation AI models or improving products by indexing content directly | Yes |
| Meta-ExternalFetcher | Fetches individual links at a user's request | May bypass it |
| FacebookExternalHit | Fetches pages shared in Meta's apps, for link previews | May bypass it for security checks |
Source: Meta web crawlers, read October 28, 2026.
Two practical points come out of that table. First, blocking Meta-ExternalAgent because you don't want your content used for training is a separate decision from blocking Meta-WebIndexer, which is the one tied to citations. A blanket "block Meta" rule removes you from both. Second, a robots.txt file only tells you what you asked for. It doesn't show what actually fetched your pages. For that you need server or CDN logs.
How much this matters became obvious in August. Promptwatch's Meta-WebIndexer report found the crawler's share of tracked AI crawler requests went from about 2.2 percent in mid-July 2026 to 37.8 percent on August 9. We covered the methodology in our Meta-WebIndexer breakdown.
Step 5: pick a tool that tracks Meta AI the way people see it
This is where most of the category falls short. Plenty of trackers list Meta AI somewhere, but often only on Enterprise or as an add-on.
Promptwatch is our first pick. As of its October 28, 2026 llms.txt, Meta AI is one of the live-search platforms it monitors through the real user interface, next to ChatGPT, Google AI Overviews, AI Mode, Copilot, Gemini, Perplexity and Alexa. That matters because the interface is what a WhatsApp user sees, web search included. Every paid plan unlocks all active models, so Meta AI is available from Essential at $95/mo. The free Explore plan is ChatGPT only. Models are picked per monitor with no cap on the number, only on responses. So budget it as prompts times engines times days: 50 prompts on four engines, run daily for 30 days, is the 6,000 responses Essential includes.
The part the trackers can't match is step 4. Promptwatch's Agent Analytics logs Meta-WebIndexer, Meta-ExternalAgent and Meta-ExternalFetcher, verified against Meta's IP ranges so spoofed bots are filtered out, and shows a crawl-to-citation path per page. Crawler logs start on Professional at $245/mo. Essential has none.
Rankability includes Meta AI among 11 platforms on every plan from $97/mo, checked daily for mentions, citations and sentiment. Its listing notes the crawler check reads robots.txt evidence rather than server logs. That is a real limit for Meta, since the fetcher that acts on a user's request may ignore robots.txt anyway.
OmniSEO tracks Meta AI, but only on Professional at $349/mo. Essentials at $89 covers four engines without it.
Profound lists Meta AI among up to 10 engines on Enterprise, which our listing puts at around $40k a year.
LLM Pulse sells Meta AI as an Enterprise add-on. Ahrefs Brand Radar has no Meta AI in its native indexes.
Step 6: read it weekly, act monthly
Daily data is noisy. Look at Meta AI weekly next to your other engines, and look for the gaps between them. If ChatGPT names you and Meta AI doesn't, compare the cited sources for the same prompt. That usually points at one or two pages, or at a crawler that never got through.
Once a month, take the three biggest gaps and fix something concrete: a page that answers the question plainly, a robots.txt rule, a listicle you can get added to. Then check whether the citation follows.
Common mistakes
- Blocking all Meta crawlers to opt out of training, and losing citations with it
- Tracking Meta AI through an API model and assuming it matches what app users see
- Copying a 300-prompt ChatGPT set into Meta AI and burning the response budget on B2B questions nobody asks in WhatsApp
- Reporting mentions without the cited URL, so nobody knows what to fix
Where to start
If you already track ChatGPT and Perplexity, adding Meta AI is a monitor change, not a new tool. Start with 20 category prompts, check your robots.txt for the four Meta user agents above, and give it two weeks before drawing conclusions. If you're choosing a platform, Promptwatch is the one we'd trial first: real-UI Meta AI tracking from $95, plus the Meta crawler logs that explain why an answer did or didn't cite you. Compare the rest in our full rankings.