Meta AI is already integrated into Instagram search and direct messages, and when a user asks for a recommendation in your sector, the system suggests specific brands — among which you probably aren't, but your competitors are. Instagram has the largest user base in Italy, and you're handing over visibility right there every day without knowing it. Becoming visible to Meta AI requires different steps from those for ChatGPT, but the results are measured against a huge audience.
Instagram has integrated Meta AI into DMs and the search bar. Every query like “best Pompeii experiences” or “Herculaneum tour in Italian” triggers Meta AI inside the app: a generated answer, with suggested brands, before the user even scrolls the posts.
Your competitors are there. You, probably, are not. And no one told you, because Meta AI doesn’t show up in Google Search Console and leaves no trace in your referrers.
In this series on AI engines I’ve already explained how ChatGPT, Gemini and Perplexity work. Today I’m taking you to a different channel, based on Llama (Meta’s model) and integrated inside WhatsApp, Instagram and Facebook. Billions of active users, citation rules entirely its own, training data that doesn’t coincide with OpenAI’s or Google’s.
For anyone selling experiences to a tourist audience — archaeological tour operators, licensed Pompeii-Herculaneum guides, small brands of the Vesuvius area — it’s the channel where the user actually looks for you. And where today you aren’t.
What changes when the AI engine lives inside a social network
ChatGPT is a separate website. Gemini sits in Google’s SERPs. Perplexity is a vertical app. Meta AI doesn’t: it lives inside the same feed where the user is already scrolling your posts.
When a German traveler opens Instagram in Naples and types “best Pompeii guided tour Italian English” into the search bar, two things happen at once. The classic results appear (accounts, hashtags, reels) and, above or beside them, an answer generated by Meta AI with brand, account and sometimes external-link suggestions.
The answer isn’t infinite. It’s 2-4 names, rarely more. If your tour operator isn’t among those names, the user doesn’t see you — even if you have 8,000 followers and viral reels.
Here lies the difference from the other AI engines: the competition for the citation isn’t the entire web, it’s the subset of brands that Meta AI considers relevant within its own ecosystem. Verified business accounts, active Facebook pages, engagement signals consistent with the sector.
Why the signals that matter are different from Google
In previous articles I talked to you about E-E-A-T for AI and about backlinks as a citation proxy. For generalist AI engines (ChatGPT, Gemini, Perplexity) those web signals carry a lot of weight: who talks about you, where, with what authority.
For Meta AI the picture shifts. Llama was trained on the public web + corpora that Meta owns (public posts, business pages, public Instagram content). And in the in-app integrations, the system can draw in real time on signals from the platform itself: who posts what, who is verified, who has an active business page with reviews and bookings.
Translated for a Pompeii tour operator: if your Facebook business page is stuck in 2022, if the Instagram account doesn’t have the “Tour Agency” category set correctly, if you have no recent public reviews — you’re less citable than a competitor with half your followers but a well-maintained business profile.
It’s not a magic factor, of course. The Meta signal doesn’t replace site + schema + web authority. But on this specific channel, where the user is already inside the app, it’s the main lever.
I’ve seen licensed guides categorized as “Public Figure” or “Personal Blog”.
The reverse engineering I recommend you do in 30 minutes
This is a test anyone can run. You don’t need any paid tool.
- Open Instagram on mobile (the Meta AI integration works best in the app)
- Go to the search bar and type in 5 realistic queries for your sector. For Pompeii and the surrounding area, for example:
- “Pompeii tour in Italian”
- “Herculaneum half-day guide”
- “Pompeii visit with kids”
- “Vesuvius archaeological tour”
- “best Pompeii tour Italian guide”
- For each query, see whether an answer generated by Meta AI appears (sometimes it shows up as a card, sometimes as a message if you start a chat with @Meta AI)
- Note down: which brands/accounts get named? Are they verified business accounts? Do they have the correct category? Do they have Facebook reviews?
- Repeat inside WhatsApp by writing the same queries to @Meta AI
Honest caveat: this is an entry level check, indicative. Meta AI’s answers vary by geography, user’s language, account history. For a serious coverage analysis on Meta AI you need professional tools and a much larger sample of queries. But 30 minutes of testing already tell you whether you’re in or out, and which competitors are occupying the space.
If you sell archaeological tours in Pompeii, it must be “Tour Agency” or “Travel Company”, not “Personal Blog”
The test I ran myself (and what came out)
I ran this reverse engineering on 12 tourist queries tied to the Pompeii-Herculaneum-Vesuvius area, in Italian and English, from an Instagram account different from the ones I manage professionally to reduce personalization bias.
Pattern that emerged on a small but clear sample: the 3-4 brands that recurred in the Meta AI answers all had three characteristics in common. A verified and active Facebook business page (posts in the last 2 weeks). A consistent Instagram category (“Tour Agency”, “Travel Agency”). A public reviews section with at least 50 reviews and replies from the brand.
The tour operators with more followers but a poorly maintained business profile — generic “Personal Blog” category, a stale Facebook page, no public reviews — never appeared, not even on the queries in Italian.
A note on method: 12 queries are not a study. It’s an operational observation that helps me formulate a hypothesis, not close a case. If in your sector you want to seriously measure coverage on Meta AI, you need 50-100 queries and repetitions over time.
The mistakes I see most often in the area’s tourism brands
I often work with businesses in the South that sell experiences (guides, tour operators, experiential B&Bs). The patterns that keep recurring are always these:
- Abandoned Facebook business page. All the investment is on Instagram and TripAdvisor. For Meta AI it’s a weak signal: the system sees a brand that lives only halfway inside its ecosystem.
- Wrong Instagram category. I’ve seen licensed guides categorized as “Public Figure” or “Personal Blog”. For “tour” or “guide” queries, Meta AI doesn’t pick you up.
- Dormant Facebook reviews. You have 400 five-star reviews on Google and Tripadvisor, zero on Facebook. On that platform, for that engine, you’re new.
- Disconnected multiple accounts. The owner’s personal Facebook page + business page + a second “tour” account not linked. The brand signal fragments.
These are basic hygiene problems, not advanced strategy. And they get fixed in a week’s work.
What to do concretely this week
- Open your Instagram business account and check the category. If you sell archaeological tours in Pompeii, it must be “Tour Agency” or “Travel Company”, not “Personal Blog”
- Go to the linked Facebook business page: add hours, address, category, a description consistent with your sector’s queries
- Enable (or re-enable) public reviews on Facebook and ask recent customers to leave one with the name of the tour
- Verify with Google Business Profile that the same information (category, hours, NAP) is consistent — for entity consistency across platforms
- Publish at least 1 post a week on the Facebook business page with explicit sector keywords (Pompeii, Herculaneum, archaeological tour, licensed guide)
- Compare with the 3-5 competitors that Meta AI actually cites for your queries: what do they have that you don’t?
It doesn’t guarantee immediate appearance in the Meta AI answers. But it removes the basic blockers that today keep many otherwise valid tourism brands out.
Where the user actually looks for you
The most costly strategic mistake I see in tourism brands is optimizing only for Google and ChatGPT, ignoring that the traveler — especially the international one, especially on the move — opens Instagram to look for experiences. And inside Instagram today there’s Meta AI answering them, citing 3-4 names.
Being visible in AI answers doesn’t only mean appearing in ChatGPT or Perplexity. It means being there in every channel where an AI answers on the user’s behalf. WhatsApp, Instagram search and Facebook search are three new entry points you didn’t have a year ago.
In the next articles of this series I take you inside the other emerging channels: AI aggregators and meta-search, the vertical AIs by sector, and how to maintain cross-platform consistency when the AI engines are now 5 or 6 and no longer 2.
