Getting recommended by ChatGPT isn't luck. It's the result of specific, measurable signals that AI models use when generating local recommendations. The good news: most of these signals are within your control, and most of the fixes are operational changes you can start this week.
Start with the audit
Before anything else, run an AI Visibility Audit to know your starting point. You need to know:
- Is AI mentioning you at all? (If yes — how often? At what position?)
- Which of your six score components is lowest?
- Who are your top competitors in AI responses, and what are they doing that you're not?
Without this baseline, you're guessing. With it, you have a ranked list of specific fixes with known impact.
Fix 1: Get your rating above 4.3★
This is the highest-leverage single action for most restaurants. AI models consistently recommend businesses above ~4.3★ and exclude those below it. Moving from 4.2 to 4.4 can shift you from invisible to regularly recommended.
How:
- Launch a review request campaign to recent customers who had a positive experience. SMS or WhatsApp outperform email for response rate.
- Make it easy: a single link directly to your Google review page. Every extra click loses 40% of people.
- Timing matters: request the review within 24–48 hours of the visit, while the experience is fresh.
- Train floor staff to mention reviews naturally: "If you enjoyed your meal, we'd really appreciate a Google review — it helps us a lot."
What to avoid: review gating (only asking happy customers to review, filtering out unhappy ones) is against Google's policies and can result in your listing being removed. Ask everyone; follow up on negative feedback directly.
Fix 2: Respond to every review — within 24 hours
AI models interpret review response rate as an authority signal. A business that responds to 90%+ of reviews is signalling active management and customer care. A business at 15% response rate is signalling neglect.
How:
- Set up a unified review inbox (tools like BrandCompanion pull Google and TripAdvisor into one place)
- Set a response SLA of 24 hours for all reviews, 4 hours for 1–2 star reviews
- Use templates for common patterns (thank you for 5-star, acknowledge-apologise-resolve for complaints) to make this fast without making it generic
- Never copy-paste the same reply to multiple reviews — AI can detect generic responses and this actively hurts your authority signal
Time investment: once set up, responding to reviews for a single restaurant takes 20–30 minutes per day. For multi-location operations, use a unified inbox and templates to scale this without proportionally more time.
Fix 3: Fix your listing inconsistencies today
Check these right now:
- Open Google Business Profile and TripAdvisor side by side — do your opening hours match?
- Check your website footer/contact page — does the phone number and address match Google exactly?
- Is your cuisine type / category tagged correctly on GBP? (Missing tags are a common issue for restaurants that changed format — e.g. added brunch, went alcohol-free)
- Are your photos recent and representative? Outdated or low-quality photos drag your listing authority.
These fixes are permanent once done. AI models will re-index your updated profiles within 1–2 weeks.
Fix 4: Address sentiment complaints before they compound
AI reads your reviews, not just your star average. Recurring complaints about specific issues create a negative sentiment fingerprint:
- "Service is always slow" appearing in 40% of reviews
- "Difficult to park" mentioned consistently
- "Noisy, hard to have a conversation" repeated
AI models pick up these patterns and apply them as qualifiers. A restaurant with a 4.4-star average but consistent complaints about wait times may still be recommended less than a 4.3-star restaurant with uniformly positive themes.
How to find your sentiment issues: look for patterns in your 3-star reviews — these are the most honest feedback, often containing both what's working and what isn't. Create an action item for each recurring complaint.
This is operational, not marketing. You can't review-respond your way out of a real service problem.
Fix 5: Get listed on the directories AI cites
Run a few AI queries about your category and city. Read the responses carefully: "according to [source]", "featured in [guide]", "listed on [platform]". These are the citation sources AI is drawing from.
For most restaurants in major cities, the key sources are:
- OpenTable, Resy, or your region's dominant reservation platform
- Time Out, Eater, or local food publication equivalents
- Google's "Best of" lists and featured snippets
- Category-specific directories (e.g. The Infatuation, Michelin Guide where applicable)
Getting listed on platforms you're missing is a one-time task with permanent benefit.
Fix 6: Audit your competitors' scores
Your score doesn't exist in isolation. If your top competitor is at 78/100 and you're at 54/100, AI will recommend them in nearly every head-to-head comparison. You need to know specifically where they're stronger.
Look at:
- Their Google rating vs yours
- Their review response rate (you can estimate this by looking at their recent reviews)
- Their listing completeness vs yours
- Which directories and publications feature them that don't feature you
This tells you the specific gap to close, not just the general goal of "improve."
What to prioritise
If you've run an AI audit and have your component scores, prioritise in this order:
- Review Signal (if below 60) — it has the highest weight (25%) and the fixes are clear
- Listing Consistency (if below 70) — easy wins that are permanent once done
- Sentiment — requires operational change, but compounding impact over time
- Citations — one-time work, permanent benefit
- Competitor Gap — informs how aggressive to be on all the above
AI Recommendation score will follow naturally as you improve the underlying signals. You can't directly "optimise" for AI mentions — but you can build the conditions that cause AI to recommend you.
Timeline and expectations
- Week 1–2: Listing consistency fixes deployed; review request campaign launched; review response rate improving
- Week 3–4: Google rating starting to move; sentiment complaint themes being addressed operationally
- Week 6–8: AI Recommendation component beginning to move as underlying signals improve
- Week 10–12: Full score impact visible; repeat audit to confirm progress
For most restaurants in the Invisible or At Risk band, moving to Visible within 10–12 weeks is realistic if fixes are executed consistently.
Run a free AI Visibility Audit for your restaurant at brandcompanion.ai/ai-visibility — results in 1 minute, no login required.