When you ask ChatGPT "best Italian restaurant in [city]", it doesn't return a ranked list of every Italian restaurant. It recommends 3–5 specific businesses. The rest are invisible.
What determines who makes the cut?
After running hundreds of AI Visibility Audits across different cities and categories, a pattern emerges consistently: businesses with Google ratings below approximately 4.3 stars are systematically excluded from AI recommendations, regardless of how many reviews they have, how established they are, or how well they rank on Google Maps.
This is the 4.3★ rule. It's not an official feature or documented behaviour — it's an emergent property of how large language models have learned to be helpful.
Why does this threshold exist?
AI models like ChatGPT are trained to maximise user helpfulness. When a user asks "where should I eat?", the model is optimising for a good outcome for that user. Over billions of training examples, it has learned that recommending a 4.2-star restaurant risks a mediocre experience — and a mediocre experience is a failure.
The exact threshold isn't fixed. It's not a hard filter where 4.29★ is excluded and 4.30★ is included. It's more like a probability gradient: as you move above 4.3★, your probability of appearing in AI recommendations increases significantly. Below 4.3★, it drops sharply.
In practice, the pattern looks like this:
| Rating | Typical AI mention rate |
|---|---|
| Below 4.0★ | Near zero |
| 4.0–4.2★ | Rare (1–3 of 18 runs) |
| 4.2–4.3★ | Occasional (3–6 of 18 runs) |
| 4.3–4.5★ | Regular (8–13 of 18 runs) |
| Above 4.5★ | Consistent (12–16 of 18 runs) |
(These are indicative patterns from audit data, not published benchmarks.)
What "4.2★ but 400 reviews" looks like in practice
A common situation: a restaurant with 400 Google reviews and a 4.2-star average. By traditional metrics, this is a solid business with strong review volume. On Google Maps, they rank well. In TripAdvisor's filter, they're in the "Good" category.
But when we run an AI audit:
- 18 ChatGPT runs on relevant "best [category] in [area]" queries
- Mentioned in 2 of 18 runs (11%)
- When mentioned, ranked #4 or #5 — an afterthought, not a recommendation
- Competitors with 4.4★ and 150 fewer reviews are mentioned in 13 of 18 runs
The rating is the filter. 400 reviews at 4.2 is worse for AI visibility than 100 reviews at 4.4.
Why this matters now specifically
Six months ago, this was a curiosity. Today, it's a revenue leak.
AI-powered local discovery has grown from 6% to 45% of all local search queries between 2025 and 2026. The trajectory suggests AI search will be the dominant channel for local discovery within 2 years.
Businesses at 4.2★ are being systematically excluded from a channel that's growing faster than any other. Every week at 4.2 is a week of compounding lost visibility.
The good news: 4.3★ is achievable
Moving from 4.2 to 4.4 is not a heroic effort. It requires:
A review request campaign — asking recent customers who had good experiences to leave a review. Most satisfied customers don't review because no one asked. A simple SMS or WhatsApp link to your Google review page, sent within 24–48 hours of their visit, converts at 15–25%.
Consistent review responses — responding to every review, especially negative ones. Businesses with high response rates tend to attract more balanced reviewing: customers who feel heard are more likely to update or add a positive review.
Operational fixes for recurring complaints — if your 3-star reviews consistently mention a specific issue, fix the issue. Reviews that say "service has improved since my last visit" are gold for AI sentiment signals.
The math: for a restaurant with 400 reviews at 4.2★, receiving 20 new reviews per month: if 15 of those 20 are 5-star (plausible with a good review request campaign), the average moves to ~4.25 in month 1, ~4.3 in month 2, ~4.35 in month 3.
Not overnight. But measurable and achievable.
What happens when you cross the threshold
Businesses that move from 4.2→4.4 through our Fix & Track program consistently show the same pattern in their weekly re-scores:
- Weeks 1–4: Review Signal component score improving; overall AI score climbing 5–10 points
- Weeks 5–8: AI Recommendation component beginning to move as AI models re-index updated ratings
- Weeks 9–12: Mention rate in audits noticeably higher; competitors who used to dominate appearing less consistently
The transition isn't instant — AI models don't update in real time. But it's reliably directional once the underlying signal improves.
Beyond the rating: what else AI filters on
The 4.3★ rule is the most impactful single threshold, but AI recommendation is multi-signal. Even businesses above 4.3★ can be filtered out by:
Low review response rate: AI interprets a 10% response rate as low engagement. Even at 4.5★, a business that ignores most reviews signals lower authority than a 4.3★ business with 95% response rate.
Listing inconsistencies: Hours that differ between Google and your website tell AI "this listing may be inaccurate." Inaccurate listings = reduced confidence = fewer recommendations.
Citation gaps: Not listed on directories AI regularly cites (Time Out, OpenTable, local food guides) means less corroborating evidence. AI recommendations that cite multiple sources are more confident than those based on a single source.
The business case for taking this seriously
For a restaurant doing AED 150K/month in revenue, with 30% of new customers currently coming from digital discovery:
- If 45% of digital discovery is now AI search, that's ~AED 20K/month potentially influenced by AI recommendations
- A restaurant currently at 4.2★ (Invisible in AI) vs 4.4★ (Visible in AI) could be leaving AED 10–15K/month on the table
- The fix — a review request campaign and consistent review responses — costs a fraction of that
The 4.3★ rule is one of the highest-leverage, lowest-cost improvements a local business can make right now. And it compounds: a higher rating improves both your Google Maps ranking and your AI visibility simultaneously.
Find out if you're above or below the AI threshold. Run a free AI Visibility Audit at brandcompanion.ai/ai-visibility.