In every audit we run, we capture two parallel data streams: what paid-tier AI users see, and what free-tier AI users see. The difference between the two is consistent enough that we treat it as a structural feature of the AI hospitality recommendation market — not noise. On average, across 200+ audits, the gap is 22 points. For some businesses, it's 40 points or more. This is the proprietary insight Score My Business was built around. No other audit measures it.

What the tiers actually are

ChatGPT splits its user base across two model tiers. Free users get gpt-4.1-mini, a smaller, faster, less capable model that handles the daily query volume of roughly 800 million users. Paid users (ChatGPT Plus or Pro) get gpt-4o, a flagship model with stronger reasoning and tighter recommendation discrimination. Gemini's tier split mirrors this: Gemini AI Pro subscribers get Gemini 2.5 Pro; free users get Gemini 2.5 Flash.

The two models are trained on similar data but fine-tuned differently and have different inference-time compute budgets. The result is two slightly different "selves" living inside the same product brand — and they describe your restaurant differently.

What the gap looks like

Concretely: a typical mid-tier London restaurant in our audit base shows up in 77% of paid-tier ChatGPT responses to "best restaurants in London" and 55% of free-tier responses. The gap is real, repeatable across runs, and consistent across question phrasings.

The gap widens for businesses with subtle positioning. A chef-led modern Anatolian restaurant with no Michelin star but strong English-language press: 80% paid, 32% free. Free-tier models lean on training-data popularity and TripAdvisor sentiment density; paid-tier models do more reasoning about specific positioning ("rediscovering forgotten Anatolian recipes") and surface those businesses confidently.

Who's on each side of the gap?

ChatGPT Plus and Pro subscribers are roughly 12% of ChatGPT's total user base globally — but they're a disproportionately valuable customer segment. Internal hospitality data we've reviewed (anonymized) shows paid-tier AI users are 3.6× more likely to book a $200+ tasting menu and 4.2× more likely to spend $400+ per night on hotels than free-tier users. They also book further in advance and have higher show-up rates.

Free-tier users are ~88% of the AI user base and overrepresented among casual diners, family travelers, and budget-conscious customers. They convert at lower per-customer spend but higher absolute volume.

Both segments matter. The question isn't which one to optimize for — it's which one you're currently invisible to.

The asymmetric implications for your strategy

If your paid-tier AI score is high (say, 80+) but your free-tier score is low (say, 30): you're winning with high-value customers and missing the volume segment. This often happens with chef-led places with strong press depth — paid-tier models reward your specific positioning, but free-tier models default to higher-aggregator-density places.

If your free-tier score is high but paid-tier is mediocre: usually means strong TripAdvisor and Yelp presence but thin press depth. You're getting volume but missing the higher-spend audience.

If both scores are low: you have a general visibility problem, not a tier-specific one. Address fundamental signal hygiene first (Google profile, NAP, structured data) before worrying about tier-specific gaps.

Closing the gap

Free-tier ranking responds most to: review aggregator presence (TripAdvisor, Yelp), volume of recent positive reviews, and Google Business Profile completeness. Paid-tier ranking responds most to: high-trust English press mentions, specific cuisine positioning, Wikipedia presence, and structured data on your own site.

Most owners can move both scores together with one set of actions, but understanding which side is your gap focuses the effort.

This split is captured on every Detailed and Full Audit report — page 3 of your audit specifically — with both scores side by side and intent-by-intent breakdowns.