We didn't read this on a blog: we measured it. 600 questions to ChatGPT, Gemini and Claude about Miami businesses, across five categories, five areas and two languages. The most useful finding for a bilingual city: asking in Spanish surfaced 166 distinct med spas, asking in English 106. Same city, two different maps.
Promeseo study, July 2026. Open method and downloadable data under CC BY 4.0.
How we work it
We measure before we say anything
Several questions, three engines, both languages, and the evidence of each answer. A single question proves nothing — models don't answer the same thing twice.
We work what depends on you
Your listing, your reviews, your own pages and the third-party sources that talk about you. That's what a model reads.
Sources beat adjectives
A business only its own Instagram talks about has one source. Models prefer the one with ten.
And we measure again
The point isn't the first report, it's the second: whether it moved, and in which engine.
Frequently asked
Can you guarantee ChatGPT will recommend me?
No, and that's the honest answer. Nobody controls what a model replies. We promise the work and the measurement, and we tell you the result even when nothing moved.
Why should I care if Google works fine for me?
Because they're two different things. In our study there are businesses ranking well on Google that appear in zero AI answers. Ranking in a search engine isn't the same as being named by a model.
How much does it cost?
The measurement is free right now: we hand it over as useful content while we finish testing the full service. No small print.
The other services
Shall we look at your business?
We review your Google listing at no cost and tell you what's wrong and in what order to fix it. No calls, no strings.
Get the free review