Our own studies · Original study · Bogotá, August 2026
Category by category, we compared who really leads according to Google with who an AI assistant actually names. They are almost never the same — and this time we measured Google's ranking properly.
Four categories (bakeries, coffee shops, barbershops, veterinary clinics). The real ranking comes from the official Google Places API: 1,147 verified businesses in Bogotá, not a sample of ten. The AI side comes from the 26 sources an assistant with web search retrieves and cites for the same question.
Our earlier study, in Medellín, built each "Google top 10" from a single query returning ten results. That is not a city's top 10: it is the ten most relevant results for one phrasing, re-sorted afterwards by review count. Obvious leaders were missing.
Three distinct failures were stacked in that one call, and all three are fixed here:
The result: 1,147 verified businesses across four categories, against 40 in the Medellín study. Everything below rests on that base.
On the left of each block, Google's top 10 by review volume — the best public proxy for how much local demand each business captures. On the right, the businesses the AI named when answering "best [category] in Bogotá", with the position they actually hold.
This is the part that explains everything above. We classified the 26 sources the assistant retrieved and cited across the four questions.
Not one of the 26 sources was Google Maps. The place where a business's entire competitive advantage lives — its reviews, its rating, its history — is exactly the one the AI does not read when building a recommendation.
Having the best reviews in the city does not put you in the AI's answer. Having a page the AI can read and quote does.
The clearest case in the study is a bakery: Ene Pan Artesanal, 12 reviews, position #315 of 329. Practically last in its category in Bogotá by volume — and the AI recommends it, because pages exist that talk about it.
In the other direction, the #4 barbershop is literally called "Gentleman's Barbería Tradicional | Cedritos La mejor barbería" — "the best barbershop" — has 1,487 reviews and 4.9 stars, and appears in no answer at all. Saying it in your listing's name does nothing; a page has to say it.
Five of the 26 sources — 19% — were the business's own website. In barbershops, three of seven. They didn't buy ads or accumulate thousands of reviews: they published a page that answered the question, and the AI quoted it.
This is a figure almost nobody publishes, because it requires sweeping the entire category. With 1,147 verified businesses we can say, per category, how many reviews it takes to reach each tier in Bogotá.
| Category | Businesses | Top 10 | Top 25 | Top 50 | Median |
|---|---|---|---|---|---|
| Panaderías | 329 | 3,313 | 1,694 | 964 | 236 |
| Cafeterías | 298 | 2,096 | 928 | 368 | 28 |
| Barberías | 270 | 880 | 521 | 263 | 52 |
| Veterinarias | 250 | 1,332 | 923 | 521 | 181 |
The practical reading: the distance between the median and the top 10 is brutal — a median Bogotá bakery has 236 reviews and the tenth has 3,313 — but the distance between the median and the top 50 is entirely reachable. Chasing the leader is a bad goal; reaching your category's median, then the top 50, is a real one.
Since the API returns the website declared on each listing, we measured something usually asserted without data: what share of businesses have a website, by position.
| Category | Top 50 | From 51 down | Whole category |
|---|---|---|---|
| Panaderías | 62% | 24% | 29% |
| Cafeterías | 68% | 19% | 27% |
| Barberías | 80% | 42% | 49% |
| Veterinarias | 80% | 44% | 51% |
The gap is consistent across all four categories: among the first fifty, a website is the norm; below that, the exception. Among the businesses the AI named, 86% have their own site; among Google's ten leaders, 93%.
And here it's worth being honest about which way the arrow points. This does not prove that adding a website raises your position. It's equally plausible that the business already earning more can afford a website. What the data does support, without argument, is the other claim: a website is a requirement for being cited by an AI, because an AI cannot cite what it cannot read.
A measurement caveat: "no website" here means no website declared on the Google listing. Mustaches Barbería shows no site on its listing, yet its own page was one of the sources the AI cited. An empty field doesn't prove the business has no website — it proves the listing doesn't link it, which is a different error and also a costly one.
In Medellín the result was 2 matches out of 20, with two categories at zero. In Bogotá it is 5 of 21. The underlying conclusion doesn't change — the two rankings differ and Google Maps does not feed the AI — but the gap is less extreme than the earlier study suggested.
It's worth saying why, because part of the explanation is ours: Medellín's "top 10" came from a defective method, so some of those "non-matches" were probably businesses that did lead and that our query never saw. The right number for Medellín is likely somewhere between the two.
What holds identically in both cities, and is all the argument needs: zero Google Maps sources, most review leaders without a single mention, and a fifth of the citations coming from the business's own site.
What about your business? Visquore runs this same comparison for one specific business: your real position on Google, your verified competitors, and whether AI assistants name you.
Run my audit →searchText. Per category: 3-4 phrasings citywide, paginated to 3 pages, plus 6 Bogotá zones searched separately to 2 pages. 111 requests in total. Verified data, not estimates.An assistant with web search answers from what it finds written on indexable pages. Google Maps is not one of them in practice: across our four Bogotá queries, none of the 26 cited sources was Google Maps. Reviews decide your position on the map; the pages that talk about you decide whether an AI names you.
It depends heavily on the category. Per our August 2026 measurement: 3,313 reviews for bakeries, 2,096 for coffee shops, 1,332 for veterinary clinics and 880 for barbershops. Each category's median is far lower — 236, 28, 181 and 52 respectively — and is a much more realistic target.
It's a requirement, not a guarantee. An AI cannot cite what it cannot read, so without a page of your own you depend on someone else writing about you. In our data 86% of the businesses the AI named have their own site. But be careful with causality: among the top 50 in each category a website is the norm (62-80%) and below that the exception (19-44%), which is also consistent with better-earning businesses affording one.
Yes, and it's the most striking finding in the study. The AI recommended a bakery with 12 reviews sitting at position #315 of 329 in its category in Bogotá. You don't need to lead the market: you need a page to exist that answers the customer's question.
Our first measurement, August 2026: which businesses an AI assistant names when recommending in Medellín, and where those answers come from. None of the 22 sources it cited was Google Maps.