THREE DATASETS WE BUILT
AND KEEP UPDATING.
Property records, neighborhood perception by zone, and observed demand signals. All three are Quintana Roo, because that is where we have been longest, and we would rather say so than round it up. The measuring is what travels. This August we read 63 agencies by hand across five states, 100 paid ads with the country set to Mexico, and 705 more in English to test the result. When a question lands where we hold no dataset, it gets answered from the public record, the registry and field work, and the answer says which of those produced it.
Claude® is a product of Anthropic, PBC. FEUDO® Group is an independent API user, not affiliated with or endorsed by Anthropic.
The deep layer · Quintana Roo
Three datasets.
One state, read all the way down.
We built three datasets before we built any tool, and all three cover Quintana Roo. That is the honest edge of what we own. What follows is each one, and what the tools can answer because it exists.
Quintana Roo property database by zone, typology, delivery stage, and price band. Every archetype the Reality Filter runs against comes from this dataset.
What buyers say about each Quintana Roo zone when nobody is selling to them. Reddit threads, YouTube comments, forums. Scraped, structured, and mapped by neighborhood.
Real inbound demand in the Riviera Maya: zone, sub-zone, budget range, typology, intent. 15 per day on average. The gap between what's offered and what's actually wanted, made visible.
Every recommendation comes with the filter logic and the discard reasons attached, so the score can be read and checked line by line.
Where FEUDO Group holds a dataset, the tools read it. Where FEUDO Group holds none, the same questions get answered from the public record and field verification, and the output names which layer answered.
Decision tools that appear at the point in the process where the analysis changes the outcome.
“We built it so the answer comes with the data behind it.”
The wide layer · Mexico
What we measured
outside Quintana Roo.
Our datasets stop at the state line. The measuring keeps going. Everything below was read by hand, on criteria written down before the first screen was opened, and it is published with the date, the sample and the holes we know about.
Sixty three real estate agencies read by hand across five states: Quintana Roo, Jalisco, Nuevo Leon, Yucatan and Baja California. Six of them publish a method. Four publish how they charge. One publishes what it turns down. Fifty three offer to buy and to sell at the same time.
Read 17 Aug 2026. The scoring criteria were written down three minutes before the first screen was captured.
One hundred active ads from 83 advertisers in the public Meta ad library, country set to Mexico. None of the hundred say they work for the person buying. None say how they charge. The Spanish word for buyer, comprador, does not appear once in the hundred. Twenty three of them are offers to buy a house from someone who can no longer keep up the payments.
Read 17 Aug 2026. One advertiser contributes at most three ads, so 100 ads come from 83 accounts.
The zero above was put under pressure on purpose, with five more searches: realtor, realty, buyers agent, agente del comprador. Seven hundred and five ads from 478 advertisers. Ten of them do declare they work for the buyer, and not one of the ten is advertising to the Mexican market. The ones named in the file sit in Australia and the United States, where buyer representation is a settled advertising category, and it is not advertised in Spanish here.
The buyer this line was built for has usually seen the category already, in the country where it is advertised.
Eight Mexican listing portals swept by hand against one real search in Merida: land, northwest quadrant, under 825,000 pesos all in. The same twelve month project came back on four different portals. A portal is good for learning that a project exists. Location, price and payment plan get verified somewhere else before any of it reaches a client.
Read 9 Aug 2026. Facebook Marketplace was left out on purpose, because location cannot be verified there.
Eighteen developments reviewed in Merida on one search, against paperwork from the developer rather than the listing. Two survived. On the one opened furthest, the inventory map inside the developer CRM showed 477 lots with 193 still available, at a level of detail no portal carried.
Read 1 Aug 2026. The surviving two are held with a reservation: one still has no price confirmed in writing.
What this does not say. Both samples were read in the order the sources returned them, so neither one is random. The agency reading runs 9 to 20 per state, which describes the market and does not rank one state against another. The ad reading covers the text of the ad, not the video and not the image. And the three datasets above are still Quintana Roo only, which is the point of separating them.
Coverage
Declared for Mexico.
Graded by evidence.
A firm that claims the whole country and shows one state has told you something without meaning to. Here is the grade, in three levels, so a reader in Guadalajara, Merida or Monterrey can see exactly what would be behind an answer given to them.
OWNED DATA · QUINTANA ROO
The three datasets above. Recorded and updated by FEUDO Group, zone by zone and month by month, for Tulum, Playa del Carmen and Cancun. This is the deepest we go anywhere, and it is one state.
MEASURED · FIVE STATES
Quintana Roo, Jalisco, Nuevo Leon, Yucatan and Baja California, on the readings above. Merida also carries case work verified line by line against developer files.
DECLARED · MEXICO
The rest of the country is inside what FEUDO Group takes on, and it arrives without a dataset of ours behind it. When that happens the answer says so, and it shows the public record, the registry entry or the site visit it came from.
The order matters more than the map. A market moves up a level only when there is a file behind it.
Across sectors
The same three datasets,
pointed at another sector.
The three-dataset framework applies to any sector where information asymmetry is the buyer's main risk, which is most of Mexico. Quintana Roo residential is where it was built first.
Residential · Riviera Maya
Zone perception by neighborhood, pre-sale developer risk, beach condition data, demand signals by typology and budget. The founding application of the methodology.
Industrial · Site Selection
Territorial analysis for corporate and manufacturing relocation. Zone perception maps to logistics and workforce data. Demand signals map to nearshoring pipeline requirements. Same framework, different inputs.
The method travels between sectors. What has to be rebuilt each time is the data underneath it, and rebuilding it takes the same months it took the first time.
Where it started
Four tools.
Real work done.
These were structured tools built on written criteria. They ran on ChatGPT. Now they are being rebuilt on the datasets above.
Stress-tested appreciation and equity growth for Riviera Maya properties. Modeled scenarios with every assumption written down.
Listings ranked by the FEUDO Group score, with the criteria that produced each rank. It exposed the gap between how a project presents and what its fundamentals hold.
Flagged risky clauses in Mexican presale contracts before signing. Built to surface what standard legal review either misses or does not flag loudly enough.
Replaced the projected occupancy in the sales material with occupancy modeled from local supply and demand.
These tools did real work. They are not gone. They are being rebuilt on the three datasets, which did not exist when the tools were first written.
Early access
Be first to test
FEUDO® Intelligence.
If you want to be among the first to test FEUDO® Intelligence, reach out and tell us what decision you are trying to make, and where.
Client intelligence portal
Clients who engage with FEUDO® Group receive a private portal built for their specific case. It holds the title flags, the comparables, the demand signals for the target zone, FEUDO Group's written strategy, and the next step, all for one case, in one place.
In the meantime
The methodology is already live.
The intelligence tools are in development. The discipline behind them started in Quintana Roo, the hardest market we could pick, and that is still where the Reality Filter runs today.