Every source, checked every hour
Collectors for portals and dealer sites in ten countries. Hourly reports compare each source with its usual volume, so a source that quietly returns less gets flagged before the numbers drift.
Our own product · data
Three million cars, one record each, priced against the market every day
Mycarro reads about six million public car adverts from portals and dealer sites in ten European countries, works out which adverts are the same car and gives every car a market price. We built the whole pipeline and run it, from collection to the numbers that buyers and dealers see.
ClientMycarro (mycarro.app), our own product
ScopeCollection, matching, valuation, search, dealer software
StatusIn production in 10 countries
The problem
The same car on three sites, at three prices
A used car is often listed on several portals and on the dealer's own site at the same time. Each advert has its own photos, its own spelling of the model and sometimes its own price. Adverts appear, change price and disappear every hour.
Before you can say anything true about the market (what a car is worth, how long cars take to sell, whether a price is fair), you need one clean record per real car. That record has to stay current every day, across whole countries.
Ready-made feeds cover one portal or one country, and they sell adverts, not answers. So we built the pipeline ourselves.
What we built
From raw adverts to answers
One record per real car
Adverts for the same car are folded into one listing across sources. About 5.9 million adverts become about 3.1 million cars.
A market price for every car
Each car is priced against comparable cars, so the site can show how far an asking price is above or below the market.
What changed, and when
Price changes and the day a car leaves the market are kept for every car. Days-to-sell and market indices are built from this history.
Results
What it does today
Live adverts read
From portals and dealer sites, refreshed continuously.
Cars after matching
One record each, with a market price.
New adverts a day
Collected, matched and priced on the day they appear.
One data model
UK, Ireland, France, Spain, Portugal, Italy, the Netherlands, Belgium, Luxembourg and Switzerland.
Figures from Mycarro's own hourly reports, 7 November 2026.
How it holds up
The hard parts of data at this scale
Sources fail quietly
A site changes its layout or blocks a collector, and nothing errors: it just returns fewer cars. Volume checks per source catch that within the hour.
One car, one record
Matching that splits one car into two inflates the market. Matching that merges two cars hides one. Both cases are measured, not guessed.
Deletes are checked
A short page from a source must not wipe its stock. Removals are compared against what the source normally holds before cars are taken off.
Too good to be true
Prices far below the market are treated as suspect, not as deals, because that is what scam adverts look like.
Built on
- Web data collection
- Record matching
- Valuation models
- Search
- Hourly health reports
We build the same kind of pipeline for clients: data engineering and web data at scale.
Free 30-minute call
Tell us what has to work, and by when
You talk to an engineer who would work on it, not a salesperson. You leave the call knowing what it would take and what the first step costs, or we tell you that you don't need us.