AutoScout24 shows you what sellers are asking. It doesn't tell you if that price is fair. This is a product concept that fixes that.
I spent a year at Bähr & Fess Analytics doing residual value research — calculating what cars would be worth in 3 and 5 years for clients like Škoda, Range Rover, Volvo, and Kia. Part of the job was understanding the German, French, Polish, and UK used car markets in detail. I know what makes a used car price go up or down.
Then I tried to buy a used car on AutoScout24. And I realised that all that knowledge I had professionally — none of it is available to a normal buyer.
I searched for a 2015 Škoda Octavia. There were hundreds of listings. Same model, same year, prices ranging from €8,000 to €16,000. No explanation for the difference. I had to open listing after listing, manually compare mileage, check fuel type, guess at condition. It was exhausting — and I knew more about car valuation than most people ever will.
For someone who doesn't have that background, it's even worse. You just pick a number and hope you're not overpaying.
Used car pricing is not random. The value of a specific car depends on a handful of factors that are knowable: mileage, fuel type, generation, trim level, colour, number of previous owners, and market (prices differ by country). The data exists. Dealers use it every day.
Buyers don't have access to any of it in a usable form.
AutoScout24 gives you a "fair price" tag on some listings — but it's vague, inconsistently shown, and doesn't explain anything. You can't ask "what should THIS car cost?" You just scroll and guess.
| What buyers deal with today | What Wert would change |
|---|---|
| See a price, no idea if it's fair | Get a fair price range for any specific car in seconds |
| Compare 20 listings manually in separate tabs | Paste a listing URL, get an instant overpriced / fair / good deal verdict |
| No idea how mileage or fuel type affects value | See exactly which factors are raising or lowering the price |
| Don't know if now is a good time to buy | See how prices for this model have moved over the last 12 months |
The product is simple: you tell Wert what car you're looking at, and it tells you what that car is actually worth — and whether any listing you're considering is priced fairly.
The first-time used car buyer. Usually 25–35, buying their first car independently. Has no frame of reference for prices. Spends hours on AutoScout24, paralysed by choice and unsure what to trust. They're the most vulnerable to overpaying and the most likely to pay for confidence.
The upgrader. Already owns a car, looking to switch models. Knows more than a first-timer but still can't tell if a listing is overpriced by €1,500 or actually reasonable. Wants to negotiate but doesn't know what number to negotiate toward.
I'd build for the first group first. Their pain is stronger and their decision is more emotional — which means the "aha moment" (seeing a fair price range for the first time) hits harder and creates more loyalty.
A used car in good condition is worth more than the same car in average condition. Condition varies. So I'd show a range — €9,500 to €11,200 — not a single number that would feel authoritative but be wrong half the time. Ranges are honest. Single numbers are false precision.
Telling someone "€10,500 is fair" isn't enough. I'd show the factors: diesel adds €800, 120k km reduces by €1,200, this generation holds value well. People trust a number more when they understand how it was built. And they learn something about used car buying in the process.
Most users arrive with a specific listing already open. They don't want to fill out a form — they want to paste a URL and get an answer. That's the fastest path to value. The manual search is for when they're still exploring.
Residual values differ by country. A Škoda Octavia in Germany holds value differently than in Poland. Building accurate models for multiple markets at once is a data problem that would kill an MVP. Germany first — it's the biggest used car market in Europe and the market I know best from my Škoda work.
Four screens — search, result, market trends, and a listing checker.
First use is curiosity. Return use means the product is part of the buying process. That's the behaviour I want to create.
From opening the app to seeing a price range. If it takes longer than 30 seconds the product is too complex. Used car buying is already slow — Wert has to be fast.
Unlike Claro, NPS works here. Buying a car is social — people talk about it. High NPS means organic word-of-mouth from people who just got a good deal because of Wert.
Enter make, model, year, mileage, fuel type. Get a price range. No listing scan yet. Validate that the price data is trusted and useful before building anything else.
Paste an AutoScout24 URL and get overpriced / fair / good deal. This is the high-value use case — but it only works if the price model is already validated.
Show WHY the price is what it is. Mileage, fuel type, generation, condition. This is the feature that builds trust and teaches users how to evaluate cars themselves.
Track how prices for a model move over time. Set an alert when a car you're watching hits your target price. This is what turns a one-time tool into a product people keep coming back to.
Most product thinkers would approach this from the user research side. I can do that — but I also come at it from the data side. I spent a year building the exact models that power something like this. I know which variables matter, how they interact, and how to get the data.
At Bähr & Fess I built 3-year and 5-year depreciation forecasts for Škoda models across four countries. That work was bought by Škoda's Global RV leadership team. The methodology behind Wert's price ranges is essentially a consumer version of that work.
That's not something most PMs can say about their product concepts.
Case study by Yuvaraj Devadoss
← Back to portfolio