Method
No gut-feeling scores.
Every number on OpenCarlytics can be rebuilt. Here are the weights, formulas and limits: if you think something is wrong, owner votes exist precisely to correct us.
1. The overall score
It's the weighted mean of the ten metrics. Balanced-profile weights reflect what matters most to a used-car buyer: first of all, not breaking down and not overpaying.
OC = Σ (score × weight) / Σ weight| Metric | Weight | Basis |
|---|---|---|
Reliability ADAC and TÜV breakdown stats, owner surveys, known engine faults. | 18% | Aggregated sources |
Safety Euro NCAP stars, with a penalty for tests before the 2018 protocol. stars → {0: 1.5, 1: 3.0, 2: 4.4, 3: 6.0, 4: 7.6, 5: 9.2} − 0.08 × years before 2018 (max −1) | 12% | Hard data |
Value for money Overall quality (reliability, safety, economy, comfort, space) against the average used price. 5.6 + (quality − 6.5) × 1.5 − log₂(avg used price / 15,000) × 1.5 | 12% | Calculated |
Fuel economy WLTP combined consumption (l/100 km), or kWh/100 km for EVs. 10 − (l/100km − 3.8) × 1.15 | EV: 10 − (kWh/100km − 13) × 0.45 | 10% | Hard data |
Comfort Ride, refinement and seats from road tests. | 10% | Aggregated sources |
Practicality Access, flexibility, storage and everyday usability. | 9% | Aggregated sources |
Space Boot litres, overall length and seat count. 1 + (length mm − 3500)/1500 × 5 + litres/700 × 4 (+0.7 for 7 seats, −1 for 2 seats) | 8% | Hard data |
Resale value Depreciation observed on the European used market. | 8% | Aggregated sources |
Performance 0–100 km/h time of the mainstream version. 10 − (0-100 s − 3.5) × 0.7 | 7% | Hard data |
Driving fun Steering, chassis and engagement from road tests. | 6% | Aggregated sources |
2. Owners correct the sources
Each metric starts from its base score (data + sources) and moves towards the owners' average with a weight that grows with the number of votes. At 8 votes owners weigh as much as all other sources combined; at 40 votes they weigh 83%.
score = (base × 8 + media_proprietari × n) / (8 + n)3. Driver profiles
In the catalog you can switch profile: only the weights change, never the scores. The same car can rank first for a commuter and tenth for an enthusiast.
| Profile | Reliab. | Safety | Value | Econ. | Comf. | Pract. | Space | Resale | Perf. | Fun |
|---|---|---|---|---|---|---|---|---|---|---|
| Balanced | 18 | 12 | 12 | 10 | 10 | 9 | 8 | 8 | 7 | 6 |
| Family | 18 | 20 | 10 | 8 | 10 | 14 | 14 | 4 | 1 | 1 |
| Commuter | 22 | 10 | 14 | 22 | 12 | 6 | 3 | 7 | 2 | 2 |
| Tight budget | 24 | 8 | 28 | 16 | 4 | 6 | 4 | 8 | 1 | 1 |
| Enthusiast | 10 | 8 | 6 | 4 | 6 | 3 | 3 | 10 | 22 | 28 |
4. Analysing a listing
- Yearly mileage — below 8,000 km/year we urge caution (possible clocking), 8–20,000 is normal, above 28,000 is high.
- Typical value — we interpolate the generation's used price band by year and adjust it for mileage. It's indicative: trim and condition matter.
- Engine faults — if the ad mentions a component with a known issue (PureTech, DSG, EcoBoost…), we flag it.
- Analysis text — written by an AI model, but the numbers are ours: for catalog models the AI cannot change the scores.
Limits, plainly stated
- Technical data refers to the most common version of each generation: a 2.0 TDI and a 1.0 TSI of the same car can differ a lot.
- Editorial scores are a starting point, not the truth: they become more reliable as owners vote.
- Price bands are indicative of the Italian/European market and are updated periodically.
- No rating replaces a pre-purchase inspection by a trusted mechanic.