How the Shortlist score works

We don't test anything ourselves. We collect results from independent testers, put them on one scale and combine them. Every product page lists each result with its weight, so you can follow the arithmetic.

1. Collecting results

For each source we record the result exactly as published: the grade, percentage, stars or label, with the test year and a link. Several testers are paywalled (Stiftung Warentest, Which?, Consumer Reports). For those we only use what they publish openly, such as test winners, Best Buys and grades quoted in their press releases. Results seen only in a press summary are marked as such and count 85% as much.

2. One scale from 0 to 100

• German school grades (ADAC, Stiftung Warentest, ÖAMTC, TCS; 0.5 is best, 5.5 worst) are mapped linearly: 1.0 → 90, 1.5 → 80, 2.0 → 70, 2.5 → 60, 3.5 → 40, 4.6 → 18. The band edges line up with the testers' own verbal grades: "sehr gut" lands in Excellent, "gut" in Good.
• Percentages (dTest, OCU, TCS, Konsument, BabyGearLab) are used as they are.
• Stars are a share of the maximum: 4 of 5 stars → 80.
• Verbal ratings get fixed values: sehr gut 90, gut 70, befriedigend 50, ausreichend 30, mangelhaft 12; IIHS Best Bet 90, Good Bet 70, Check Fit 45, Not Recommended 10.
• Folksam grades relative to the other products in its test, so its grades get fixed values: 4 → 80, 3 → 65, 2 → 45, 1 → 20.
• Editorial picks and pass/fail tests (Wirecutter picks, Swedish Plus Test) carry no score. We show them as badges but leave them out of the number.

3. Counting each test once

Many European organisations run the same lab test together through ICRT and each publish it. ADAC, Stiftung Warentest, ÖAMTC, TCS, Which?, dTest and others do this for car seats, twice a year. Counting every copy would let one crash test look like five opinions. We keep one copy of a joint test, from ADAC where available, and list the other publishers next to it. For car seats, results from ÖAMTC, TCS Touring Club Schweiz, Stiftung Warentest, Which?, dTest, OCU (Organización de Consumidores y Usuarios), Consumentenbond, Test-Achats / Test Aankoop, Konsument (VKI) are treated as copies of ADAC's test whenever ADAC has published that seat.

When a source has tested a product more than once, its newest tests count and older ones are shown but superseded. If one test covered several configurations, such as a car seat on two different bases, we use the best one, since you can choose it. Some editorial sites score products relative to each comparison article, so the same stroller can get 84 among joggers and 55 among full-size strollers. When a source has several recent tests of one product, we average them and the source still counts once.

4. Weighting

• Lab and consumer-organisation tests count fully (1). They measure things: crash sleds, chemical analysis, durability rigs.
• Editorial reviews count 0.6. They use products hands-on, and some run their own measurements.
• NHTSA (National Highway Traffic Safety Administration) Car Seat Ease of Use Ratings counts 0.35: its stars rate ease of use only, not crash protection.
• Folksam counts 0.6: it assesses seat design without crash tests.
• Tests up to 2 years old count fully. After that a result loses 10% of its weight per year, down to 40%.

5. The Shortlist score

First we take the weighted average of the counted results. A product with a single glowing review shouldn't outrank one that several testers rated highly, so we then add a small amount of doubt: the score is computed as if every product also had 0.5 of a test scoring 60. This only ever lowers a score, never raises it. With one lab test of 80 the Shortlist score is 73; with three lab tests averaging 80 it is 77.

Safety overrides. If a source's latest test of a product reports a safety failure or a Don't Buy, the score is capped at 35, whatever other tests said. Harmful-substance findings are shown as warnings. Testers like ADAC and ÖKO-TEST already downgrade the overall grade for them.

Older products. Products whose newest test is more than 7 years old are hidden from the rankings by default, because many are no longer sold. You can still include them, and their pages stay up.

Bands

• 80–100
  Excellent
• 60–79
  Good
• 40–59
  Fair
• 20–39
  Weak
• 0–19
  Poor

Limits worth knowing

• Testers weigh things differently. ADAC's car seat grade is mostly crash safety and handling, while a BabyGearLab score also reflects comfort and convenience. A shared scale makes them comparable, not identical.
• Product names differ between countries and years. We match them by name and a reviewed alias list, so occasionally two versions of a product are merged or one product shows up twice.
• Car seats are certified per market. The ranking shows where a product was tested, which tells you where it is sold.
• A missing product usually hasn't been tested by anyone we track. It doesn't mean the product is bad.