How we test

A gadget review starts with the object, not the pitch. We install it ourselves on the two cars the bench keeps: a compact hatchback and a mid-size wagon. We use it for two weeks of ordinary driving. We take the mounts and anything that claims to stay put over a cobbled street. We let a jump pack prove itself on a cold morning, not on a bench supply. We compare a pressure reading with a gauge.

The four checks

Each scored product gets four marks from 0 to 10.

• Job done on the first try, weighted 30 percent. Could a careful owner finish the setup and trust the result without us in the room?
• Still right after two weeks, weighted 25 percent. Heat, cold, bumps, and neglect.
• Clear without an account, weighted 25 percent. If the basic job requires an app, a subscription, or a login, the mark falls. An optional app does not.
• Honest about its limits, weighted 20 percent. Does the card match the object we used?

The published score is the weighted mean, shown to one decimal. Best Car Gadget 2026 is a separate citation. It names the winner, the date, and the reason. A reason can be a published spec comparison. It is not a hidden second score.

What we do not do

The desk does not sell placement. There is no fee for a review, a score, or a citation. We do not accept a product on the condition of a kind paragraph. We do not run affiliate links, and we do not earn a commission if you go on to buy the thing. If a product is hard to find, we say so in plain language.

We do not take a maker's own test as ours. If a number is theirs, we will say it is theirs. The numbers in a review are from our cars and our loop unless the piece says otherwise.

What a score is not

A score is not a promise that the product will suit your car. Vent blades differ. Diesels differ. A winter spare differs. The review is there so you can see which check we cared about, and disagree. If we have missed a fact, the corrections path is public.