How we rank

1. Find the lists

We search the web for what you typed. Store pages, videos and forum threads drop out first. An AI model then looks at each result and keeps only review lists: pages where someone other than the seller recommends several things.

A brand's list of its own products doesn't count, and neither do news stories or pages that only link to other lists. If too few lists are left, we stop and tell you, instead of ranking on too little.

2. Read each list

We open each list and pull out every pick it recommends, honorable mentions included. Things it mentions in passing, sidebars of other articles, and picks it no longer recommends are left out. For each pick we keep one sentence on what that list thinks of it.

Some sites, such as Wirecutter, block automated reading. We don't work around that. Those lists appear under "Couldn't read" with the reason, and you can add one to your own ranking with the browser extension.

3. Match the same pick

Lists name the same thing in different ways, like "All-Clad D3 10-inch" and "All-Clad D3 Stainless 10 in." An AI model groups mentions that share a brand and model into one pick. It keeps versions apart: an All-Clad D3 and a D5 are different pans.

4. Count

Each list counts once for each pick it recommends, however often it mentions it. Picks are ordered by how many lists recommend them, and the top ten make the ranking. Where a pick sits on a list doesn't change its count: first on one list and eighth on another is two lists either way.

5. Describe each pick

For each pick in the ranking, an AI model merges the lists' sentences into a short description. It uses only what the lists said, with no specs, prices or claims from anywhere else. Every pick links to the lists that recommend it, so you can check the source.

What doesn't count

Nobody pays to be ranked, and we don't add picks of our own. Amazon links carry our Associates tag, so we earn from qualifying purchases there, but that plays no part in the count. Links to other stores stay the ones the list used.

How fresh it is

We save each ranking, so the next person to search the same thing gets it right away. Saved rankings get run again from time to time. While a new run is in progress, you see the saved one.