How we test recipes

A recipe earns its place here by beating the alternatives in a direct comparison — and then working again, on a different day, for someone following only the written page.

1. Choosing the contenders

Before cooking anything, we read widely and note where well-known versions of a dish actually differ. Most dishes come down to three to five real decisions: a technique (charring peppers before the onion goes in, or not), a ratio (yolks to whole eggs), or one ingredient (porcini, buttermilk, black vinegar). Those decisions become the test plan. Versions that differ only in garnish are left out.

2. Cooking side by side

Contenders are cooked in pairs, at the same time where the kitchen allows, with ingredients from the same shop and pans of the same size. Each round changes one thing. A typical recipe goes through four to eight rounds before a winner is clear.

3. Tasting without labels

Tasters — the household and whoever is around for dinner — get two plates and no explanation. We ask a deliberately plain question: which would you make again? It turns out to be more useful than asking which one people like.

4. Repeating the winner

The winning version is cooked at least three more times on different days, and at least once by someone working only from the written recipe while we stay out of the kitchen. Anything they stumble over gets rewritten.

5. Measuring

• Dry ingredients and most liquids are weighed in grams; small amounts use level measuring spoons (1 tsp = 5 ml, 1 tbsp = 15 ml).
• Oven temperatures are for a conventional oven and checked with a standalone thermometer. For a fan oven, reduce by about 20°C unless the recipe says otherwise.
• Timings are ranges, and every step also says what to look for, because hobs and pans vary more than recipes admit.
• Eggs are large (about 60 g out of the shell is a reasonable guide).

6. US measures

The recipes are developed in metric. The US toggle converts weights to ounces and pounds and liquids to cups and spoons, then rounds to practical fractions. For baking, a scale gives far more reliable results than cups.

7. What “best” means here

We are one kitchen, not a laboratory. “Best” means the version that won our comparisons for cooking at home on an ordinary evening, with ingredients from an ordinary shop. Every recipe lists the versions it beat under What we tested, so if your priorities differ — speed over depth, say — you can see which way to go.