Guide
How to test a recipe before you publish it
Your recipe works in your kitchen. Does it work in anyone else's? A practical way to test recipes: what to change, what to write down, and the one test most creators skip.
You've made the brownies four times. Two batches were perfect, one was a bit dry, and one you ate standing at the counter before you could decide. Is the recipe ready to publish?
Probably not yet. Not because the brownies aren't good, but because "it works when I make it" and "it works when you make it" are two different claims. A recipe is a set of instructions for someone who can't see you cook. Testing is how you find out what you know but forgot to write down.
Every test needs a question
Making a dish again "to see" gets you dinner, not answers. Before each round, decide which of these you're trying to find out:
- Is the dish what I want yet? You're still developing it.
- Why did that batch work? You're trying to understand a change.
- Can someone else follow what I wrote? You're testing the instructions, not the food.
Before you start, write one sentence about what success looks like: "fudgy in the middle, crackly on top, one bowl, under an hour." It's much easier to judge a batch against that than against "better." And if the recipe promises a quick weeknight dinner, the test has to check that too. Bon Appétit's test kitchen judges recipes against what they promise readers, not just on taste. A delicious sheet-pan dinner that secretly needs a second pan for the sauce hasn't kept its promise.
The trap: changing two things at once
Here's an example that catches almost everyone. You're working on a snack cake:
| Batch | Flour | Pan | Result |
|---|---|---|---|
| A | 200 g all-purpose flour | 20 cm square metal pan | A bit dense in the middle |
| B | 200 g cake flour | 23 cm square metal pan | Exactly right |
Batch B wins. So cake flour is the secret, right?
You don't know that. You changed the pan too. A bigger pan means a thinner layer of batter, which bakes through faster, and that alone could explain the lighter middle. It might be the flour, it might be the pan, it might take both.
Whether that matters depends on what you plan to publish:
- Publishing batch B exactly as made? You don't need to know why. Bake B once more, exactly as written, to confirm it, then publish cake flour and the 23 cm pan.
- Want to tell readers all-purpose flour works too? Now you need to know. Bake both flours in the same 23 cm pan, keeping everything else the same.
- Want to offer a smaller pan option? Same idea: one flour, two pans.
The rule is simple. If you want to explain why something worked, change one thing at a time. If you just want a recipe that works, test the exact version you'll publish. And don't write "all-purpose or cake flour" into a recipe unless you've actually baked it both ways.
If you're moving to a different pan or yield, our guides to scaling a recipe and changing cake-pan size will help you plan the version to test.
Write down what you did, not what you meant to do
The most dangerous moment in recipe testing is the one you won't remember. The batter looked thick, so you added a splash of milk. The onions weren't browning, so you turned the heat up. The dish came out great, and the written recipe still says nothing about either fix.
Keep a short log for every test. Food Blogger Pro recommends writing down ingredients, measurements and steps as you cook, including places where the instructions felt unclear. A note this size is enough:
- Version and date, plus what this test is meant to find out.
- Ingredients as used: weights where you can. Cups depend on how you fill them, and weight is more precise. If your recipe uses cups, note the grams alongside.
- Equipment: pan size and material, oven setting, burner level.
- What you did and when: timings, plus what the food looked like at each stage.
- Every change you made on the fly. This is the line that matters most.
- The result, and what you'll try next.
Keep each tested version as its own dated copy. If you keep editing a single document, you'll eventually lose track of which quantities produced which batch.
One more thing: for meat, poultry, eggs and anything else with a safe minimum temperature, check it with a thermometer and write the reading in your log. Color isn't a reliable test. The FDA's safe food handling guide lists the numbers.
The test that changes everything: someone else's kitchen
Once the dish tastes the way you want, the question shifts from "do I like this?" to "can someone follow it?" You can't answer that yourself, because you know too much. You fill in gaps without noticing they're there.
Professional test kitchens handle this with a cross-test: after a developer finishes a recipe, a different editor cooks it exactly as written and reports back. You don't need a magazine staff to do the same thing. A friend, a sibling or a reader who cooks at about the level of your audience is enough.
Give them the written recipe and ask them to note:
- Anywhere they had to guess
- Ingredients or equipment that differed from yours
- How long each step actually took
- What they changed, and why
- How it turned out
The hard part is staying quiet. If they text you mid-recipe asking "how thick should this be?", answer them, then write the question down. Every question they ask points to something missing from the recipe.
Here's how that plays out. Your recipe says "mix well." Your tester asks whether a few flour streaks are okay. That question tells you "mix well" isn't an instruction, it's a guess. Check your notes: if you stirred just until no dry flour remained, write exactly that. If your notes don't say, you have to bake it again and pay attention to that step. Confident wording can't fix a step you never pinned down.
When a stage is hard to describe in words (soft peaks, the dough after kneading), that's where a photo earns its place. Our guide to recipe photos covers which moments to capture.
When is it ready?
There's no magic number of tests. A recipe is ready when:
- The version you'll publish has produced the result you described, and you've made it more than once.
- The written quantities and steps match what you actually did.
- Someone else has cooked it from the page without getting stuck.
If a tester got stuck because a step was vague, rewrite it and have someone check that passage again. If they got stuck because you aren't sure how the step should go, go back to the kitchen. Baking more batches won't fix a sentence that's still vague on the page.
Then turn your notes into the final recipe. Our guide on how to write a recipe walks through it. Keep the test log with the recipe, too, even once it's sitting in your recipe archive. The next time a reader asks "can I use all-purpose flour?", you'll know exactly what you tested.