How We Test Recipes
Two cooks minimum, ordinary home equipment, and a clear list of the things that send a recipe back for another round.
Last updated:
Every recipe on this site has been cooked before it was published. This page explains exactly what that means, because "tested" is a word that gets used very loosely online.
The minimum bar
A recipe is not published until it has been cooked at least twice: once while developing it, and once more by following the written draft exactly, as a stranger would. That second cook is the real test. It is where we find the step that said "grind until smooth" when it should have said "grind until it floats in water".
Home equipment only
All testing happens in an ordinary domestic kitchen: a gas hob, a standard pressure cooker, a regular mixer-grinder, a cast iron tawa. We do not test on restaurant equipment, because almost nobody reading this has any.
Batter recipes are tested on both a wet grinder and a mixer-grinder, and both methods are written up. A batter guide that only works on equipment most readers do not own is not a finished guide.
Climate and fermentation
Fermented batters behave completely differently at different temperatures, so we test them across the range we can produce, including a deliberately cold run using an unheated room in winter. That is why our batter guides give timings banded by temperature instead of a single number. It is the single most common reason a batter recipe fails for a reader.
What sends a recipe back
- Any instruction that cannot be acted on without already knowing the dish.
- A timing with no sensory check attached to it.
- A failure we could not explain. If we cannot say why a test batch went wrong, the guide is not ready, because we cannot then write the troubleshooting section.
- A claim about a regional version that neither of us can source or has cooked.
Troubleshooting sections
Every guide ends with named failures and their causes, and those come from real failed batches rather than from imagination. When readers report a failure mode we had not seen, we try to reproduce it, and if we can, it gets added.
Re-testing
Published guides are re-cooked periodically, and always when a reader reports that something did not work. If a re-test contradicts what is on the page, the page changes and the update date moves.