We set out to build a tracker and ended up deleting it.

The reason they gave was never “logging is too slow”. We had made logging about as fast as it can get. The reason was that the app only ever told them, after the fact, that the day had already gone wrong. It was a very efficient way of finding out you had under-eaten protein by four in the afternoon, when there was not much left to do about it.

That is a design problem, not an effort problem. A tracker asks you to make every food decision yourself, unaided, and then does arithmetic on the result. All of the difficulty is in the part it does not help with.

So we moved the work to the front. Decide the menu once, properly, with the targets in hand — and then stop deciding. That is the whole of Protn now. Everything else got deleted, including features we liked.

What it does instead

You answer four questions: your macro targets, the foods you would happily eat on repeat for weeks, which meals you want in a day, and whether to track water. Protn builds a week of meals from your foods that lands on your targets. After that the app is a checklist.

Every main meal is one protein, one carbohydrate, one vegetable and one fat, given in units a person actually uses — two eggs, one banana, 150 g of chicken. No recipes. Nothing to shop specially for. The menu is boring on purpose, because boring is what you can keep doing in week six.

The decisions we would defend

1. 1

   A plan that misses your targets is never saved

   Not by the generator and not by your own hand edits. Protn would rather tell you it could not build something than hand you a menu that quietly does not add up.

2. 2

   Calories always come from the macros

   Four per gram of protein and carbohydrate, nine per gram of fat — everywhere, every screen. Food tables disagree with their own macro columns, and letting both methods into one app is how a plan passes every check and is still wrong.

3. 3

   Portions snap to what you can measure

   A whole egg, half a banana, a weighable number of grams. An instruction to eat 215 g of banana is arithmetically correct and useless.

4. 4

   Meat raw, starches cooked

   Because that is when the food is on the scale. Cooked chicken is about a third higher in protein than raw, so getting this wrong inflates every plan built from it.

5. 5

   Nothing is taken away if you stop paying

   A lapsed subscription keeps the full menu, readable. You paid for that plan; taking it back would be punishing you for having bought it.

What we are not claiming

Protn records the edits you make and turns repeated ones into preferences that later menus take into account. The mechanism is built and running. It has no results behind it yet, and we are not going to describe it as proven until it does. You can see every preference it has stored, where each came from, and delete any of them.

We also will not pretend the app suits everyone. If what you want is to record whatever you happened to eat, scan a barcode, or get a nutrition breakdown of a restaurant meal, Protn does none of that and is not going to. We would rather lose the download than take the money and disappoint you in week one.

And it is general nutrition guidance, not medical advice. The targets come from a standard metabolic formula applied to what you entered about yourself, which is an estimate about a population rather than a measurement of you. If you are managing a health condition, pregnant, or under 18, talk to a doctor or dietitian before you change how you eat.

Who is behind it

Protn is built and run by one developer. That is why support is an email address rather than a ticketing system, and why the reply usually comes from the person who wrote the code that broke. If something is wrong, or you want to argue with one of the decisions above, write to hello@protnapp.com.