Ninety days, and the three things worth writing down

The hardest part of judging a supplement is that you are the instrument, and the instrument has opinions. Memory rewrites itself towards whatever was expected, and a quarter is long enough for that to happen twice — once towards hope in the first month, once towards disappointment in the third.

The fix is boring and it works: write three things down before starting, and then do not change them.

They should be things that were already being tracked, or could have been. A number from a training log. The hour of the day when work stops being difficult. How many nights in a week end with a decision to go to bed rather than a discovery that it is one in the morning. Anything that can be answered without an interpretation.

What does not belong on the list is anything shaped like the marketing. If the reason for buying was a word on the front of the jar, that word will be found in the data whether or not it is there — not through dishonesty, but because a person looking for something finds it.

The timing matters as much as the metrics. Day seven is too early to mean anything and is exactly when most people decide. Day thirty is when the first honest reading is available, and it is worth treating as a reading rather than a verdict. Day ninety is the decision, because it is the first point at which a season has passed: a holiday, a bad fortnight at work, a week of poor sleep, a couple of missed evenings. A result that survives all of that is a result. A result measured across three good weeks in spring is a description of spring.

And it is worth deciding in advance what would count as a no. A protocol that cannot fail is not being evaluated — it is being subscribed to.