Run the breakfast experiment: two weeks, one variable

Almost every client arrives having tried something total. Whole30, keto, no sugar for a month, a plan from a book. Some of them worked for a while. Almost none of them produced knowledge, because when you change eleven things at once and feel better, you have learned nothing about which of the eleven mattered.

What I ask for instead, usually in the first session, is a two-week experiment with a single changed variable. The one I choose most often is breakfast, and there are three reasons for that.

1. 1It is the most repeatable meal of the day. Most people eat roughly the same breakfast every weekday, which means it is the easiest thing in your diet to hold constant or to change deliberately.
2. 2It is usually the worst-constructed meal of the day. The typical breakfast lands around 10 grams of protein and 2 grams of fiber, which is a poor start to a day in which you will make roughly two hundred food-adjacent decisions.
3. 3Its effects show up later, at the times you are struggling. The 4pm crash and the 9pm grazing are often downstream of what happened at 7am, and almost nobody connects them.

The protocol

Week one: change nothing. Eat your ordinary breakfast. Every day, at three fixed times — 10:30am, 3:30pm, and 8:30pm — write down two numbers on a scale of one to five. Hunger, and energy. That is the entire data collection. It takes four seconds and you must do it at the fixed times rather than when you happen to remember, because when you remember is not random.

Week two: change exactly one thing about breakfast. Get it to 30 grams of protein. Do not change lunch, do not change dinner, do not start exercising, do not cut anything else out. Keep taking the same two numbers at the same three times.

At the end, compare the fourteen sets. You are not looking for a dramatic difference — you are looking at the 3:30pm column, which is where this change shows up if it is going to.

Why 'it did nothing' is a real and useful result

Perhaps a quarter of clients run this and report no difference at all. That is genuinely informative, and I want to be clear that I mean it rather than saying it kindly. It means their afternoon problem is not a breakfast problem, and we have eliminated the most common explanation in fourteen days rather than spending three months assuming it.

In those cases we usually go looking at sleep, at whether lunch is actually being eaten, or at the structure of the afternoon itself — because a 4pm biscuit that happens every single day at the same desk after the same meeting is not hunger. It is a cue. And cues respond to different interventions than hunger does.

The generalisable part

The breakfast is not the point. The method is. One variable, two weeks, a fixed measurement taken at fixed times, and a result you are willing to have go against you. Everything we do in a twelve-week block is this shape, run four or five times in sequence, and by the end a client has a small body of knowledge about their own physiology that is worth considerably more than a meal plan.

A meal plan tells you what to do this month. An experiment tells you something about yourself that is still true in five years.