What to Measure in Your Training and What to Ignore
Introduction
A watch that estimates calories, a scale that calculates body fat percentage, an app that assigns a recovery score, a ring that rates your sleep. It has never been this easy to measure, and never this easy to confuse measuring with knowing.
The problem with over-measuring isn't the time it takes: it's that noise looks a lot like information, and acting on noise is worse than not measuring at all.
If a number won't change what you do this week, don't record it.
The criterion: action and speed
Two questions filter almost everything.
Can I act on this? A data point with no associated decision is entertainment. Knowing your heart rate variability without being clear what you'd do differently depending on the value contributes nothing.
Does it change more slowly than I'm measuring it? Here's the most common error. Body weight fluctuates a kilo or two from water, salt and gut contents; measuring daily and reacting to each reading is reacting to noise. Body fat changes over months; measuring it weekly with a home scale produces variations that correspond to nothing real.
The rule that follows: measurement frequency has to match the speed at which the thing actually changes.
What is worth measuring
Sessions completed. The most important data point and the most ignored. One mark on the calendar per session, and by month's end you have the answer to 80 % of your questions about why you aren't progressing.
Load and reps on your main exercises. The direct measure of progression. Without this there's no training, there's physical activity.
Body weight as a weekly average. Weigh yourself daily if you like, but look only at the seven-day average. The individual reading means nothing; the three-week trend does.
Hours of sleep. The variable with the best ratio between what it costs to record and what it explains.
Waist circumference, once a month. Simple, cheap, and better correlated with metabolic health than weight.
One effort rating per session. A number from 1 to 10 when you finish. Useful for spotting overload weeks when you look at them together.
What to ignore
The calories your watch estimates. The least reliable metric on the device, and we'll come back to that.
The scale's body fat percentage. Bioimpedance scales have high error for the absolute value and are sensitive to hydration, time of day and recent meals. As a single number it's useless.
Composite recovery or sleep scores. They're proprietary formulas no company publishes or independently validates. You're acting on a black box.
The daily weight reading. As an isolated figure it informs nothing, and it does generate anxiety.
Watch-estimated VO₂ max. An approximation of an approximation. Useful for an annual trend; useless for deciding Thursday's training.
The trap of measuring what's easy
There's a phenomenon worth keeping in mind: we tend to measure what the device offers, not what matters. And there's a known effect — formulated as Goodhart's law — that when a measure becomes a target, it ceases to be a good measure.
You see it constantly in training. Someone chasing their watch rings ends up walking in circles at eleven at night. Someone chasing burned calories picks the exercise that estimates the most, not the one that helps most. Someone chasing a sleep score sleeps worse from monitoring it.
The number indicated something. Turning it into a goal breaks its link to the thing it indicated.
How to build your minimum dashboard
1. Pick five metrics, not fifteen. Sessions, load on two or three key exercises, weekly average weight, hours of sleep, monthly waist. 2. Set each one's frequency according to how fast it changes: per session, weekly or monthly. 3. Define what decision each one drives. If you can't write the decision, drop the metric. 4. Review weekly, decide quarterly. The weekly look is for spotting drift; plan changes are made with twelve weeks of data. 5. Turn off notifications for the rest. It isn't technophobia: it's keeping noise from competing with signal.
Common mistakes
Measuring to feel in control. Measuring is reassuring, and that feeling is independent of whether the data helps.
Changing the plan over one reading. No training decision should depend on a single data point from a single day.
Comparing measurements taken under different conditions. Weighing yourself dressed then undressed, measuring your waist relaxed then braced.
Trusting the precision the device displays. That it says 14.3 % doesn't mean the third digit means anything.
Dropping the boring measures. Sessions completed and weight lifted are the least attractive metrics and the ones that explain the most.
Conclusion
Measuring well isn't measuring a lot: it's choosing a few indicators you can act on and watching them at the speed they actually change.
Make it concrete today: write down your five metrics and, beside each, the decision you'd take if it worsened. Delete any for which you can't write that decision. You'll end up with a short, boring list that's far more useful than your watch's dashboard.
References
- Harkin, B., et al. (2016). Does monitoring goal progress promote goal attainment? A meta-analysis of the experimental evidence. Psychological Bulletin, 142(2).
- Shcherbina, A., et al. (2017). Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. Journal of Personalized Medicine, 7(2).
- Strathern, M. (1997). "Improving ratings": Audit in the British University system. European Review, 5(3). (Formulation of Goodhart's law.)
- Ross, R., et al. (2020). Waist circumference as a vital sign in clinical practice: A consensus statement from the IAS and ICCR. Nature Reviews Endocrinology, 16(3).
- Buchheit, M. (2014). Monitoring training status with HR measures: Do all roads lead to Rome? Frontiers in Physiology, 5.