How to Tell If Your Training Program Is Actually Working

Soreness does not measure progress. It measures novelty and microstructural damage, and it peaks exactly when you change an exercise or come back from a break. A program that reliably leaves you sore is usually a program that changes too often.

"Feeling the muscle" misleads the same way. Pump and local burn depend on tempo, rest and rep range, not on whether your volume dose fits your capacity to recover. A meta-regression of 67 studies shows the volume–outcome relationship is real but with clearly diminishing returns [1] — while the sensation of fatigue keeps climbing linearly far longer than the effect does.

The third misleading metric is the mirror. Changes in appearance depend on hydration, glycogen, lighting and time of day to a degree that completely swamps real weekly gains. Muscle accrues at a rate measured in grams per week; water swings can be ten times larger.

That does not make appearance irrelevant — for many people it is the whole point. It means appearance is unusable as a decision metric in any window shorter than eight weeks.

The first is your e1RM trend on the main lifts, read as a three-session rolling average. A single workout depends on sleep, caffeine and time of day; a three-session trend filters out most of that noise. If the average climbs 1–2% per mesocycle, the program is working.

The second is the share of prescribed reps you complete. If the plan says 3×8 and you deliver 8/7/6, you have a volume gap that will not close by itself. The third is load at a fixed rep count — if the same 8 reps now happen with more weight than four weeks ago, adaptation is happening regardless of what the scale says.

The fourth is consistency: the percentage of planned sessions you actually did. Below 80%, judging the program becomes impossible, because you are judging a plan that was not executed. Load7 lines those four up side by side, so it is visible whether the problem is the dose or the delivery.

Those four metrics have another advantage: read together, they point at a specific cause. A rising e1RM with falling consistency means the plan works and you are not doing it. A flat e1RM with full attendance and a high completion rate means the dose is too low.

The worst combination is a falling e1RM alongside a falling completion rate despite full attendance — an unambiguous recovery deficit, and the one situation where cutting volume is the first-choice answer.

Four weeks is the minimum, six is better. A shorter window mixes adaptation with day-to-day variance and leads to changing the plan every fortnight — at which point every stimulus is new, so every one produces soreness and none gets time to work.

That patience has a limit. If after six weeks your rolling e1RM average is flat or falling while completion rate drops, this is not a matter of time — the current dose exceeds your capacity to recover.

The weekly volume meta-analysis shows how uncertain these thresholds are at population level: the confidence intervals for low, medium and high set counts overlap heavily [2]. Your own data is a better source than the study average in that situation.

It is also worth defining in advance what counts as success. "I want to get stronger" is not a criterion; "I want 10 kg on my squat in 12 weeks at the same body weight" is. Without a written target, any result can be rationalised and no plan is ever rejected.

Set the target at a level appropriate to your training age. For an intermediate, 1–2% e1RM growth per mesocycle on the compounds is a realistic rate. Expecting 5% every four weeks leads to abandoning plans that were working correctly.

Once you diagnose a stall, the temptation is always the same: change the split, the exercises and the volume at once. Then you learn nothing about what worked. Start with one variable — usually volume down, not up — and give it four weeks.

The order that usually works: sleep and session completion first, then volume, then load and rep ranges, then exercise selection last. Periodization helps strength relative to no structure at all (effect size 0.31), but it will not substitute for volume you are not performing [3].

The order of variables matters because each successive one is more expensive to verify. Fixing sleep shows up within a week. Changing volume needs four weeks. Changing exercise selection resets data comparability and really costs two mesocycles before you know whether it helped.

That is why exercise selection sits last, even though it is intuitively the most tempting. A new exercise produces immediate improvement from motor learning, which is very easy to mistake for having solved the problem.