RepStack
Training guides

Add reps before weight.

Double progression is the boring rule that keeps a workout log honest: keep the weight, add reps inside the target range, and only increase the load after every working set reaches the top.

Written and maintained by Alex Tirim, founder of RepStack — this is the rule his own training log runs on. Published June 10, 2026.

The rule

Do not add weight just because the calendar changed.

A useful progression rule should answer one question before the next session starts: should the lifter keep the load, add reps, add weight, or back off? Double progression answers that without pretending strength moves in perfect weekly steps.

If your target is 3 sets of 8-12, a session of 80 kg x 10, 9, 8 does not earn more weight. It earns the same weight again, with a better rep target. A session of 80 kg x 12, 12, 12 earns the next load jump.

Decision diagram by RepStack. It is a programming rule, not a medical rule: pain, injury, or unusual symptoms need a qualified professional, not a heavier next set.

Logbook example

The win is clearer before the bar gets heavier.

Session 1

80 kg x 9, 8, 8

Keep 80

Build reps inside the range.

Session 2

80 kg x 11, 10, 9

Keep 80

Progress happened, but the last set is not at 12.

Session 3

80 kg x 12, 12, 12

Add load

All working sets reached the top.

Session 4

82.5 kg x 10, 9, 8

Keep 82.5

The new load resets the rep climb.

Evidence note

The evidence supports progression. The exact rule is an implementation choice.

The American College of Sports Medicine's 2009 position stand on progression describes progressive overload, specificity, and variation as core resistance-training principles. It does not say every lifter should add load every week. Load, reps, sets, rest, velocity, and training status all affect the prescription.

More recent work points the same direction. A 2021 network meta-analysis by Lopez and colleagues found that adults doing resistance training built muscle across a broad load range when work was hard enough, while heavier loading remained more specific for maximal strength. A 2022 umbrella review of hypertrophy variables reached the same conclusion across rep ranges, and Grgic and colleagues' meta-analysis found training to failure is not required for most strength or size gains. That is why the rule here is conservative: earn the load jump instead of forcing it.

When it fails

Double progression is not a fix for every stall.

The jump is too large

Some dumbbell racks jump by 5 kg per hand. In that case, the next weight may be too big even after the range is complete. Use smaller jumps, more reps, slower tempo, or a different movement before pretending the jump is earned.

The exercise changed

A machine row, chest-supported row, and barbell row should not share one progression history. The exercise identity has to stay stable for the log to mean anything.

Technique is degrading

Reps only count if they are comparable. If every extra rep is shorter, sloppier, or more dependent on momentum, the log is recording a different lift.

Pain changes the lift

This page is not injury advice. Stop chasing progression and get qualified help if pain changes your setup, range of motion, or willingness to load the movement.

How RepStack runs this rule

The diagram above is the engine that picks your next session.

RepStack's progression engine walks a fixed ladder over your logged sets before every workout. These are the actual triggers, from the code that ships in the app:

  1. 1

    Every working set reaches the top of the range

    Weight goes up by the smallest practical jump — 2.5 kg on a barbell, 2 kg on dumbbells (5 lb in pounds) — and the rep target resets to the bottom of the range.

  2. 2

    Any working set lands below the bottom of the range

    The weight holds. The target is the range bottom on every set.

  3. 3

    Two sessions in a row below the bottom

    The load drops 10% so you can climb back instead of grinding.

  4. 4

    You rate a session “too hard”

    Same call: 10% off the bar next time, reps rebuilt from the bottom.

  5. 5

    Your estimated 1RM is flat across 4 sessions

    A 10% reset, but only if your reps are not already at the top — a plateau at max reps is not a stall, it is one session away from a weight jump.

  6. 6

    Three identical sessions — same weight, same reps

    The engine changes the stimulus instead of the load: heavier weight for sets of 6 if you were working at 8 or above, lighter for sets of 10 if below.

  7. 7

    Rebuilding after a deload, every set back at the top

    It skips the slow climb and returns you straight to the pre-deload weight.

RepStack's Today's Plan sheet: six push exercises queued with the note 'Same weight, one more rep per set', each lift showing its load and a rep bump such as 230 lbs, 6 to 7 reps.
A real plan from the app, the session after a lifter finished inside the range on every push exercise: same weight, one more rep per set. Every number is a suggestion computed from the logged sets — you can override any of them, and the engine adapts to what you actually did.

That is product logic from your workout history. It is not a diagnosis, recovery score, injury screen, or proof of your true one-rep max — the estimated 1RM behind triggers 5 and 6 is a coaching estimate from your logged sets, nothing more. The full math behind that estimate, and the Strength Score it feeds, is documented on the Strength Score methodology page.

Run the rule on your next workout

Free on iOS — double progression is not paywalled.

The bottom line

The bar earns weight. The calendar does not.

Pick a range, log every working set honestly, and let the log make the call: reps first, weight second, 10% off when two sessions in a row say the load is lying to you. Most lifters who stall are adding weight on a schedule. Most lifters who keep progressing let the reps set the schedule.

Sources

What this page is based on.

  1. ACSM 2009 position stand on resistance-training progression
  2. Lopez et al. 2021 network meta-analysis on training load
  3. Grgic et al. 2022 meta-analysis on training to failure
  4. Bernardez-Vazquez et al. 2022 umbrella review on hypertrophy variables

Spotted an error? Email [email protected]. Factual errors are corrected within 48 hours, with a visible note on this page.