You finished barbell rows with 60 kg (132 lb) for 12, 12 and 12. Every guide says add weight. The smallest pair of plates in your gym is 1.25 kg, so the next load is 62.5 kg — a jump of just over 4%, and you will probably still get eight or nine reps. Fine.

Now do the same thing on a dumbbell lateral raise with 10 kg bells. The next pair up is 12.5 kg. That is a 25% jump on a small muscle in a long lever, and it will not cost you two reps, it will cost you five. The advice that worked on the row falls apart.

Double progression is the rule that handles both cases. It is not a beginner method you graduate out of — it is what most well-written intermediate programmes are doing underneath, and it is the reason "just add 2.5 kg" stops working around your second year.

The rule, in one paragraph

Pick a rep range instead of a rep target. Keep the load fixed and add reps session to session until every working set reaches the top of the range. Then, and only then, add load — and drop back to the bottom of the range. You have two variables progressing in sequence, which is where the name comes from: reps first, load second, round and round.

On a 3 × 8–12 row at 60 kg that looks like 10/9/8, then 11/10/10, then 12/11/11, then 12/12/12 — and now you add weight and start again at eight reps. Every one of those sessions added work.

Adding reps is real overload, not a consolation prize

The usual objection is that reps are what you do while you wait to add weight. The evidence does not support treating them as a lesser option. A 2022 PeerJ trial by Plotkin and colleagues took 43 people with at least a year of consistent lower-body training and split them for eight weeks: one group added load while holding reps constant, the other added reps while holding load constant. Muscle thickness differences came in under 1 mm at most sites, endurance under 1%, countermovement jump 0.1 cm. Rectus femoris growth modestly favoured the rep group; dynamic strength slightly favoured the load group, by an amount the authors called of questionable practical significance.

The authors' conclusion is the one worth carrying: both are viable strategies over a training cycle. A session where you went from 10 reps to 11 at the same weight was not a holding pattern. It was the overload.

The older ACSM progression position stand reaches the same place from the other side. Its instruction is to apply a 2–10% load increase "when the individual can perform the current workload for one to two repetitions over the desired number" — a load rule triggered by a rep achievement. That is double progression, written into a position stand in 2009.

Choosing the rep range

A range of 8–12 is the sensible default for most accessory and hypertrophy-oriented work, and it is what the 2021 re-examination of the repetition continuum points at as the practical middle of the useful zone. But the width of the range matters more than most people think, and it is the lever you have for the dumbbell problem.

A narrow range (say 5–8) means you reach the top quickly and add load often. That suits barbell lifts, where 2.5 kg is a small fraction of the load. A wide range (say 8–15) means you bank far more reps before each jump, which suits any lift where the smallest available increment is large relative to the weight — dumbbells, most cable stacks, plate-loaded machines with 5 kg steps, and anything you do with a kettlebell.

The test is arithmetic, not preference. Take the smallest jump available on that lift and divide it by your current working weight. Under about 5%, a narrow range is fine. Over 10%, you need a wider one, because you are going to need more accumulated reps to absorb the jump.

Sizing the jump

The percentage most programmes use

The ACSM range of 2–10% is wide because it has to cover a 180 kg squat and a 6 kg lateral raise. A multi-joint lift tolerates the larger end; a single-joint lift needs the smaller one. A reasonable split is around 5% for compounds and 2.5% for isolation work, rounded to whatever your gym can actually load.

On the 60 kg row, 5% is 3 kg, which rounds to the nearest available 2.5 kg step: 62.5 kg. On a 100 kg squat, 5% is 5 kg, which is two 1.25 kg plates a side. On a 12 kg dumbbell curl, 2.5% is 0.3 kg, which rounds to nothing at all — and that is the case the percentage alone cannot solve.

When the smallest available jump is still too big

Whenever the percentage rounds below one increment, you are forced to take one full increment or none. There are four honest ways out, and they are worth knowing in order:

• Widen the range on that lift. Run the lateral raise at 8–15 instead of 8–12. You will take the 25% jump from a much higher rep base, so landing at eight reps afterwards is realistic rather than fantasy.

• Add a set before you add load. Going from three sets of 12 to four sets of 12 at the same weight is a 33% increase in working volume with no jump at all, and it buys you several more weeks on that dumbbell.

• Micro-load. Fractional plates of 0.5 kg or 1.25 kg, magnetic add-ons for fixed dumbbells, or an adjustable dumbbell with 1 kg steps turn a 25% jump into a 5% one. This is the only option that actually fixes the problem rather than working around it.

• Change the exercise rather than the load. A single-arm variation, a slower eccentric, or a longer pause at the hard position all raise demand without touching the number on the bell. Use this sparingly — it makes progress harder to read in your log.

There is a clean test for whether the jump you took was the right size: after the increase, you should still be able to complete the bottom of your rep range on every set. If the new load drops a set below the bottom of the range, the jump was bigger than double progression can absorb, and you should either go back and bank more reps or find a smaller increment.

The sets that do not all reach the top

This is where most people quietly cheat: you get 12, 12, 9 and call it a top-of-range session because two sets hit twelve. Judge the session by your worst working set. If the third set was nine, the lift is at nine, and the load stays where it is. This one rule is what keeps double progression honest, because the alternative — progressing off your best set — has you adding weight to a lift that is already failing at the end of the session, and it is the most common way a well-intentioned programme turns into a stall.

The other half of the rule: if every set cleared the bottom of the range, you had a successful session. Nothing is wrong. Chase the reps.

When you stall

Sooner or later a load stops moving. You get 8, 8, 7 at 62.5 kg, then 8, 7, 7, then 7, 7, 7. Three sessions in a row where a working set fell below the bottom of the range is enough evidence. Do not keep grinding it, and do not overhaul the programme.

Take roughly 10% off, reset to the bottom of the range, and climb again. On 62.5 kg that is 56.25 kg, rounded down to 55 kg with the plates you have. You will usually pass the old sticking point within three or four sessions, because what stalled was rarely the muscle — it was accumulated fatigue meeting a load you had stopped being able to recover from.

For the volume and intensity this sits inside, our write-up of the 2026 ACSM resistance training guidelines covers what the current evidence says about set counts and load for healthy adults.

Eight weeks on one lift

Here is the whole method on a single exercise — barbell row, three working sets, 8–12, starting at 60 kg, with 2.5 kg the smallest jump available.

1. 60 kg — 10, 9, 8. Worst set is 8: bottom of range cleared. Hold the load.

2. 60 kg — 11, 10, 10. Hold.

3. 60 kg — 12, 11, 11. Hold.

4. 60 kg — 12, 12, 12. Top of range on every set. Add load.

5. 62.5 kg — 9, 9, 8. Bottom cleared, so the jump was correctly sized. Hold.

6. 62.5 kg — 10, 10, 9. Hold.

7. 62.5 kg — 11, 11, 10. Hold.

8. 62.5 kg — 12, 12, 11. Hold — one set short.

Eight weeks, one load increase, and every session added either a rep or a kilogram. That is the pace double progression runs at once you are past your first year — and it is faster than adding 2.5 kg every week for a month and then deloading twice.

Where Hefty fits

Hefty runs exactly this rule for you, per lift, so you are not doing the arithmetic between sets. It reads your last session, and when every working set reached the top of your range at the reference load it prescribes the next weight — 5% on multi-joint lifts, 2.5% on single-joint ones, rounded to the smallest jump you told it you have, with a floor of one increment so a rounding error never produces "no change". When you clear the bottom of the range but not the top, it holds the weight and gives you a rep target instead. Three sessions where a set falls below the bottom of the range and it takes 10% off and restarts you at the bottom.

The two settings this article is really about are per-exercise and free: the rep range, and the smallest weight jump for that lift — 0.5, 1, 1.25, 2, 2.5 or 5 kg, or 2.5, 5 and 10 lb if you work in pounds. Dumbbells default to 1 kg steps and kettlebells to 2 kg. Set the jump to what your gym actually has and the prescriptions stop asking for weights that do not exist. Changing the percentage itself, the number of misses before it backs off, or the size of the drop are per-exercise controls under Hefty Pro; the progression is free.

If you already log somewhere else, our comparison of Hevy and Strong covers what those two do well — and the decision neither of them makes for you.

Pick your hardest lift, set its rep range and its smallest jump to what you actually have available, and see whether the weight Hefty prescribes next session matches the one you would have picked yourself. If it gets something wrong, tell us what it got wrong.