reckoner
Most programs add weight on a calendar. Monday comes around, the number goes up, and nobody asks whether last week earned it. That works until the week it does not, and then it keeps going anyway, because a spreadsheet has no way of knowing that your last set was a grind. Progression models built that way assume your body adapts at a predictable rate and that you turn up able to express it every session, and neither of those is a safe assumption3.
Reckoner works the other way. Nothing moves until the training says it should, and when something does move you are told which rule moved it. The input is not the date. It is what you lifted, set by set, the effort you reported when you finished, and what you said about the form.
Reps and load are both progressive overload
Progressive overload is what keeps training working: once your body has adapted to what you ask of it, the demand has to keep rising or it has no reason to keep changing6. What gets missed is that the bar is only one of the things that can rise. Reps count. Sets count. Over ten weeks of leg extensions in people new to training, one leg was given more load as it became available and the other more reps, and both raised that leg's one rep max and the size of the muscle, with no significant difference between them7.
So the order is a decision rather than a law, and it is one Reckoner makes on your behalf every week. Reps first, because that is the safer of the two routes up: you earn control of the bar across a full range before you make it harder to control. The limit on that is worth saying out loud. For a trained lifter chasing maximum strength, load is the variable that has to move eventually, and no amount of reps substitutes for it2. Reps first is a sequencing rule, not a way of avoiding heavy weight.
Reps come before load
Every lift carries a rep range rather than a rep target. The first thing the engine asks is whether you finished every set at the top of that range. Until you have, the load stays exactly where it is and you climb the range instead, a rep at a time. The effort you report does not decide this: the reps are counted and the effort is an estimate, so the effort decides whether you are owed more rest or a lighter week, and never whether the bar moves.
That order matters more than the size of the jump. Reps buy you the strength to hold a heavier bar in a position you can still control. Skip that step and you get to week five with a bar path that has quietly gone crooked and a number you cannot repeat.
None of that is Reckoner's invention. The American College of Sports Medicine's 2009 progression guidance said the same thing about the order: put the load up once the current one can be carried a rep or two past the target, and not before1.
For a lifter still in their first months the range itself is fixed wider, at eight to twelve reps, which is the range that guidance wrote for someone new to training rather than for someone with years under the bar1. The engine also waits a little longer there before it takes the jump: the load only moves once the reps you can do at the new weight would still land at eight or more, so an early set that feels easy is not mistaken for a green light.
The form answer holds the reps too. Rough holds them where they are, and so does fair on a set you rated nine or ten, because the next rep would be asked for right where your technique was already slipping, and technique measurably drifts as a set gets heavier and more fatiguing4. Fair on a set with two or more good reps left still climbs, because a flaw that shows up that far from the end is a skill to practice rather than a sign of fatigue. The form answer only ever holds or lowers. It never adds. How it is used is our coaching judgment rather than a tested rule: nobody has yet studied a self rated form answer as a gate on progression.
Build the range before the load.
Then the load, in the smallest honest step
When the range is clean the engine adds one step, and the reps come down by what that step costs rather than back to the bottom of the range by rule. It reads the cost off a published table of how many reps people manage at a given share of their best5, so a small step might take you from ten to eight. It never lands you below a floor: six reps on a main lift and eight on an accessory, or the bottom of the lift's own range where that is lower. Here Reckoner departs from that 2009 advice, which put that step at two to ten per cent of the load1. The step is not a percentage: it is the smallest thing you can actually pick up in the gym you train in. A barbell takes the smallest pair of plates you own. A dumbbell rack goes up in twos. A pin stack goes down one pin. A percentage of a bar is a number nobody can select, and a prescription you cannot load is worse than one that is slightly too small.
Small steps mean more weeks where the reps are good, and more weeks of good reps is the whole game. A five per cent jump on a lift you have just barely finished is a week of grinding followed by a week of backing off.
Six weeks of one bench press
This is not an illustration. The table below is the output of the same engine that writes your week, run for six weeks against a lifter who finished every set as prescribed and reported the same effort each time. Nothing here was typed in by hand.
| Week | Prescribed | What it did |
|---|---|---|
| 1 | 4 x 6 at 60 kg | First week Where the run starts. Nothing is logged yet, so nothing has changed. |
| 2 | 4 x 7 at 60 kg | More reps 6 → 7 repsYou finished every set with three reps left in the tank (RPE 7) and room left in the range, so the reps climb before the load does. |
| 3 | 4 x 8 at 60 kg | More reps 7 → 8 repsYou finished every set with three reps left in the tank (RPE 7) and room left in the range, so the reps climb before the load does. |
| 4 | 4 x 9 at 60 kg | More reps 8 → 9 repsYou finished every set with three reps left in the tank (RPE 7) and room left in the range, so the reps climb before the load does. |
| 5 | 4 x 10 at 60 kg | More reps 9 → 10 repsYou finished every set with three reps left in the tank (RPE 7) and room left in the range, so the reps climb before the load does. |
| 6 | 2 x 10 at 60 kg | Deload 4 → 2 setsFive weeks of building, then one week lighter. About 40% of the sets come off. The load holds and your reps stay where you earned them, so you keep the strength and lose the fatigue. |
| Next | 4 x 8 at 62.5 kg | Load up 60 kg → 62.5 kgYou earned this before the lighter week. The lighter week was for recovery, not a test, so the rise waited for it to end. Now the bar goes up to 62.5 kg. The lighter week is over, so this is back to four sets. |
Real engine output, not an illustration. Every row was produced by running the engine itself when this page was generated.
Two things worth reading twice. Week six is the planned lighter week: five weeks of building and then one where about 40 per cent of the sets come off, with the load and the reps held exactly where you earned them. And the bar has not moved once in those six weeks: the first load step is the decision on the last row, once the reps have climbed the whole range.
Then sets, and then the movement itself
Some lifts have no load to add. A push-up and a pull-up are the weight you happen to be, so when the reps run out the engine adds a set instead, up to the ceiling that lane was built with. A hold has neither, so it gets seconds.
And when none of that is working, the lift itself is the question. A lift whose load, reps and sets have not moved for three trained weeks, with no extra rep on the first set either, is reset: about 10 per cent comes off the bar and you build back through it. If it stalls again after that, it is offered for a swap rather than another week of the same, which is the subject of its own guide. Waiting three weeks is our coaching judgment and taking off 10 per cent is a common coaching convention; no trial has tested either number.
What to do with this in your next session
Find the top of your range
Look at what the lift is asking for rather than at the weight. The number that decides your next session is the top of the rep range, and whether every set reached it. One set short of the top is not the top, and it is the commonest reason a load sits still for a month.
Report the effort and the form honestly
The engine can only read what you tell it. The effort does not move the bar, the reps do; what the effort decides is whether you are owed more rest or a lighter week, so flattering it costs you the recovery you needed. The form answer is the one that can hold the bar: call a lift fair or rough when it was, and a load your reps would have raised waits until you can move it cleanly.
Take the smallest jump your gym has
When the range is finished, the step up is the smallest pair of plates on the rack, one notch on the dumbbells, or one pin on the stack. Not a percentage. A jump you cannot load is a prescription you cannot follow, and a jump twice the size of the one you needed is two weeks of grinding.
Give it three weeks before you blame the lift
A load that has not moved for a week is normal. One that has not moved for three trained weeks is information. The first answer is a lighter bar and a clean climb back, and a lift that stalls again after that is the one worth changing for another movement rather than driving the same load into the same joint for another month.
Why you are told all of this
Every one of those sentences is written by the engine at the moment it decides, from the numbers it had in its hands. There is no second copy of the rules written down somewhere by a person, which means there is nothing to go stale: change what the engine does and every sentence on this page changes with it, including the ones in the table above.