The method

The zone, and why it is a range rather than a percentage

One piece of vocabulary, used throughout this page and nowhere on the rest of the site. RPE — rating of perceived exertion — is just a number for how close a set came to failure. An RPE of 7 means you finished with roughly three reps left in you; an 8 means about two; a 9 means one; a 10 means the next rep was not happening. Nothing more technical than that.

The plan's phrasing is deliberate: a typical landing zone of RPE 7–8. It is where sets are expected to land, not a target you chase rep by rep.

Why a felt range instead of a percentage of your one-rep max? Three reasons, and none of them is that percentages are unscientific.

• A percentage needs a max, and this program never tests one. To prescribe 70% you have to know 100%, which means either a maximal attempt you do not otherwise need, or an estimate that drifts out of date within weeks — and across twenty-odd exercises, most of which have no meaningful max at all. What is 70% of a hanging windshield wiper?
• The right percentage is not the same on Tuesday as it was on Monday. Sleep, food, stress, the session before it. A fixed percentage says 100 kg on a day your body can handle 92 and on a day it could have handled 108. A felt range absorbs both without anyone editing a spreadsheet.
• The whole design runs on frequency, so the cost of overshooting is high. A set taken to failure on Monday is still being paid for on Wednesday, and Wednesday is another session. RPE 7–8 is roughly the intensity a movement can absorb two or three times in seven days at four to six sessions a week.

It is worth doing the arithmetic here, because “one set” sounds thinner than it is. The volume guidance people quote is stated per muscle per week, not per exercise per session — and across a week at six days this program puts roughly twelve sets into the back, nine each into quads, shoulders and the posterior chain, and six into the chest. Arms look thin at three direct sets each until you count properly: biceps are working in every row, pull-up and pulldown, and triceps in every bench and overhead press, which puts both nearer fifteen working sets a week. That sits inside the usual range rather than at the bottom of it. Frequency is what buys it: each pattern comes round two to three times in seven days, so the sets accumulate across the week instead of being crammed into one punishing session.

For an intermediate — someone who already knows what a hard set feels like — one hard set per exercise at this frequency is enough to keep progressing. The honest caveats are narrower than the headline suggests, and they are set out at the bottom of this page.

Two sessions up, three sessions down

The two rules look symmetric. They are not, on purpose.

The asymmetry follows from what each signal actually means. Below 7 is outside the zone. It is not the comfortable end of correct, it is an under-dose, and an under-dose costs you the entire session's stimulus for that lift — you did the work, you paid the time, and nothing was asked of the tissue. That is worth correcting fast. Two sessions is exactly enough evidence to rule out a good night's sleep and a light previous day.

A hard 8 is the top of the zone, not an error. A single one is the system working correctly; it is what a well-chosen load feels like on an ordinary day. Only persistence means the load has genuinely drifted above where it should be rather than the week having been heavy, so it takes three before anything moves. Reacting to one hard session would have you backing off from loads you can handle, every time you sleep badly.

Being at 8 is allowed. Living at 8 is not.

Put another way: the two thresholds are not measuring the same quantity. Going up is a question about whether the dose was too small, and a small dose has an obvious, cheap fix. Coming down is a question about whether you are accumulating fatigue faster than you can clear it, and that question needs more than one data point before you spend real load on the answer.

One consequence worth stating: a streak is evidence about this load. Change the weight and the evidence is void, so streaks never carry across a jump — and after a long enough layoff they go stale on their own rather than being treated as current.

Why the log is the mechanism

Read the rules above again and notice what they all have in common. Every one of them is a comparison between this session and the last: two sessions under 7, three at a hard 8, warmup weights creeping, aborts clustering, cuts piling up, the same elbow mentioned twice. Not one of them can be evaluated from inside a single session.

Now multiply. Roughly two dozen exercises, four to six sessions a week, each with its own independent streak counter running on its own timeline because each lift comes round every two or three days. Nobody holds that in their head. What people actually do is remember the two lifts they care about and guess at the rest, which is how a program like this quietly becomes "whatever felt right today".

The log is what makes the autoregulation real rather than a vibe.

So the app keeps the two grids the plan asks for — one for A, one for B, consecutive sessions of the same workout side by side, which is exactly the shape the rules need to read — plus two footer rows per session: whether you cut, and anything that felt off. The note line is where forearm and elbow niggles get caught before they become a pattern, and two notes about the same elbow in a fortnight is a signal to act on rather than train through.

One tap is the entire input

A mark is the only judgement you make. On paper it is a single character written only when something deviates. On screen the paper polarity is deliberately not reproduced: - means the set was easy and drives a load increase, which is the opposite of what a + reads as on a screen. So the control is a thermometer instead — under target, in the zone, over target, aborted.

IN ZONE takes about 44% of the width, because the honest answer is usually "nothing to report". But a deviation still costs exactly one tap — the honest answer must never be more expensive than the convenient one.

The abort rule sits outside the load rules. If a set reaches RPE 9 mid-set you stop the set, and usually cut the session short too. A STOP counts toward the same streak a hard 8 does, but on its own it does not change the load; it holds. What it changes is the block: one mid-set abort is a bad day, two in a fortnight is a statement about the block, not about those two days.

Not every lift is judged this way, and the exceptions are honest ones. Power cleans run on bar speed, because a slow, ugly clean can feel easy — the sensor there is whether set three moved like set one. Pull-ups have their own absolute rule from the plan (add load at 15+ bodyweight reps, assist under 8). Push-ups progress on rep count, because there is no weight to add. And naturally loaded movements — windshield wipers, ab wheel, step-ups, single-leg calf raises — log a leverage description instead of a number, because leverage is their progression variable.

That is roughly a dozen numbers and usually zero marks. Under a minute. If it takes longer than that it will not survive, and the grid should shrink before the plan does.

Why the sessions are short: the order does the resting

About fifteen minutes to the cut line, about thirty for the whole session. Those are the author's own measured times running this program — one experienced lifter in a familiar gym, which is a real data point and not a study. Learning the lifts, or sharing a rack at a busy hour, will cost you more.

The reason it is that quick is not that anything gets rushed. It is the exercise order. After the opening pair, no two consecutive exercises in either workout train the same thing. The opening pair is the deliberate exception: five front squats sit directly before the back squat, or directly after the clean, to prime the pattern and set an upright spine for the heavier lift that follows. The cost is that the back squat runs on legs that have already worked, so it lands at a lower weight than it would cold. That is an accepted trade, not an oversight: the number on the bar is not what this program is chasing. A body that works is, and legs that did five front squats and then twelve back squats have done more work than legs that only did the twelve.

The consequence is that the muscle you have just finished using is recovering while you do the next item. You are not standing still waiting for a timer; you are doing rows while your chest recovers, and pressing again while your back recovers. Rest becomes something you take when you want it rather than something the program schedules, and no individual set has to be hurried to keep the session short.

That ordering is the whole explanation for the fifteen and thirty minute figures. One working set per exercise sets the floor; the alternation is what stops the gaps between them from filling with dead time. With thirteen items and no clock between them, the limiting factor is how fast you can walk between stations and load a bar.

It also makes the two figures two different offers rather than one number with a discount. Fifteen minutes gets you everything above the cut line — the axial lift, both patterns of the day, the single-leg work, the loaded core — which the plan calls a complete session in its own right. Thirty gets you those plus the arms, calves, hamstring curls and prehab. Six of the long version is three hours a week; four of the short version is one. Both are less gym time than a conventional four-day split of hour-plus sessions, which is the arithmetic worth doing before deciding that training most days is a lot.

The exception proves the rule. Power clean day is three sets of three with two minutes between them, and the reason is specific: short rest turns a power lift into conditioning. That is the one place in the week where you stand there and wait, and it is the one place where waiting is the point. Two minutes, not five: the load is nowhere near a max, so it is four minutes of waiting across the whole session and clean day still finishes inside the same half hour as every other one.

Two honest caveats. Grip is the one thing that does get asked twice in a row, since pull-ups and the suitcase deadlift sit next to each other — if your hand gives out before your obliques do, put straps on the suitcase deadlift. The trunk is the thing that lift is training, and your grip already gets worked everywhere else in the week. And thirty minutes is time under the bar; a heavy squat still needs its warmup sets before the one working set counts.

The cut line

Exercises 1–8 are the session, and they take about fifteen minutes. Both workouts open with one set of five front squats — the one exception is power clean day, where the clean goes first and the front squat follows it, because cleans are graded on bar speed and leg fatigue takes speed before it takes anything else. Both workouts are ordered so that by the cut line the day's whole argument is done — the axial lift, both patterns of the day, the unilateral work and the loaded core. Stopping there is a complete session, not an abandoned one.

That is designed flexibility, not a failure state. A program you can only do all of is a program you stop doing, because life produces days with fifteen minutes in them. Building the short version into the plan means the answer to a bad day is "do the the ones above the cut line" rather than "skip it and feel bad", and they are chosen so that answer is defensible rather than a consolation prize.

It is membership, not position. You can reorder freely when the squat rack is busy and the cut line does not drift, because it is attached to the seven core items rather than to whatever you happen to do seventh.

Every use is logged, and that is the part people skip past. Using the cut line is an exception — off days, time pressure — and how often you take it is a diagnostic rather than a failure tally. More than two cuts a fortnight means something structural: the session is too long, recovery is short, or life has changed shape. The fix might be trimming an exercise permanently rather than truncating ad hoc, and if the cuts cluster on one workout, that is the one to shorten.

The reason the count matters is specific. Cutting always drops the same things: the arm work, the calf work, the hamstring curls and the prehab. Cut habitually and those are no longer in your program, while the tables still claim they are. A cut line you can use without recording it is just a program that shrinks without telling you.

Deloads

Autoregulating each exercise independently means nothing is watching cumulative load. Every lift can read 7 while weekly tonnage outruns recovery — that shows up as stalled progression rather than as a bad RPE, which is precisely the failure mode a per-exercise rule cannot see. Deloads are the layer that watches the whole block.

Three paths can call one:

• Aborts. Two mid-set aborts inside a fortnight. One is a bad day; two is a statement about the block. That rule is written into the program itself.
• Several lifts backing off at once. One exercise hitting three hard sessions in a row is a load problem — dial that lift back and leave everything else alone. Three different lifts doing it inside a fortnight is not a load problem, it is cumulative fatigue. Also written into the program, quantified at three because it says "several" and something has to fire.
• A scheduled cadence, counted in sessions. Thirty sessions. This one is our default rather than part of the program as written — it describes block-level pullbacks but never names a cadence, so the app picks one and lets you change it. It is counted in sessions rather than weeks on purpose: thirty sessions is about six loading weeks if you train six days, and closer to nine if you train four, so a lighter week never burns deload credit on a program whose entire logic is session-to-session.

Fatigue signals outrank the calendar. If aborts or a block problem are firing, that is the reason you are shown, not "thirty sessions elapsed".

A deload is a light week at about 80% across the board, with prehab kept in and the cut line off the table. Each of those three clauses is doing work. Eighty percent keeps the movement patterns and the schedule intact while dropping the load that caused the problem. Prehab stays because it was never the fatiguing part and it is the thing you least want to interrupt. And the cut line comes off the table because a light week is already the short version — cutting a deload leaves you with almost nothing, and the one week you are most tempted to skip is the one that fixes things.

Marks are still logged during a light week, but they touch no streak. An 80% set is meant to feel easy, and feeding those readings into the rules would end every deload by proposing increases across twenty lifts.

Coming out, the rule is restore, don't creep. The pre-deload weight comes back exactly as it was, and autoregulation decides from there with recovered tissue. The deload treated fatigue, not load selection.

One signal is worth watching yourself, because it appears before any single set misbehaves: warmup weights starting to feel like working weights. That is systemic. Soreness lags and lies; bar speed on the first rep of the working set, session over session, is the better read.

Reps and load are one ladder

The written program fixes reps — usually 12 — and floats the load. That works when the load is finely adjustable, and it breaks the moment it is not, which is most of a real gym.

The leg press problem. Take a leg press that steps in 20 kg. You have been doing twelve reps at 100 kg, it has felt below 7 twice running, and the rule says add load. There is no load to add. 100 is easy; 120 is a 20% jump and there is no version of you that gets twelve reps of it at RPE 7–8. The machine simply has no weight that satisfies the prescription. Follow the written rule literally and you either stall at a weight that has stopped doing anything, or you take a jump that turns the set into a grind and files a hard 8 that is not really a hard 8.

It is not an exotic case. Dumbbell racks jump 10% or more at the light end. Cable stacks step 5 kg on a 40 kg setting. Anything that is not a loaded barbell has this problem somewhere in its range.

The resolution is the one every lifter reaches on paper anyway. Reps are allowed to float in a band — 8 to 16 by default. When no jump is available, add reps where you are. When the jump is too big, take it at fewer reps and climb the reps back. Load and reps are two axes of a single ordered ladder, and "one increment" means the next rung, whichever axis it happens to lie on.

1. 100 × 12

   The written prescription. Twelve reps at RPE 7–8, and for a while this is simply the answer.

2. 100 × 14

   Twelve has gone easy twice. There is no load between 100 and 120, so the rung that moves is the rep count.

3. 100 × 16

   Easy again. Reps climb once more — and 16 is the top of the default band, so this is the last rung on this axis.

4. 120 × 10

   Now the 20 kg jump is takeable, because it is taken at fewer reps. Load moves, reps reset downwards to pay for it.

5. 120 × 12

   Reps climb back to the written target, and the lift is home. The next easy streak starts the staircase again.

The shape is always this: climb reps to the top of the band, spend them on the jump, climb back. The exact rung the app proposes comes from your own numbers — rungs are ordered by estimated 1RM, which is not a claim about your true max, just a monotone way to compare "twelve at 100" with "ten at 120".

Two things keep this from turning into permanent rep drift. Rungs are ranked first by how faithfully they hit the target intensity, and then — among options that are near enough equivalent — by how close the rep count sits to the exercise's written target. That second key is what pulls you back toward 12 rather than letting reps climb forever, and it is why a big weight jump is taken as soon as it becomes takeable within the band. A rep count that differs from the written target always comes with a line explaining why, because otherwise it reads as a bug.

On a well-stocked barbell, none of this ever happens. With 1.25 kg plates the next rung is 1–2% away, so it is always a weight change and the reps never leave their written target. Squats, bench, rows, overhead press and deadlifts all behave this way. The band exists to absorb coarse equipment — the leg press that steps 20 kg, the dumbbell rack that jumps from 20 to 22 — not to reinterpret the program.

The band is also absent wherever the rep count is part of the protocol rather than a free variable. Power cleans are triples because they are cleans.

Why there is no gym setup step

The obvious way to make the weight suggestions land on real numbers is to ask what your gym has: bar weight, plate denominations, the dumbbell rack, the step on every machine stack. It is the wrong trade. That is a wall of questions before you have lifted anything, it goes stale the moment you travel, and most of it is guesswork dressed up as configuration — very few people know the step on the lat pulldown they have used for two years.

The app does not actually need to know your gym. It needs two things, and it can get both for free:

• Which direction to move. That comes from the marks, which is exactly what the program specifies, and all it specifies.
• Roughly how big a step is. Every weight you have ever logged is evidence. If your dumbbells go 20 → 22 → 24, the app sees 2. If your leg press steps 20, it sees 20. Nobody has to type it in.

The observed step is simply the smallest real gap between distinct loads you have logged on that exercise. Before there is any history, a magnitude-based default gets close enough to suggest something sane — nothing moves a 140 kg squat by half a kilo, and a 15 kg lift moves in 1s.

The honest cost. On a brand-new lift the first suggestion is a guess, so the app can propose a weight your gym cannot actually make, and you fix it by typing the weight you really used. It cannot know about the 1.25 kg plates you own but have never used. And because the step is inferred from the smallest gap you have ever logged, one mistyped weight can convince it your gym is finer-grained than it is until more history washes that out. Every one of those failure modes is repaired by the same action — overwriting the number — which is the reason the trade is worth making. You are never blocked, you are only ever mildly wrong, and being mildly wrong for one session costs less than a setup form you fill in badly and then forget you filled in.

What the app decides, and what you decide

It is worth being precise about this, because "the app picks your weights" is both roughly true and misleading.

The direction is the only thing the app is confident about, because the direction is the only thing the program actually specifies. Everything downstream is a proposal, and the proposal loses every argument with the lifter. Overwriting a suggested weight is not fighting the app — it is how the app finds out what your gym can make.

What this method does badly

The landing page carries the full argument. In short, the same design decisions that make this method work are the ones you should be suspicious of:

• Four days a week is a different program from six. The weekly set counts above assume six sessions. At four they drop by about a third, which is fine for steady intermediate progress and more than fine for maintenance, but genuinely light if maximum hypertrophy is the goal. Stopping a rep or two short of failure also leaves something on the table by design — the plan raises this itself as its open question, and its written answers are to push the accessories to RPE 8–9, or to run two sets on positions 1–3.
• Everything depends on a judgement people are bad at. Estimating reps in reserve is a learned skill, and accuracy is worst far from failure, at high reps, and on lower-body lifts — which describes a set of twelve squats at RPE 7. Two-up/three-down filters noise, not bias.
• Nothing forces progress. Every rule fires on a set falling outside the zone. A set that lands inside it moves nothing, so a cautious rater can idle for months without the app ever objecting.
• Session position is unmodelled. The rules assume this session and the last are comparable. Reordering exercises, or cutting one week and not the next, changes the fatigue a lift is performed under and therefore changes the mark at the same load.
• Cumulative load is only watched crudely. Deloads are the whole of it, and two of the three triggers require something to have already gone wrong.

None of that makes the method wrong. It makes it a set of trades — an intermediate lifter's trades, specifically, chosen by someone who would rather train four to six short days than three long ones and would rather under-reach than under-recover.

The full case against the program →