You had a bad bench day. The bar felt heavy in the warm-ups. You got six on a set that was supposed to be eight, then five, and drove home convinced you'd hit a wall.
So next week you "deload." You take twenty pounds off, do some easy sets, feel fine, and two weeks later you're back at the weight that beat you, which beats you again, because nothing about it changed.
The deload wasn't wrong as a tool. It was wrong as a diagnosis. You prescribed it for a plateau, and what you had was a Tuesday.
A plateau is a pattern, and patterns need more than one point
One session is a single measurement, and the thing it measures is noisy. Sleep, food, stress, the time of day and whether someone had the rack you wanted all move the number a few percent in either direction. Lifters who track for long enough see it plainly: perfectly healthy training blocks are full of sessions a rep or two worse than the one before.
That's why the oldest rule of thumb in linear progression doesn't trigger on a bad day. You miss your reps, you repeat the weight. You miss them again, you repeat it again. Only after the third session at the same load that doesn't move do you call it a stall, take something like ten percent off, and climb back through it. Three is a little arbitrary, but it's the right kind of arbitrary. It's the smallest number that lets a bad day stay a bad day.
The point isn't the number three. The point is that a plateau is a claim about a trend, and you can't read a trend from one point. You can only feel one.
Two things that look like a stall and aren't
Once you start judging across sessions instead of within one, two common mistakes fall apart.
Missing the top of the range while the reps are climbing. Say the plan is three sets of six to eight. Week one you get 6, 7, 7. Week two, 7, 7, 8. Week three, 8, 8, 7. You've "missed eight" three sessions running, and if your only question is did I hit the target, that reads like three failures. It's actually the exact thing double progression is supposed to look like: same load, total reps going 20, 22, 23. Deloading here punishes a lift for working.
Hitting the top of the range every single time. Three weeks of clean 3×8 at the same weight also produces a flat line, and someone squinting at a log can mistake flat for stuck. It's the opposite. Flat at the ceiling is the signal to add weight, not take it off.
So the real definition is narrower than "it didn't go up." It's this: same load, three sessions, total reps not improving, and the most recent session still short of the range. Reps stopped moving and you aren't at the top. That's a stall. Most of the other things people call stalls are just noise or progress they didn't recognise.
Why almost nobody can apply this
Read that definition again and look at what it needs: the load, the reps on every working set, and the totals, for three sessions, side by side. Most people have none of it. They have a feeling about last week, and a feeling about the week before that has already merged with it.
This is the same trap as trying to see your own technique change: the change you're trying to detect is smaller than the resolution of your memory. "Twenty, twenty-two, twenty-three" and "twenty-one, twenty-one, twenty-one" both feel like about the same, kind of stuck. One is a lift progressing on schedule and the other is a real stall, and from memory they're the same lift.
So people fall back on the two instruments they do have. One is feel: the session felt bad, so something must be wrong. The other is the calendar: it's week four, so deload. Feel can't tell a bad day from a pattern. The calendar is a legitimate choice, and plenty of good programs schedule deloads, but it's answering a different question. It manages fatigue on a timetable. It doesn't tell you whether this lift is stuck.
And the cost of guessing wrong isn't just one wasted week. It's what comes next. A lifter who deloads off a bad day and then hits the same wall concludes the program has stopped working, and goes looking for a new one. A perfectly good block gets thrown out because of an instrument that can't tell three numbers apart.
The honest, unglamorous fix
You don't need anything clever. You need the three sessions on a page.
Write down every working set, not just the top one. The top set is the most variable number in the session. The total across your working sets is much steadier, and it's where a stall actually shows up.
Before you call it, look back three sessions at the same load. If the totals are climbing, you're fine, even if you missed the top of the range. If you hit the top every time, add weight. Only if they're flat or falling, and you're still short of the range, is it a stall.
When it really is a stall, back off on purpose and climb again. Something around ten percent, then work back up through the same range. It's not a punishment and it's not a reset to zero. You're stepping back far enough to build momentum through the sticking point instead of grinding into it a fourth time.
Don't let one grinder vote. A rep you barely made at the very limit is useful information. It's still one rep on one day. It goes in the log with the rest.
Where we come in
We build Flexion, and the relevant part is the least exciting thing it does: it remembers every set you log and puts last time's numbers in front of you when you're about to lift. That's the whole requirement for telling a bad day from a plateau, because the judgment above is impossible when the three sessions only exist in your head.
A notebook does it too. If you already keep one and you actually open it before you load the bar, you have everything this post asks for. The tool matters much less than the habit of looking at the record instead of the feeling.
So the next time a session goes badly, don't rush to fix it. Write it down. Then wait for two more, and see what they say.
