How SteelRep Recommends a Deload

The deload most apps get wrong

Most workout apps handle deloads in one of two ways. Either they ignore them entirely — leaving you to notice the stall yourself, weeks too late — or they bolt a deload onto a fixed calendar: every fourth week, drop the weight, regardless of whether you actually need it.

Both are blunt instruments. A calendar deload tells you to back off when you might be progressing fine, and it stays silent when you have been grinding the same failed weight for a fortnight.

SteelRep takes a different approach: it watches your actual performance, one lift at a time, and recommends a deload the moment the numbers say you have stalled. No guesswork, no calendar, no black box. This article walks through exactly how that decision is made — because the logic is simple, deterministic, and worth understanding.

If you want the broader picture of what a deload is and why it matters for long-term strength, start with the deload week guide. This piece is about the mechanism: how SteelRep decides.

The signal: consecutive failed sessions, per lift

SteelRep does not track “how you feel.” It tracks one thing precisely: consecutive failed sessions at a working weight, for each lift independently.

Two details matter there.

What counts as a failed session. A session fails for a lift only when none of your working sets at the current weight hits the target rep count. Miss your last set by a rep after nailing the first four? That is not a failure — you hit the target on the sets that count. A failed session is a genuine miss: you loaded the bar and could not produce the prescribed reps on any working set.

Per lift, not per program. Your squat and your bench press are tracked separately. If your bench stalls while your squat keeps climbing, SteelRep deloads the bench and leaves the squat alone. There is no blanket “deload week” forced across every lift just because one of them is struggling.

The counter is honest in both directions. Fail a session, it ticks up. Hit your target on the next one, it resets to zero. You have to actually stall — repeatedly, at the same weight — for a deload to trigger.

The trigger: how many misses it takes

By default, SteelRep recommends a deload after three consecutive failed sessions on a standard linear-progression program (the 5x5 Power Builder and Linear Barbell families). Three genuine misses at the same weight is the clearest possible signal that linear progress has run out — one bad day is noise, three is a pattern.

Two program types trip the wire earlier, at two consecutive failures:

• Percentage / periodized programs like Periodized Strength Cycles (the 5/3/1 family). These run closer to your true capacity, so a stall here means more, sooner.
• Joint-Friendly Strength 50+, which is tuned for older lifters who benefit from backing off before fatigue compounds.

That is the whole trigger. Consecutive misses, counted per lift, compared against a threshold. Nothing hidden.

What happens when it fires

When the threshold is hit, SteelRep cuts the load by 10% — dropping to 90% of where you were — and rounds to the nearest 2.5 kg so the number is loadable. A 100 kg squat becomes 90 kg; a 105 kg squat becomes 95 kg. You then climb back up from there, and the extra runway almost always carries you past the weight you were stuck on.

Where that cut lands depends on the program type — and this is the part most apps cannot do, because they do not model the difference:

• Linear programs deload the working weight itself. Next session pre-fills 10% lighter, and you resume linear progression from there.
• Percentage / 5/3/1 programs deload the Training Max instead. Because every set in these programs is calculated as a percentage of your Training Max (85%, 90%, 95%…), dropping the TM rebuilds the entire weekly staircase lighter in one move — the whole program rescales together, not just one set.

Then the counter resets and the lift is flagged as recovering, so you can see at a glance which lifts are rebuilding.

How you recover from a deload

Recovery is as deterministic as the trigger. One successful session clears it. Hit your target reps at the deloaded weight and SteelRep marks the lift as advancing again, clears the recovering flag, and resumes normal progression. There is no penalty box, no mandatory number of weeks at reduced load — the moment you prove you can do the work, you are back on track.

Auto-deload vs the programmed deload week

It is worth separating two things that both use the word “deload.”

• The auto-deload described above is reactive — triggered by your stalls, on the lift that stalled.
• A programmed deload week is planned into a program’s structure. In the 5/3/1-family cycle, for example, week four is a built-in deload at 60% of Training Max — scheduled recovery to clear accumulated fatigue before the next cycle starts heavier, whether or not you stalled.

SteelRep runs both. The programmed week handles the predictable fatigue of a hard cycle; the auto-deload catches the unpredictable stall. You get the calendar’s discipline and a response to what actually happens on the bar.

It is your call, not a black box

The defaults above are sensible, but they are defaults. You can tune the deload to your own tolerance:

• How much to drop — anywhere from 50% to 95% of your current load.
• How many misses trigger it — from 1 (cautious) to 5 (stubborn).
• Off entirely — if you would rather manage stalls by hand.

Because the rule is just “count misses, compare to a threshold, cut by a percentage,” there is nothing opaque to override. You can see why every recommendation happened.

Why stall-detection beats a calendar

Here is the philosophy behind the design. A deload is only productive when it lands at the right moment — late enough that you have earned the fatigue, early enough that you have not spent three weeks grinding a weight you cannot move. A fixed calendar cannot know that moment, because it cannot see your sets. It deloads the lifter who is flying and ignores the one who is buried.

Counting your actual misses, per lift, is the closest a no-nonsense algorithm can get to what a good coach does: watch the bar, notice when it stops moving, and pull back exactly then. (When the deloads start stacking up faster than you can climb out of them, that is a different signal entirely — it means you have outgrown linear progression. See what to do after 5x5 and the 3-way program comparison for where to go next.)

And all of it runs on your device, with no account and no backend. Your training history never leaves your phone; the deload logic is computed locally from your own logged sets. It is deterministic and the same every time — the antithesis of an AI that guesses.

Train with the brakes built in

Progressive overload only works if something catches you when the overload outruns your recovery. SteelRep is that something — it watches every lift, recommends the deload exactly when the numbers call for it, and gets out of the way the moment you recover.

Pick a program, log your sets honestly, and let the algorithm handle the stalls. Pick up the bar.