A back-off set is not a percentage. Most of the guides that rank for this topic will hand you a number: take ten to twenty percent off the top set and do three more sets. That number is a starting guess dressed up as a rule. The real instruction is quieter. The top set already told you what the back-offs should be. You just have to read it.
Here is the frame. The top set is a measurement. The back-off sets are the dose you derive from the measurement. Two lifters can open with the same 3 @ RPE 8 and walk away needing completely different back-off work, because the number on the bar was the same and the reading behind it was not.
What the top set measures
When you finish a heavy top set, three things are on the table, and none of them is the load.
The first is proximity. Not the RPE you planned, the RPE you actually hit. You wrote 3 @ RPE 8 in the program. The bar said RPE 9. That gap is information. It means today's readiness is under where the program assumed it would be, and every set after this one inherits that gap.
The second is bar speed. The last rep of the top set has a velocity, and you felt it whether or not you measured it. A grinding last rep and a snappy last rep at the same RPE are two different days. Bar speed is the cheapest autoregulation signal you own, and the top set is where you take the reading.
The third is rep quality. Did the bar path hold? Did the third rep look like the first, or did your hips shoot up and your chest cave? A top set that held its shape earns more volume behind it. A top set that fell apart on the last rep has already told you the back-offs are going to be ugly.
The load is the one thing on that list that the program handed you. Everything worth knowing, you learned by lifting it.
Turning the reading into a dose
Now you set the back-offs. Not from a percentage table. From what you just read.
Start with the honest RPE. If the top set came in on target, at the RPE you programmed, the back-offs behave. Drop the load enough to buy the reps you want at a manageable cost, and run the volume the block calls for. On a strength-leaning day that might be two to three sets across, two RPE below the top set. The top set was the intensity. The back-offs are the tonnage that makes the intensity stick.
If the top set came in heavy, over the programmed RPE, the reading changed and the dose changes with it. This is the case the percentage rule gets most wrong. A fixed minus-fifteen-percent applied to a top set that already cost you RPE 9 is still going to bury you, because the percentage never saw how much the top set took. When the top set runs hot, I take more off the bar and I cut a back-off set, not add one. The goal of the back-offs on a hot day is to hold quality reps at a fatigue you can recover from, not to chase a tonnage target the day did not earn.
If the top set came in light, under the programmed RPE, you have room. The bar moved fast, the shape held, the RPE undershot. This is the day to add a set or keep the load closer to the top set. The reading gave you headroom. Spend it.
A worked example
Take a lifter running a squat day built on one top set of 3 @ RPE 8, then back-offs.
Monday, the top set comes in exactly at RPE 8. Last rep moved with control, bar path clean. The dose is easy: 3x5 at around RPE 7, roughly ten percent off the top set, and the day does what the program intended.
Two weeks later, same prescription, the top triple grinds to RPE 9. The last rep stalled halfway and the lifter fought it up. The percentage table still says three by five, minus ten percent. The reading says otherwise. I cut it to 2x4, take fifteen percent off instead of ten, and tell the lifter to end each set with two clean reps in reserve. Same slot in the program. Different day, different dose.
A third session, the triple flies up at RPE 7 when RPE 8 was on the sheet. The lifter has more in the tank than the program assumed. Now the back-offs earn an extra set, 4x5, held tight to the same load, because the reading opened the door.
Three identical prescriptions. Three different back-off schemes. The variable was never the percentage. It was what the top set said on the way down.
Why this beats a fixed rule
A fixed percentage is a program written for a lifter who does not exist, the one whose readiness never moves. Autoregulation is the correction. The top set is the cleanest autoregulation checkpoint you get in a session, because it is the heaviest honest effort before fatigue muddies the signal. You spend it to learn something, then you spend the back-offs acting on what you learned.
There is a ceiling on this. The reading can tell you to add a set, but the block's volume target is still the wall. If your back-offs are creeping up week after week because every top set feels light, that is not headroom. That is a program that under-shot its intensity. Fix the top set, not the back-offs.
This is not permission to freelance. The block still sets the shape. The rep range, the movement, the rough volume target, the RPE cap all come from the program. What the reading controls is the fine adjustment inside those walls. You are titrating a dose, not rewriting the prescription.
The SRA curve is doing the accounting underneath all of this. Every back-off set is fatigue you are choosing to buy against a recovery window you have to pay back before the next session on that lift. Let that debt run long enough across a block and the bill comes due as a deload. A back-off scheme set by a percentage buys the same fatigue every day regardless of what you can afford. A back-off scheme set by the reading buys only what the day can recover from. Over a block, that is the difference between volume that compounds and volume that just accumulates.
The habit
Log the reading, not just the load. Next to the top set, write what it cost: the true RPE, whether the bar slowed, whether the shape held. That one line is what tells you how to set the back-offs today, and it is what tells you in six weeks whether the top sets are trending the right direction or whether the back-offs have been quietly out-spending your recovery the whole time.
That habit is the reason Logbook has a notes field on every exercise and on the session itself, not just a number pad. Logbook is the iOS training notebook I built. One-time purchase. No social feed. No AI coach. When a top set costs more than the sheet said, you write it down next to the set, in the moment, and it is there next week when you program from it. That is the difference between a tracker that stores your numbers and a notebook that keeps your reasoning.
The top set is a question you ask the bar. The back-offs are what you do with the answer. Stop reading the answer off a percentage table and start reading it off the rep.