RPE and percentage each capture half the lift, so a good log keeps both
The RPE-versus-percentage debate is about programming. Your log faces a different question: which entry can you still read six months from now?
The entry you can't read anymore
You open the log to plan tonight's squat and scroll back to the same week last cycle. The entry reads 315 x 5 @ RPE 8. You know exactly what you lifted. You have no idea whether that was a good day or a forty-minute fight. Were you fresh or three days into a bad sleep stretch? Did the bar speed hold across all five, or did the last two grind to a crawl? The number recorded the load and the rating. It recorded almost nothing about the lift. The percentage lifter has the mirror-image problem: an entry that reads 80% x 5 is precise about the plan and silent about the max that 80% was actually a slice of that morning.
The readable-later test
Here is the test that almost no one applies to the RPE-versus-percentage argument: six months from now, can you read this entry and reconstruct what happened? Call it the readable-later test. As far as your log is concerned, RPE and percentage are not rival programming philosophies. They are two recording formats, and each one drops a different half of the story. A percentage answers what you were supposed to do. An RPE answers what it cost you. Neither answers the other. The block you can actually learn from later is the one whose log captures both, plus the variable moving underneath them both.
Why the usual argument misses the log entirely
The internet argues which method gets you stronger. That is a programming question, and the honest answer is that both work for the lifter who runs them consistently. But the debate treats the log as a passive container that just holds whatever the program prescribed. That is where it goes wrong. Choosing RPE over percentages, or the reverse, is not just choosing how to assign load. It is choosing which half of each session you write down by default and which half you have to remember to capture manually, by week three of a long block, when you have stopped writing notes. The method you pick quietly decides what your training history will and won't be able to tell you a year from now.
What the research actually measures (and what it doesn't)
The studies people cite in this argument measure the wrong axis for our purposes. Controlled comparisons of RPE-based loading against fixed percentage loading have reported a small advantage for autoregulated RPE on 1RM outcomes RPE-based loading may provide a small 1RM strength advantage over percentage-based loading in a majority of individuals. ([source](https://pmc.ncbi.nlm.nih.gov/articles/PMC5877330/)), and the coaching orbit around Stronger By Science has worked this ground for years. Useful work. But every one of those trials is measuring which method produced more strength, not which method left behind a more readable record. Those are different questions. A method can edge out the other in a twelve-week trial and still hand you a log you cannot interpret when you go back to plan your next peak. Progress over a block lives in the record, not in the prescription that generated it.
Where each method lies, walked through a block
Run a percentage block first, 5/3/1 style. You program off a training max and log 85% x 3 as 290 x 3. Six months later that 290 is fossilized against a training max you have since raised twice, and without the TM and a note on how the bar moved, you cannot tell whether it was a comfortable triple or a near-miss. Percentage lies by hiding readiness: it assumes the max that was true on the spreadsheet, not the one that was true that morning. Now run an RPE block. You work up to a top single at RPE 8 and log 315 @ 8. RPE lies by drift: an 8 in week one and an 8 in week five of a fatiguing block are not the same bar speed, because your internal scale recalibrates as fatigue accumulates. Read the rating alone across the block and the stall hides inside a flat column of eights.
What a log has to keep for either method to be readable
For either method to survive the readable-later test, the log needs three things, not one. The prescription, whether that is the percentage and its training max or the RPE target. The actual load moved. And at least one objective anchor the rating can be checked against later: reps left in the tank, a note on last-set bar speed, a sleep or bodyweight flag, something the number alone can't fake. The honest version of 80% x 5 is 80% of a 365 TM, 290 x 5, last rep moved well. The honest version of 315 @ 8 is 315 x 5 @ 8, week four, speed held. Captured that way, percentage stops hiding readiness and RPE stops drifting silently, because the anchor catches both. The method becomes a detail. The record becomes the thing you actually own.
An instrument that keeps both halves
This is the part most logs leave to a free-text note you abandon by week three. Platepusher records the prescription, the load, and the RPE as separate native fields, and surfaces them side by side across the block so the drift is visible instead of buried. When your load at a fixed RPE 8 stops climbing while the rating holds flat, the chart shows the stall before the third stalled week makes it obvious. The math runs on what you logged. It does not tell you whether to autoregulate or run percentages; that is still your call, and your coach's. And if you are carrying a multi-year history, you bring it in from a CSV and read it back in the same shape, so the record you already built stays legible. Recognition, not motivation. The instrument hands back what your own training has been recording.
What we're watching next
The open question is the objective anchor itself. RPE is readable when it sits next to bar speed, but few lifters want a velocity device on the bar for every working set. The interesting thread is whether wrist-worn capture can stand in as a good-enough proxy for the speed an RPE 8 should hold, so the anchor rides along with the session instead of being one more thing to log by hand. Not solved. Worth tracking, because the method debate gets a lot less interesting once the record reads itself.
Log the prescription, the load, and the RPE as separate fields, then watch the drift across the block. Bring your history in from a CSV and read your own training back.
Built for lifters with multi-year logs who autoregulate off RPE, run percentage blocks, or move between the two. The point isn't picking a side; it's keeping a record you can still read when the cycle is long over and you're planning the next one.
