How to Track Personal Bests in the Gym (Without Guessing) 2026

A personal best is a comparison between one set and every previous set you have done on that exercise. That definition is doing more work than it looks. A PR only exists if the comparison actually runs, and most of the time it doesn't. You can know your all-time best bench to the half-kilo and still have no idea whether last Tuesday's 5 × 5 was a record, because nobody checked at the time.

This guide covers the three kinds of record worth keeping, how to compare sets that look nothing alike, and how to get the checking to happen without you.

What counts as a personal best

Three kinds of record answer different questions.

Weight at reps is your best load for a given rep count: the heaviest single, the heaviest triple, the heaviest set of eight. Each rep count keeps its own record, so 100 kg × 1 and 80 kg × 8 are two separate PRs on the same lift. This is what most people mean by a PR.

Estimated one-rep max is a single strength figure computed from any hard set, which lets a week of fives and a week of eights sit on the same line. It answers "am I stronger than last month" without a max attempt.

Session volume is weight × reps summed across every set of an exercise in one session. Volume PRs record work done, and they keep moving through the weeks when the top set doesn't budge. That makes them the most reliable sign a block is going somewhere.

Two rules keep all three honest. The exercise has to match: a paused bench and a touch-and-go bench are different lifts. And the conditions have to match within reason. A rep with a belt and a rep with a spotter's fingers under the bar are not the same rep, so decide once what your log counts and stick to it.

How 80 × 8, 90 × 4 and 100 × 1 compare

Suppose you benched 80 kg for 8 one week, 90 kg for 4 the next, and 100 kg for a single the week after. Which was the best set? Weight-at-reps says all three are records at their own rep counts, which is true and unhelpful. Estimated one-rep max gives one answer.

The common formula is Brzycki's:

Estimated 1RM = weight ÷ (1.0278 − 0.0278 × reps)

Working the three sets through it:

All three land within 2 kg of each other. In strength terms that is the same lifter on three different days, which is the whole use of the estimate: it stops you reading a change of rep scheme as a change in strength. The 80 × 8 set is a marginally better strength result than the 90 × 4 set, and six times the volume of the single. Run the same three weeks in reverse order and the log shows a lifter getting no stronger while doing much more work. Same sets, different story, and only one of the two readings is real.

The formula has limits. The NSCA's load chart, which maps rep counts to percentages of a max, shows the same shape: estimates are tight at 1 to 5 reps and loosen as reps climb, and past 10 or 12 individual differences in endurance swamp the arithmetic. Use it for direction, and for comparing sets in the 1 to 10 range.

Why the checking has to be automatic

Here is the failure mode. You hit 82.5 kg for 5 on Tuesday. It felt heavy, but fives usually do, so you racked it and left. The following week, scrolling back to set up the same lift, you notice the 82.5 and realise your previous best five was 80. That was a PR, a week ago, and nobody marked it.

Attention isn't the problem. A PR is a comparison against every earlier set on that exercise, and a human doing that between sets has to recall a number from three weeks ago while breathing hard. Paper can't compare; you flip back. A spreadsheet can, but only if you built the lookup, and only once the set is typed in, which mid-session it will not be. Evening logging from memory is worse: weights round up, reps round down, and next week's comparison runs against fiction.

What a missed PR costs you is the signal. If you don't know Tuesday was a record, you don't know the block is working, and a stall two weeks later reads as the program failing rather than a normal wave after a peak. Records are the earliest evidence training is doing something, and evidence that arrives a week late isn't much use.

So the requirement is specific: the comparison runs at the moment the set is saved, against the complete history for that exercise, with no action from you. Anything else works until the first busy week.

Setting up automatic PR detection

Most of this is discipline about inputs.

Start by keeping one history per exercise. Every set you want compared has to land under the same exercise name, so pick one catalog entry per movement and give a variation its own entry when it differs enough to matter. A tracker with a proper exercise library does this for you; a notes app does not.

Then log weight and reps for every set, during the session. Volume PRs need all the sets and weight-at-reps PRs need the honest rep count, and both only survive if logging costs seconds. Speaking the set works when the gym allows it: voice logging turns "bench, 82.5 for 5" into a structured set and shows the numbers for review before anything saves. When it is too loud to talk, the keypad prefills last session's weight and reps, so a repeat set is one tap and a heavier one is three.

The step that does the work is the comparison. In Strongbro, every saved set is checked against your history on that exercise, whether it arrived by voice or keypad. Two records update on their own: your best weight at that rep count, and your estimated one-rep max, computed with the Brzycki formula from your best set. When a set beats the history, the app says so by name and number at the moment it happens. Records carry the date of the set that earned them, so a session back-logged from yesterday slots into the timeline where it belongs.

Read the results as a timeline. One row per exercise per month, showing the heaviest lift and the day it happened, tells you whether this quarter beat the last one. A list of all-time bests can't do that; it only ever shows you the peak.

Last, fold it into a weekly review. Records are one of the four numbers in a full progress-tracking system, alongside volume, sessions and body weight. Once a week, ask which lifts set a record of any kind. One is enough. A week with a single rep PR on rows and nothing else is a good week.

If your tool can't run the comparison itself, the rest is at risk, because you become the comparison engine, and the first tired Tuesday you won't run it.

How often to expect personal bests

Expectations matter here, because a lifter who expects a PR every session will read an ordinary fortnight as failure. The pattern follows training age, meaning consistent logged training rather than years since joining a gym. The ACSM's position stand on progression notes that untrained lifters gain strength fastest and that the rate slows with experience. The practical shape is roughly this.

In the first six months, some record on a main lift most weeks. Weight-at-reps PRs come often, because technique is doing most of the work and the smallest plate is a small fraction of the bar. A novice training three times a week who goes three weeks without any PR should look at sleep and food before touching the program.

From six months to three years, one record per lift every two to six weeks, most of them rep PRs. This is where estimated one-rep max earns its keep, because 80 × 8 becoming 80 × 9 is a real PR that a weight-only log would miss. A month where squats stall and pull-ups climb is normal.

Beyond three years, months between records on any given lift. Estimated one-rep max is often the only number that moves, by a kilogram or two. Volume PRs keep the log alive here, and planned blocks with a deliberate peak replace the expectation of weekly gains.

Across all three stages, the strongest predictor of PR frequency is sessions per week. The NHS guidance sets the floor at strengthening work on at least two days a week, and that is a floor for health rather than a target for records; three a week held for months is what a PR schedule like the one above assumes. A tracker that counts training days for you, the way a rank ladder does, is measuring the thing that decides everything else.

What to do in a PR drought

Every lifter hits a stretch where nothing moves. The order of operations matters more than any new program.

Check the inputs first. Sessions in the last four weeks against your target, and volume against the month before. A drought that coincides with two skipped sessions a week, or with volume halving, is attendance rather than a plateau. The log usually admits this before the program looks guilty.

Then count the records you are ignoring. A weight-PR drought often hides a run of rep and volume PRs. If your best five has held at 82.5 for a month but you went from 82.5 × 5, 5, 4 to 82.5 × 5, 5, 5, the estimated one-rep max moved and the weight-only view can't see it. Give yourself credit.

Check the increment. If the next dumbbell up is a 5 kg jump, the step may simply be too large for where you are. Widen the rep range so the jump fits, add a set at the current weight, or use fractional plates. The progressive overload guide covers that loop in detail.

Then deload. If two or more lifts have stalled in the same fortnight with sleep, food and attendance unchanged, cut the load to around 80 percent for a week, keep the sets, and return to the previous weights. The first session back often produces the record you were chasing.

Change the program last, and only after four to six weeks of flat records with the inputs in order. Swapping programs at the first flat fortnight throws away adaptation you were two weeks from collecting.

Through all of it, keep logging every set. A drought with a complete log is data. A drought where you stopped logging is a hole, and records can't restart from a hole.

Get Strongbro on the App StoreFree for 3 days. $19.99 once for lifetime.

Get Strongbro on the App StoreFree for 3 days. $19.99 once for lifetime.