Progression

Should you track each arm separately?

You do single-arm dumbbell rows. Your log has one row for the exercise: three sets of ten at 30 kg.

But you did not do three sets. You did six, three per side, and they did not go the same way. Your right side got its ten without much trouble and your left ground out the last two. The log has no way to say that, so it says ten, and the difference disappears.

Which is a peculiar outcome, because noticing that kind of difference is one of the main reasons to do single-arm work in the first place.

Two chains, not one

The fix is to treat a unilateral exercise as two independent progressions that happen to share a name.

Each side carries its own baseline. Each side gets its own verdict. Your left row and your right row are tracked the way two different exercises would be, because in every way that matters to measurement, that is what they are.

This is not extra bookkeeping for its own sake. A single number for two limbs is an average, and an average of two things moving in different directions is a number that describes neither. If your right side is progressing and your left is stalling, the average may look flat, and flat reads as "nothing much happening" when in fact two quite different things are happening.

One verdict, and it is the worse one

Two chains still have to produce one answer for the exercise, because you want to know whether your rows are going well.

The rule that works is: the exercise's verdict is the worse of the two sides. Left regressed and right progressed means the exercise regressed. Left flat and right progressed means flat.

Taking the worse rather than the average or the better side is a deliberate choice and it is the conservative one. A strong side is fully capable of carrying an average upward while the weak side goes backwards for months, and by the time an averaged number finally turns, the gap has had a long time to widen. Taking the worse of the two means the moment either side has a problem, the exercise has a problem.

The cost is that a good session on your strong side does not get to shine if the other side had a bad day. That is a real cost and it is worth paying, because the reason you are logging sides separately is to catch the bad side early.

Switching between one-arm and two-arm work

Here is the case that quietly corrupts logs.

You have been doing barbell rows. You switch to single-arm dumbbell rows. Your tracker, seeing "rows", compares the new numbers to the old ones and reports something.

Whatever it reports is meaningless. A one-arm set and a two-arm set are different movements with different stability demands, different ranges, and a load that is not on a comparable scale. Comparing them is comparing two different things and calling the difference progress or regression.

Switching between unilateral and bilateral versions of a movement should break the comparison and start a fresh baseline, in exactly the way that switching from barbell to dumbbell does, or changing gyms and therefore machines does. A fresh baseline is not lost information. It is the refusal to generate false information.

What a normal imbalance looks like

Worth saying plainly, because splitting your log by side will show you an asymmetry and it is easy to over-read it.

Almost everyone is somewhat stronger on one side. Dominant hand, sport history, an old injury, or simply how you are built. A persistent gap between sides is not automatically a problem to solve, and chasing perfect symmetry is not a goal worth having.

What is worth watching is the trend. A gap that has been the same size for a year is a feature of you. A gap that is widening month over month is a change, and changes have causes. The second one is worth investigating and the first one mostly is not.

You cannot tell those apart without per-side history, which is the practical argument for the extra row.

Where not to bother

None of this applies to bilateral movements. A barbell bench press is one movement performed by one body, and there is no meaningful sense in which it has a left set and a right set. Splitting it would produce two identical chains and tell you nothing.

The line is whether the sides were genuinely performed independently. Single-arm rows, single-leg presses, split squats where each leg gets its own set, unilateral machine work: those are two exercises. Anything where both limbs move the same load at the same time is one.

The short version

For genuinely unilateral work, log the sides separately. Each side keeps its own history and gets its own verdict, and the exercise takes the worse of the two.

Treat a switch between one-arm and two-arm versions as starting over rather than as continuing a line, because the two are not measuring the same thing.

Then watch whether the gap between your sides is stable or growing. Stable is ordinary. Growing is the thing you split the log to find.

Common questions

Should I log left and right sides separately?

For genuinely unilateral exercises, yes. A single row forces you to record one number for two limbs that are not performing identically, and the side that is falling behind is exactly the information you lose.

How should progress be judged when the two sides disagree?

Take the worse of the two. If your left side regressed and your right progressed, the exercise regressed. Averaging them lets a strong side conceal a weak one, which is how imbalances go unnoticed.

Does switching between two-arm and one-arm work reset my progress?

It should. A one-arm set and a two-arm set are different movements, so comparing them produces a false verdict. Starting a fresh baseline is the honest response.

Is it worth the extra logging effort?

Only for genuinely unilateral work, where it is one extra row and it answers a question nothing else answers. There is no reason to split a barbell bench press into sides.

What if one side is always weaker?

Some asymmetry is completely normal, especially between dominant and non-dominant sides. What matters is the trend rather than the gap: a stable difference is ordinary, a widening one is worth acting on.