What a good form cue actually looks like
11 August 2026 · form review, coaching
A good form cue names three things: the rep, the joint, and the second. "Right knee drifted in on rep 6" is a cue. "Watch your form" is not. The first one tells a trainee where to look and what to change. The second one tells them only that someone was unhappy, and leaves them guessing.
The three parts
The rep. A set is not one thing repeated. Rep 1 and rep 8 are often different movements wearing the same name, because rep 8 is the one where attention runs out. Naming the rep turns "sometimes your knee goes in" into "here is where it went in". The trainee can scrub back and watch it happen.
The joint. Name the part that moved and the direction it moved. Knee in. Elbow flaring out. Ribs lifting at the top. "Core tight" names no joint, so it gets filed as general disapproval and changes nothing on the next set.
The second. Video has a timeline. A cue can sit at 0:41 instead of floating over the whole clip. Once it sits at 0:41, the trainee is not searching their memory of the set. They are watching the thing being described, at the moment it is being described.
Before and after
| Feedback | The same note, specific |
|---|---|
| Watch your form | Right knee drifted in on rep 6 |
| Core tight | Ribs lift at the top of every rep from rep 4 on |
| Go deeper | Rep 5 onward stops a couple of inches higher than rep 1 |
| Sloppy last rep | On rep 8 at 0:41, the bar travels forward over your toes |
| Nice work | Rep 2 and rep 3 look identical. That is the pair to keep |
The right column is not longer because it is more critical. It is longer because it carries the information the left column left out.
Specific is the kind version
There is a worry that naming a rep and a joint sounds cold, and that softer wording is warmer. It runs the other way. "Watch your form" is a verdict on the person. "Right knee drifted in on rep 6" is an observation about one rep, and it carries the quiet implication that the other reps were fine.
The same applies to praise. "Nice work" is pleasant and gone by lunchtime. "Rep 2 and rep 3 look identical, that is the pattern to keep" tells a trainee what to reproduce.
What a cue is not
It is not a score. A number out of ten summarises a set and explains nothing about it.
It is not a list of everything. Three cues on a video is a review. Twelve is a wall, and the trainee remembers none of them.
It is not a command. "The hinge looks cleaner when the ribs stay down" leaves room for a trainee who knows something about their own body that the coach does not. "You must brace harder" does not.
Why FormLab is shaped this way
FormLab is coaching software built around form video review, and the cue is the unit it is built on. A trainee films a set and sends the clip. On the coach side the video opens with a scrubbable timeline. The coach pauses at the moment that matters and writes the cue there, anchored to a timestamp in the video. The trainee sees the cues listed against the clip and taps one to jump to that second.
That is a small design decision with a large consequence. When the interface asks you to place a note at a point in time, "watch your form" has nowhere to go. It does not belong at 0:41 any more than it belongs at 0:12. The software does not grade the coach, and it will not write the cue for them. It just makes the vague version feel out of place, which is most of the work.
Every cue in FormLab is written by the coach, not by a model watching the video. That is a deliberate position, and it has its own post: why FormLab has no AI form checker.
FormLab is in open alpha and free to use while it is. If you want the shorter version of what it does, the FAQ has it.