You did the thing everyone eventually tells you to do. You propped the phone against a water bottle, filmed a working set, and watched it back in the car. Your right knee dives in on the way up. Fine — that's a real thing, and now you know about it.
So you picked a cue. Spread the floor. You ran it for six weeks, honestly, on every squat session, and around week five you filmed again.
And then the moment nobody warns you about: you watch the new clip and you have absolutely no idea whether it's better.
It looks like a squat. The knee still moves in a bit. Is that less than before? It might be less than before. You genuinely cannot tell, and after ninety seconds of squinting you close the app and get on with your day, and the question just quietly stops being asked.
That is not a focus problem or a discipline problem. You built a before and then let it evaporate, and the advice that got you here — film yourself — stops exactly one step short of the step that makes it work.
Change happens below the resolution of memory
Here is the uncomfortable arithmetic of technique.
Memory of your own movement is coarse. You don't store a squat; you store a verdict — it looked bad, it looked okay, my back rounded. That's roughly three settings, and it's about as much fidelity as introspection gives you for anything physical.
Real form change, though, doesn't arrive in verdict-sized jumps. It arrives in the size of change that six weeks of a decent cue actually produces: the knee that used to collapse to the arch of your foot now collapses to the ball of it. Your torso is a few degrees more upright out of the hole. You lose the bar path a quarter of the way up instead of immediately.
Every one of those is a large change. All of them are smaller than the gap between it looked bad and it looked okay. So the instrument you're reading returns the same value it did in June.
Which means the lifter who is steadily improving and the lifter who is completely stuck get the identical reading from the only tool they're using. Not similar readings. The same one.
And the expensive part isn't the frustration
If this were only demoralising, it would be survivable. The real cost is what you do next.
At about week five, having concluded the cue did nothing, you go looking for a different cue. Then a different program, because clearly the program isn't working either. This is where a very ordinary, perfectly good block of training gets thrown out and replaced — not because it failed, but because you had no way to see that it hadn't.
You abandoned the thing that was working, on the evidence of an instrument with three settings.
Notice, too, that you would never accept this anywhere else in your training. Nobody says "I think I'm benching a bit more than I was in June." You wrote the number down, because everyone accepts without argument that a human cannot hold eight weeks of numbers in their head. Movement is a far harder thing to remember than a number — and it's the one variable almost everybody tracks entirely by feel.
How to actually build a before that survives
None of this is difficult. It's just unglamorous, which is why it doesn't get done.
Film from the same place. Not roughly the same place — the same place. Two clips shot from different heights are two different lifts, and any comparison between them is worthless in a way that isn't obvious while you're doing it. Pick a spot: side-on, camera at about hip height, far enough back to get the bar and your feet in frame. Then reuse it. Use a floor tile, a plate you leave in a particular spot, the third upright of the rack — something you can find again in six weeks without thinking about it.
Keep the load comparable. A grinding top single next to a smooth set of five will show you an enormous difference that is entirely about the weight and tells you nothing about your technique.
Keep the clip, and make it findable. This is the whole thing, and it's the boring part. Not "it's in my camera roll somewhere" — findable, in week six, when your camera roll has another two hundred photos in it. However you do it, the before has to still exist on the day you want to use it.
Compare one checkpoint, not "form." Is it better has no answer and just makes you feel bad. Is the bar further back over my midfoot at the bottom has an answer you can see in two seconds. Pick the checkpoint before you watch, not after.
Freeze both at the same moment. Play them end to end and you'll compare the best frame of one to the worst frame of the other and come away with whatever you already believed. Pause both at the bottom position, put them next to each other, and look at one joint.
The honest version of this
You don't need an app for any of the above. Two clips and the photos app you already have will do it, and if you're someone who keeps and labels their footage, that is genuinely the right tool and it costs nothing. That's not a grudging admission; it's the actual recommendation for that person.
The reason this is worth a whole post is that almost nobody is that person. The clip that mattered got deleted in a storage panic, or it's the 1,400th video in the roll, and by the time it would have been useful the before simply doesn't exist.
Two other things worth being straight about. A side-by-side makes a difference visible; it does not make it objective. You're still exercising judgment about what you're looking at. And a pair of angles that are close but not identical will manufacture a difference that was never in your body — that's the failure mode worth worrying about most, because unlike a vague memory, it looks like evidence.
What we built for this
We build Flexion, and the part of it relevant to this post is deliberately mundane: you record the set, the footage gets read frame by frame afterwards, and the read is kept.
Kept is the operative word. There's a toggle called Compare With Previous. Turn it on before a form check and it lists your earlier checks of that same lift; you pick one, and the new video gets read against that specific one — what improved, what still needs work — instead of being assessed in a vacuum for the fifth time. It's off by default, which is honest about how it's meant to be used: you turn it on when you're asking a question about a six-week block, not every time you film.
The same caveats apply to us as to your camera roll. One angle. A judgment, not a measurement. And if you filmed the two sets from different spots, it will describe a change the camera invented rather than one you made — the comparison is only ever as good as the discipline of putting the phone in the same place.
One of the four tests for whether a form check app is doing anything is to change a single cue and refilm. This post is that test with the before kept — which is the only version of it that can actually answer.
