How to Read Your PaddleLens Report (And Actually Use It)

Here's a failure mode we see constantly, and it has nothing to do with the accuracy of the analysis: someone gets their report, looks at the number at the top, feels some combination of pleased or defensive about it, and closes the app. Two weeks later, nothing's changed, because nothing about "7.2" tells you what to do.

The report has more in it than the score. Here's what each part is actually for.

The headline fix — read this first, and mostly only this

At the top of a scored report is one section: a practical focus chosen from the weakest supported score. It is coaching generated from the measurements and targets, not a guarantee that this is the only change that matters.

This is deliberate, and it's the same principle behind running a video review your crew won't hate: one correction is easier to train than five. If several supported criteria score well and projected outside-arm posture does not, the headline focus will usually be the arm — while a coach remains free to choose differently.

What to do with it: turn it into a single cue you can say to yourself mid-stroke. "Straighten my outside arm at the catch" is usable on the water. "Improve my overall technique" is not.

The skeleton replay — watch it, don't just read about it

Below the headline is a colored overlay on sampled frames from your footage. It shows which joints the pose model found, its joint-confidence signal and which bones were excluded from measurements. The overlay is evidence for review, not proof that every tracked point or derived angle is correct.

What to do with it: watch three or four consecutive strokes, specifically at the moment the headline fix is about. Don't judge everything you see — you're checking one thing, the same discipline we've written about for coached video review applies just as well to reviewing your own footage alone.

The key numbers — beside the advice, not on top of it

Alongside the report sits a panel of the supported measurements: projected arm angles, torso lean, stroke rate and other view-dependent observations. Only criteria with a formula and sufficient evidence receive a score; others may be written up or left unavailable. The measurements sit next to the coaching so observation and interpretation can be checked separately. (For the fuller breakdown, including each interquartile range and available targets, save the report as a PDF.)

Worth knowing: a score of zero doesn't mean you were terrible at something. It means that measurement wasn't available in this clip — the camera angle didn't support it, or too few frames were readable. The system leaves things out rather than guessing at them.

What to do with it: mostly, ignore it for the two weeks you're working on the headline fix. It's genuinely useful for two situations: confirming why something is the headline fix (which specific number is dragging things down), and — later — checking whether fixing it accidentally moved something else, which happens more than you'd think, since these numbers aren't independent of each other.

Tracking coverage — evidence quantity, not accuracy

Near the top of the report is the share of sampled frames that contained a usable paddler track. It tells you how much tracking evidence survived, not whether each joint or angle is accurate.

A low number is a clear reason to inspect the footage. A high number is not a blanket pass: camera view, subject size, occlusion, joint trust and clear cycle count can still make an individual measurement approximate or unavailable.

What to do with it: review the annotated frames and each criterion's status. If coverage is low or the required evidence is missing, re-film using the filming guide. Do not turn the percentage into a confidence or accuracy score.

Stored score history — compare like with like

The current chart stores each report's average of scored criteria, oldest to latest. It is useful context, but reports may differ in camera view, supported metric set, targets and analysis method. Until those are versioned and matched automatically, the line is not a universal performance scoreboard.

What to do with it: compare sessions filmed from the same view and setup, under similar conditions, and inspect the underlying measurement rather than only the average score. Say that a number changed; let the coach and context decide whether it improved.

Badges — small, but they're not nothing

Your first report. Every measured area above 7.5 in a single analysis. Three reports in a row where the score climbs. Six practices without a miss. Seventeen badges in all, across attendance, scores, trends and milestones, visible on your home screen — no extra effort required beyond doing the thing that already earns them.

It's easy to dismiss this as decoration, and it is partly decoration. But it's doing something specific: a progress chart creeping up over three months is real progress and also, honestly, quite hard to feel in the moment. A badge landing on the day your average crosses 8.0 makes that moment register. Two of them are worth knowing about — Comeback, for beating your best after four weeks away, is the one that catches people coming back from injury or a busy winter; and Coach's Pick is the only one no algorithm can award, because your coach hands it to you by starring a report.

What to do with it: nothing, really — just notice them. They're not a task list.

The two-week loop that actually works

Putting all of it together, here's the rhythm that turns a report into progress instead of a screenshot:

1. Read the headline fix. Ignore the rest of the report for now.
2. Turn it into one cue you can say to yourself on the water.
3. Train that cue for two weeks — nothing else, even if you notice other things you'd like to fix. Write those down for later; they'll still be there.
4. Re-film the same way (our filming guide reduces avoidable setup differences).
5. Check the underlying measurement and evidence, not just the average score. Did the supported number move under a comparable setup? Does the coach agree with what that change means?

Half the value of an analysis is the number. The other half is resisting the urge to read it as a verdict instead of a to-do list with exactly one item on it.