How it reads

What happens in the one tap.

You tap once and get a clean read of your session. Behind that tap is the work of turning a raw watch file into something true.

Your watch hands us a list of laps and almost nothing else — no labels, no sense of which split was a rep and which was you waiting at a crossing. Reflect does the quiet work of reading that mess the way a careful coach would: sorting the hard from the easy, throwing out what the watch got wrong, and matching it against what you actually set out to do. The result reads simple because the judgment underneath it isn't.

The file, as the watch recorded it — laps in order, nothing sorted yet

The same run, read — each lap placed in the run it belongs to

Work rep Recovery Warm-up & cool-down Dropped
Watch paused mid-run — no effort to read

One run, read the way a coach would read it — the labels your watch never recorded.

01 First, throw out what the watch got wrong

It throws out what the watch got wrong.

Pause your watch at a light and a single “lap” can read as 33 min/km. Reflect spots the laps that are artifacts, not effort, and sets them aside before they distort the read.

The discarded lap leaves a faint gap, not a silence — so the judgment (“we didn't trust this one”) is felt, not hidden.

One lap, inspected

Lap 7 · recorded pace
33:04min / km
An artifact, not an effort — discarded.

02 The hard part

Hard from easy, with no labels to lean on.

Two laps can share the same pace and the same length and still be different jobs — and pace alone can't separate them. So Reflect reads where each lap sits in the run.

Working out which lap was the session — and which was you catching your breath — is the part that takes several passes.

Same pace, same length — told apart by where they sit

Opens the run
Warm-up
first lap of the run
Sits between recoveries
Work rep
boxed by recoveries

The two “?” laps are identical. One opens the run; one sits between recoveries — context is what tells them apart.

03 Then match it to the plan

A stride is not a rep.

Because Reflect knows the session you planned, it can tell a six-second pickup from a one-kilometre rep — two things a watch files identically. Matching weighs several signals at once.

And when the evidence isn't there, it asks instead of guessing.

Your run, lined up to plan

Planned session
6 × 800 m
intervals · with recoveries

Six reps · lined up to the six in the plan

The match

Six reps, where six were asked for. This was your 6 × 800. Counted.

An odd block midway, neither clearly a rep nor a recovery.

“Not enough evidence to call this.”

So it asks, rather than guessing

04 The payoff

The drift read.

Distance and pace against the band you intended; whether you ran your easy day too fast; on intervals, the rep count, the per-rep pace, and rest you over- or undershot. And the line that matters most: whether you faded across the set, or finished it strong.

Read off the same run

You faded on the back half — the last two reps slower. Not a negative split.Per-rep pace

Your recoveries ran long — you took more rest than the plan asked for.Rest drift

The warm-up was an even, calm stride — exactly the easy you intended.Easy-pace check

We didn't tune this on synthetic tests. We hardened it by replaying real marathons, lap by lap, until the read held up.

A glitch lap, a warm-up the exact length of the reps — real running produced them; test data never did.

Depth, in service of trust

When the evidence isn't there, it asks instead of guessing.

A read of your training you can take at face value.

That restraint, and the work behind it, is what you're paying for.