Accuracy Reports: your watches, measured

ChronoLog turns the timing measurements you take — taps, photos, and acoustic timegrapher sessions — into a shareable report: a health score for every watch, charts of how it actually keeps time, and an anonymous comparison against the same models owned by other people. This page explains what each part means, where the numbers come from, and — just as important — what they can't tell you.

Want to see one first? Open the example report — it's built from sample data and every chart on it is live.

The health score

The score blends two things: accuracy — how close the watch's daily rate is to zero — and consistency — how much that rate wanders between timing runs. A watch running a steady +4 s/day scores better than one swinging between −2 and +9, even though their averages look similar. The score is spec-aware: a quartz watch is judged against quartz expectations, a mechanical against mechanical ones.

Rate, drift, and the charts

Rate is the number watchmakers care about: seconds gained or lost per day. The report's chart shows it two ways. The Drift view plots every measurement as cumulative seconds within each timing run — the slope of each line is the rate, so a steep line is a fast watch and a flat line is a chronometer. The Rate view plots the pairwise rate between consecutive measurements, which reveals how stable the rate is from day to day. Acoustic timegrapher readings appear as distinct hollow markers.

The comparison: your watch among its siblings

Every rated watch is placed against a cohort: the same model when enough exist, falling back to the same brand, then to all of ChronoLog. Two percentiles are reported: accuracy (closeness of the daily rate to zero — "keeps time more accurately than 64%") and steadiness(run-to-run consistency). Cohort sizes are deliberately shown as ranges — 20+, 50+, 100+ timepieces — never exact counts.

Privacy, by construction

The reciprocity rule cuts both ways: the comparisons exist because owners contribute their timing data anonymously, so opting out of contributing also closes your own report. Fair's fair.

When a watch qualifies

A watch earns its report once it has six measurements spanning at least seven days — enough evidence that the rate is a property of the watch, not of one afternoon. Until then the app shows how far along each watch is. There's no premium gate on reports; they're built from the measuring you do.

Limitations — read this part

It measures what you measured. Reports reflect synced measurements. A watch you timed for one enthusiastic week in March and never again is honestly described — for March. Rates also legitimately differ between wrist time and drawer time, between positions, and across temperature; the report averages over however you actually measured.

Method precision varies. Tap measurements include your reaction time; photo measurements are corrected against network atomic time but read a dial with one-second resolution; the acoustic timegrapher resolves sub-second rates in minutes but needs a quiet room. Longer spans wash out per-measurement noise — which is exactly why eligibility requires a week.

The comparisons describe ChronoLog owners, not the world.Cohorts are self-selected: people who measure their watches, measuring the watches they care about. A 64th-percentile result among enthusiasts' well-kept examples is not a factory QC statistic, and a health score is a heuristic — it is not a COSC certificate, and it won't diagnose why a watch runs the way it does. For that, see a watchmaker. For whether it's running worth the trip — that's what the report is for.