GPS Watches and Running Tech

Distance and pace

Let the watch find satellites before you start, and expect a few percent of error in distance, more under trees and between tall buildings. In a 2020 comparison of eight sport watches from Apple, Coros, Garmin, Polar and Suunto in urban, forest and track areas, mean error ranged from 3.2% to 6.1%, with distances underestimated and less accurate in the urban and forest areas¹. Across ten smartwatches, error ranged from 0.8% to 12.1% over six 4,000 m runs on a track and from 0.2% to 7.5% over six hilly outdoor runs of 3.4 km, and for some watches the results deviated substantially from the true values².

So a watch distance is an estimate. On a route you know, check it against the measured distance. On a twisty, tree-covered or city route, go by effort for pace. In a race, go by the course markers when they disagree with your watch.

Heart rate, wrist or chest strap

On a steady treadmill run, wrist optical heart rate is close to a chest strap or an ECG. A 2020 systematic review of 44 articles (15 brands) found wrist readings were off by about half a beat a minute on average during treadmill walking to running, and by more during resistance training (about 7 beats) and cycling (about 4.5 beats). The authors concluded wrist devices had acceptable validity for most common activities³. An average hides the errors of individual readings. Across ten smartwatches, 30 people ran five 3-minute treadmill intervals, and the mean absolute percentage error for average heart rate ranged from 3.1% to 8.3% against a chest strap².

Use a chest strap when exact heart rate matters, such as heart rate intervals or trail running. In a 2020 study of 21 people on a 3.22 km out-and-back trail run with a chest strap as the reference, a wrist watch had a mean absolute percent error of 13%, a forearm band 6%, and a watch worn with its paired chest strap 2%. The authors concluded all the optical devices agreed poorly during variable-intensity trail running and that heart rate outdoors should come from a chest strap connected to a watch⁴.

Wear the wrist band snug. In a 2023 test of one wrist monitor in 30 people on a treadmill and a bicycle ergometer against a 10-lead ECG, the error on the treadmill was 16.1% with a normally tight band and 8.1% with the band tightened as far as tolerable, and it was higher in transitions than in steady state (9.0% against 7.6%)⁵.

VO2max, threshold, recovery and sleep numbers

These numbers are calculated from other measurements, and the studies found larger errors than for heart rate itself. In 19 people, an Apple Watch Series 7 VO2max estimate averaged 4.5 mL/kg/min below the laboratory value, with a mean error of 15.79%, and the authors described the reliability as poor⁶. In 34 people, a Google Pixel Watch 3 estimate was close on average (bias -0.53 mL/kg/min) but for an individual could be off by about 12 mL/kg/min either way, and the authors said the estimates lack individual-level validity for clinical use⁷.

For threshold, 100 recreational runners did a laboratory graded test and then outdoor tests with smartwatches. Threshold heart rate showed no significant difference from the laboratory method on average, with mean absolute errors of 8.93 to 11.44 beats a minute, and the authors said the watches tend to overestimate threshold pace⁸. For heart rate zones, Runsense reads your threshold from a hard effort you ran. See the guide on testing your lactate threshold heart rate.

Treat recovery and readiness scores as a nudge to check how you feel, not a verdict. In 39 adults wearing an activity tracker for three months, self-reported stress and nervousness were not associated with overnight heart rate variability, and feeling energized went with lower variability (64.1 against 66.9 ms). The authors cautioned against relying on HRV-based recommendations⁹.

A watch's sleep total runs high. In 70 masters endurance athletes wearing a research-grade wrist sensor and a Garmin smartwatch for 7 nights, the smartwatch recorded sleep 126 minutes longer than the research sensor and the diary 109 minutes longer, and the authors advised caution, particularly in athletes with shorter or more fragmented sleep¹⁰. Use the trend over weeks and not a single night, and see the guide on sleep.

How a watch connects to Runsense

Runsense reads runs from Apple Health or from direct connections to Strava, Suunto, Polar and COROS. Garmin has no direct connection, so a Garmin run arrives through Strava or Apple Health. A direct connection brings the whole recording the provider holds, while a run relayed through Apple Health can arrive with fewer heart rate samples, depending on the app that wrote it. Where a connection carries them, Runsense also reads a watch's own overnight sleep, resting heart rate and heart rate variability.

For Suunto, Runsense also sends upcoming sessions to the watch as SuuntoPlus guides, on Suunto watches that support SuuntoPlus. For an Apple Watch, it schedules upcoming sessions in Apple's Workout app. COROS is a connection for runs and nightly data only, and Runsense does not send workouts to a COROS watch.