Ski trackers love accuracy language — “professional grade GPS,” “precision tracking.” Rarely does anyone get to check. In 2023, researchers did: they ran four ski apps simultaneously (Strava, iSki, Slopes, Skitude) on four phones, against a 50 Hz differential-GNSS reference rig carried on the skier’s back, across days at Davos and Zermatt. It’s the only study of its kind we know of, and it’s openly published.
The findings, condensed:
So when ski apps compete on “GPS quality,” they’re competing on a layer where science says they’re tied. The honest competition is in what happens after the fix arrives. Here’s Alpy’s processing pipeline — all of it, because an instrument that hides its method is asking for faith, not trust.
Before a fix touches anything, two tests from the benchmark literature’s own recommendations: a fix reporting horizontal accuracy worse than 50 m is discarded as noise, and a jump implying more than 150 km/h between fixes is discarded as a receiver glitch. A rejected glitch is also never used as the comparison baseline for the next fix — one teleport can’t poison an honest point behind it.
Phone speed comes from satellite Doppler, which is good — and occasionally produces a single spectacular spike. The infamous “300 km/h top speed” screenshots come from exactly this. Alpy only lets a speed reading contend for your day’s maximum if the actual displacement between fixes roughly supports it (with a generous allowance, because the straight line between two fixes always under-measures a carved path). A spike with no movement behind it is dropped — not averaged in, not error-barred, dropped. And the top speed that survives is still printed with its measurement uncertainty: 54 ±3 km/h.
The study’s bluntest finding is that raw GNSS altitude is the worst channel. Alpy agrees, and fuses the phone’s barometric altimeter — sub-metre relative precision — with GPS as a slow absolute anchor. Runs, vertical totals and airtime detection all ride the fused curve. The method is written up in detail in our vertical-measurement post.
A tracker’s worst failure isn’t a noisy point — it’s a phone dying at 15:40 with the whole day unsaved. (Ski Tracks, the elder statesman of the category, got this right years ago; credit where due.) If your battery reaches 10% while tracking and you’re not on a charger, Alpy stops, saves the day, and tells you — the record survives even if the phone doesn’t.
GPS needs open sky and a minute to settle. While fix quality is still poor — first minutes, deep lodge, canyon walls — Alpy says “acquiring GPS, numbers firm up with open sky” instead of rendering confident garbage. The study found position error roughly doubled in obstructed terrain; the least an app can do is admit when it’s in that regime.
Buy whichever ski app you like — the satellites don’t care. But ask what it does between the antenna and the screen, because that’s where the real differences live: gating, cross-checking, fusing, saving early, and admitting uncertainty. Our pipeline is public, our numbers carry their error bars, and the study that keeps everyone honest is right here.
Alpy is a free, private ski app for iPhone. Every number labeled by how it was measured. The app · More from the blog.