What electronic shifting records in your ride files
Electronic groupsets do more than change gear: your head unit can log every shift with the exact chainring and cog. That turns each ride into a record of how you actually use your drivetrain.
What ends up in the file
When a Shimano Di2 or SRAM AXS groupset is paired with a compatible head unit, the device writes an event into the FIT file each time you shift. Each event carries a timestamp and the position of the front and rear derailleurs, usually as both a gear number and the number of teeth on the chainring and cog. The 1-Hz records of speed, cadence, power and heart rate sit in the same file, so every shift can be placed on the timeline of the ride.
That is all the file contains: a list of "at 10:42:17 the drivetrain went to 50×17". Everything else, such as how long you stayed in a gear, is derived by walking the timeline between events. GearWatts does exactly that and sums time, distance, power, cadence and speed for each chainring × cog combination. It stores sums rather than averages so that many rides can be added up exactly later, for a season or for one bike.
Teeth or index: why it matters
Some devices write the tooth counts, others only the index (front ring 1 or 2, cog 1 to 12). The index is ambiguous without knowing the drivetrain, and conventions differ between brands. When only the index arrives, GearWatts resolves it with the transmission stored in your bike profile (front ring 1 is the smaller one, cog 1 the largest). Teeth, when present, are used as they are.
Two practical consequences. First, enter your real chainrings and cassette in the bike profile before judging any result; a 34-tooth cog that is really a 32 shifts every figure. Second, data can be wrong. One older head unit paired to a bike wrote "chainring 7, 24 teeth" for whole rides, which is impossible for a double or triple. A sensible analyser discards such values and treats the gear as unknown instead of inheriting the previous one, and GearWatts does so.
What you can learn from it
Time and power in each combination. Which gears do you really live in? Many riders discover that two or three combinations cover most of their time, and that the extremes of the cassette are used for minutes per month.
Gear by gradient. Grouping samples into 1 % bins of gradient shows the gear you choose, the share of time on the big ring, and your cadence and speed at each slope. It is the honest picture of your gearing: if at 8 % you sit in the last two cogs with 60 rpm, the issue is not your legs but your ratios.
Shifting habits. Shifts per hour of moving time, front versus rear shifts, the median time between shifts, and "quick reversals" (undoing a shift within 10 seconds, a sign you picked the wrong gear). GearWatts also compares the cadence in the 5 seconds before and after each rear shift and counts shifts made while your power was above FTP, which is what stresses the chain and shows poor anticipation.
Cross-chaining. GearWatts flags the big chainring combined with the three largest cogs, or the small ring with the three smallest. These combinations run the chain at an angle, add friction and noise, and are often unnecessary because a neighbouring combination gives almost the same development (see gear development in metres per pedal revolution).
A worked example: should I change the cassette?
Take a rider with 50/34 and an 11-34. After 20 rides, the usage table, with time spent per gear, shows:
| Combination | Share of time | Typical cadence |
|---|---|---|
| 50×15 – 50×19 | 41 % | 86–92 rpm |
| 34×21 – 34×27 | 33 % | 78–84 rpm |
| 50×11 – 50×12 | 1.5 % | 95 rpm |
| 34×30 – 34×34 | 4 % | 66 rpm |
(Illustrative numbers.) The top two cogs are almost never used, and the easiest ones appear only on the steepest 12 % ramps. Two options appear. Keep the 34 but move to an 11-30 cassette: the middle steps become 1–2 percentage points finer and you give up 34×34 only for the few minutes where cadence already falls to 66 rpm. Or, if those ramps are the reason you ride, keep the 34 and accept the jump. The data does not tell you what to buy, but it replaces "I feel I need a bigger cog" with a number: 4 % of the time.
The same reasoning works per climb: GearWatts shows the gear of each section of a climb and a mark at every shift, so you can see whether you ran out of gears on the last ramp (see how climbs are detected and categorised).
Caveats you should know
- Not every head unit records shifts. The device must support the shifting protocol (ANT+ or Bluetooth) and the setting must be on. If a ride has no shift events, there is no way to reconstruct them afterwards. A ride recorded on the same bike before pairing simply has no gear data.
- Some services strip them. Platforms that give you the activity through an API often omit gear events and the original FIT file. When a ride arrives twice, GearWatts prefers the FIT version because it carries shifts and pedalling data. For a full export, see how to export FIT files.
- Gear before the first shift. The file starts logging at the first change. GearWatts assumes the first recorded gear for the time before it; a few minutes at the start may be wrong.
- Synchro or semi-automatic modes. If the system shifts the front derailleur for you when you reach the end of the cassette, those shifts are real events but not decisions of yours; keep that in mind when comparing "shifts per hour" between riders or between bikes.
- Sample, not truth. A ride in the wind, a group ride or a race changes the picture. Compare similar rides and look at weeks, not a single day.
Handled with care, shift data answers questions that power and heart rate cannot: not how hard you worked, but how well your gearing let you do it.