Step Counting API: How to Get Step Data

For how on-device stores differ from cloud APIs, see on-device vs cloud health data.

Where you can get steps#

Steps are available from a phone alone (its pedometer) as well as from wearables. Provider-specific field names, records, and device coverage change often, so treat this table as a starting point and confirm against each vendor's live data dictionary (as of 2026, verify).

Measured or estimated?#

Steps are counted, but the count comes from motion-sensor pattern-recognition algorithms interpreting accelerometer data, not from a sensor that registers each literal foot-fall. Treat step counts as algorithmic and close, not exact, and do not attach an accuracy percentage — none is sourced, and figures vary by device, gait, and where the device is carried. This is wellness-grade activity data, not a clinical measurement.

The headline gotcha is double-counting across a phone and a paired watch. When a user carries a phone and wears a watch, both log steps, so a naive sum roughly doubles the total. The platforms provide de-duplication, and you should lean on it rather than summing raw samples yourself:

• Health Connect: use aggregate() (for example StepsRecord.COUNT_TOTAL) rather than summing raw readRecords(). Be aware that attribution can span multiple DataOrigin package names, and that on-device step attribution changed in 2026 — legacy readings attributed to the "android" package versus a newer device Synthetic Package Name. Read and verify the current Health Connect steps guidance before you filter or de-duplicate by origin.
• HealthKit: it de-duplicates across sources when you query statistics rather than raw samples. Verify the current behavior in Apple's HealthKit docs before assuming a given query already merges phone and watch.

Because both platforms already reconcile multiple sources, the safest design is to read the aggregated total the OS gives you instead of stitching sources together by hand.

Which source should you pick?#

• Widest reach, least friction: read the on-device platform store — HealthKit HKQuantityTypeIdentifier.stepCount on iOS, Health Connect StepsRecord on Android — so you inherit the OS's own multi-source de-duplication and avoid the double-counting trap. See the setup guides for HealthKit and Google Health Connect.
• Live, in-the-moment step count on iOS: Core Motion's CMPedometer gives you low-level pedometer data directly, separate from HealthKit's stored samples.
• Server-side without the phone present: a cloud wearable API (Garmin, Fitbit, Samsung Health, and others) delivers daily or interval step summaries after the device syncs — useful when your backend needs the data and the user's phone is not in the loop.
• Many brands at once: an aggregator (Terra, Junction, Rook) hands you one normalized steps schema instead of N integrations.

Before you ship#

Field names, records, aggregation methods, and device coverage are volatile — re-verify each against the vendor's current data dictionary as of 2026. In particular, confirm the 2026 Health Connect step-attribution change before shipping any de-duplication logic. Prefer the platform's aggregated total over hand-summed sources to avoid phone-plus-watch double-counting, and frame step counts as an algorithmic wellness signal rather than an exact measurement.