"Is this tracker accurate?" has no single answer, because a tracker reports several different measurements, and each one is checked against a different yardstick. Fitness tracker accuracy for heart rate means something other than accuracy for steps, sleep or calories. A device can do reasonably well on one and poorly on another. So before you compare devices, name the metric, the reference it should be judged against, and the activity you care about. This article gives you a reference table and a check log for doing that, and it deliberately does not produce an overall score.
Start by Choosing the Metric You Are Comparing
Write down one sentence before you look at any numbers: "I want to know whether [device] measures [metric] correctly during [activity]."
Each part of that sentence changes the answer:
- Metric: heart rate during a workout, daily step count, nightly sleep time, or calories burned are four different questions.
- Activity: walking on a treadmill is not the same as lifting, cycling, or a day of mixed movement. Wrist sensors cope with these differently.
- Decision: are you tracking your own trend over time, or comparing two devices on the same day? Trend consistency and absolute correctness are separate standards.
This framing doesn't apply if you want a health interpretation. A consumer device reading is not a diagnosis, and this article doesn't cover that. Using readings to adjust training is also a separate topic; see Exercise Heart Rate Zones Explained for what heart-rate numbers are used for once you trust them.
What the Research Says About Accuracy
Validation research does not support one verdict for "wearables". Two examples show why.
- A systematic review of commercial wearables summarised validity and reliability separately for step count, heart rate and energy expenditure. It also noted that newer and smaller manufacturers often have few or no validation studies. Evidence for a metric on one device does not transfer to another.
- A 2020 validation study tested two wrist optical trackers against a chest-strap monitor in 20 young and 20 older adults doing sedentary and moderate activity on indoor aerobic equipment. Mean absolute percentage error stayed under the authors' 10% threshold in every group, but both wrist devices underestimated average heart rate, and some participants had unexpected outlier readings. That is a finding about those two devices, that sample and those activities, not about wrist heart rate in general.
When you read any fitness tracker accuracy study, a percentage belongs to a specific device model, activity, reference, and group of participants. Without those four, it is a headline, not evidence.
Reference Methods by Metric
The table below sets out what each metric is usually checked against. The reference column is the yardstick; the acceptance column is your own judgment rule, not a published threshold.
| Metric | What "accurate" means | Typical reference method | Activity and population to match | Acceptance judgment | Main limitation |
|---|---|---|---|---|---|
| Heart rate | Beats per minute agree with the reference at the same moment | ECG or an ECG-style chest strap | Steady vs. interval vs. strength work; your skin, wrist fit, and movement | Define a tolerance first; check at rest, steady effort, and peaks separately | Wrist optical sensing can struggle during rapid changes and gripping |
| Steps | Counted steps match actual steps | Manual or video count over a set number of steps | Walking, running, or daily life; stride and arm movement | Compare counts for the same bout, with the same source | Free-living totals are harder to verify than a counted walk |
| Sleep | Time asleep or stage labels match a lab record | Polysomnography | Who was studied: healthy adults, a specific condition, one night vs. many | Look for agreement on total sleep first; treat stage claims cautiously | Stages are the hardest to verify at home; you can't check them yourself |
| Energy expenditure | Calories match measured energy use | Indirect calorimetry in the lab; other methods for free-living | Activity type, body size, and whether the test was a lab bout or a whole day | Treat as an estimate; judge consistency, not exact calories | A home check isn't possible; you rely on lab-based studies |
Heart rate is one metric you can check outside a lab by wearing a wrist tracker alongside an ECG-based chest-strap monitor during the same workout. Compare readings taken at the same time, including during steady effort and changes in intensity. Treat the chest strap as a practical reference rather than a perfect measurement standard, and remember that this comparison only evaluates heart rate—not steps, sleep stages, or calorie estimates.
Sleep stages and mismatched step totals from two devices each need more than a framework, so this article only flags them. Stage claims call for a professional-grade reference. Differing step totals raise a data-source question, such as whether two sources are counting the same movement, before an accuracy one.
Check the Test Population and Activity
When you read a validation result, compare it with your situation using this sequence.
- Input: the device model and generation, the activity, the reference, and the sample.
- Judgment: does each of those match what you own and do? A study of a treadmill walk in healthy adults tells you little about kettlebell work or about a different wrist.
- Output: a narrower claim, such as "these two wrist devices stayed within 10% average error against a chest strap on indoor aerobic equipment," instead of "this brand is accurate."
If any of the four is missing, narrow the claim or set the study aside.
An Illustrative Check Log
The numbers below are made up to show the method. They are not from a real test.
Question: does my band's heart rate agree with a chest strap during a steady 20-minute cycling session?
| Time point | Chest strap (reference) | Wrist device | Difference |
|---|---|---|---|
| Minute 5 | 118 | 116 | −2 |
| Minute 10 | 132 | 131 | −1 |
| Minute 15 | 140 | 139 | −1 |
| Minute 20 (after a sprint) | 158 | 146 | −12 |
Reading the log: if you had set a tolerance of ±5 beats per minute before the session, the first three points pass and the minute-20 point does not. That is all the log shows: the two devices disagreed at that moment. It doesn't tell you why. To say anything about lag, you would need time-aligned data, such as second-by-second recordings from both devices with synced clocks, so you could see whether the wrist trace follows the strap trace a few seconds later or stays off. Repeating the session a few times would show whether the disagreement recurs. Even then, the conclusion stays within heart rate on this activity; it says nothing about steps or sleep.
A No-Overall-Score Checklist
Use this when you compare devices or read a claim:
- [ ] I named one metric and one activity.
- [ ] I know the reference method for that metric (ECG or chest strap, counted steps, polysomnography, calorimetry).
- [ ] I have the exact device model and generation, not just the brand.
- [ ] Any study I rely on states the activity, reference, and sample.
- [ ] I did not carry an old model's error figure over to a new one.
- [ ] I set my tolerance before looking at the comparison.
- [ ] I fixed the inputs I control first: fit, workout type, personal details, any calibration the device offers.
- [ ] I judged each metric separately and wrote a narrow conclusion for each.
- [ ] I did not combine metrics into one score, because no evidence supports how to weight them.
- [ ] I treated the result as a fitness estimate, not a medical reading.
A table of separate, narrowly worded results is a better fitness tracker accuracy comparison than a single number, because it shows which readings you can lean on and which you can't.
