Smartwatches can provide useful insights into health and physical activity, but many of the numbers they display are estimates rather than direct measurements of the body. A new study from the University of Michigan found that the reliability of smartwatch metrics varies considerably, highlighting the need for users to understand what their devices can — and cannot — accurately measure.

Millions of people use smartwatches to monitor sleep, steps, heart rate, exercise and other aspects of their health. As wearable technology has become more sophisticated, however, the information presented to users has also become increasingly complex. Some measurements closely reflect signals detected by a watch’s sensors, while others are calculated using algorithms and assumptions.

Adam Lepley, an assistant professor at the University of Michigan School of Kinesiology, and his research team developed a framework to help people better understand what smartwatches actually measure, what they estimate and how their data should be interpreted. Their findings, published in the journal Sensors, emphasise that smartwatch metrics should not all be treated in the same way.

“The most important takeaway is that not all smartwatch metrics should be interpreted the same way,” Lepley said. Some outputs are relatively close to what a sensor detects, while many others are estimates produced by combining sensor signals with proprietary algorithms, information about the user and other assumptions. As a result, users should be cautious about treating every number displayed on a smartwatch as an exact measurement.

Smartwatches gather information using several types of sensors. Optical sensors shine light into the wrist to detect changes in blood flow, while motion sensors monitor movement. GPS and other technologies can provide information about location, speed and additional signals. Algorithms then process these raw signals and transform them into familiar measures displayed on the watch or its accompanying app.

According to the researchers, smartwatch data are generally most useful for monitoring changes in the same person over time rather than providing precise, laboratory-quality measurements. For example, a consistent change in resting heart rate, sleep patterns or physical activity over several days or weeks may be more informative than a single unusual reading. Lepley said wearable devices are therefore often better suited to identifying trends than producing exact measurements.

Some smartwatch measurements also appear to be more dependable than others. Resting and steady-state heart rate, step counts and outdoor pace tend to be relatively reliable. More complicated estimates — including calories burned, sleep stages, body composition, hydration and recovery — should be interpreted with greater caution because they depend more heavily on algorithms and indirect calculations. Accuracy may also be influenced by movement, watch fit, temperature, sweat, skin tone, tattoos and body composition.

Another complication is that results from different smartwatch brands may not be directly comparable because manufacturers use different sensors, algorithms and definitions to calculate their metrics. For the study, the researchers conducted a narrative review using topic-focused searches of PubMed, SPORTDiscus and Google Scholar through June 2026. They also examined reference lists, technical and regulatory documents, and professional guidance. The findings suggest that smartwatches can be valuable tools for understanding personal health trends, but their numbers are best viewed as useful indicators rather than unquestionable measurements.

More information: Adam Lepley et al, Consumer Smartwatch Technology in Health and Performance Research: Validity, Limitations, and Real-World Applications, Sensors. DOI: 10.3390/s26144486

Journal information: Sensors Provided by University of Michigan