Quiet Metrics

Wearables arrived with a promise that felt almost clinical: turn physiology into information, information into better decisions. In the U.S., that promise has scaled fast. In a large consumer sample from 2020–2022, 44.5% of participants reported owning at least one wearable device.

What’s emerged alongside adoption is a more specific question—less “Do wearables work?” and more “Which outputs remain dependable enough to use without being misled by them?” Independent validation work suggests the accuracy landscape is uneven: some measurements track reasonably well under certain conditions, while others drift with device, algorithm, and context. A 2024 Sports Medicine living umbrella review of systematic reviews frames this as a core issue in consumer wearables: they are widely used, but biometric accuracy varies by outcome and circumstance.

That variability matters because modern wearables rarely present raw signals. They present conclusions: readiness, recovery, strain, stress. These composites can be useful as private heuristics, but the scientific caution is straightforward—when a metric is a proprietary score rather than a direct measurement, interpretability declines, and independent verification becomes harder.

So the most evidence-aligned way to use a wearable looks less like obeying a dashboard and more like treating the device as a trend instrument—one that is better at surfacing stable patterns than delivering nightly verdicts.

Across the literature, three signals keep reappearing as comparatively defensible for everyday use because they’re either more directly measured or strongly linked to meaningful health outcomes in large datasets: movement volume (steps), sleep duration and timing regularity, and resting heart rate trends.

Steps are the blunt metric that doesn’t require a translator. They don’t capture every form of training well, but when researchers look at device-measured movement volume, step counts repeatedly map to long-term outcomes. In U.S. adults using NHANES accelerometer data with mortality follow-up through 2019, the number of days per week people reached 8,000 steps mattered: those who hit 8,000 steps on 1–2 days/week or 3–7 days/week had lower all-cause and cardiovascular mortality risk than those who never reached that threshold, with a curvilinear pattern and a plateau after several days. The implication isn’t that 8,000 is magic; it’s that step patterns can capture something real enough to show up in outcomes research.

Sleep is the domain wearables market most aggressively—yet it’s also where measurement can become most psychologically sticky. In the U.S., CDC surveillance using BRFSS frames the baseline problem clearly: adults are recommended to get at least 7 hours per night; reporting less than 7 hours is classified as insufficient sleep, and national trends show insufficient sleep has remained a persistent public health issue. Wearables can help by making duration and timing visible, but the limitation is built into the technology: consumer devices infer sleep from movement and physiological proxies, which can be good enough for broad patterns, and less reliable for fine-grained staging or nightly “quality” scoring.

Where sleep science has moved in the past few years is toward regularity as its own signal—how stable your sleep-wake timing is over the week, not only how many hours you accumulate. A large device-based prospective study in Journal of Epidemiology & Community Health examined sleep regularity and major adverse cardiovascular events and reports that irregular sleep was strongly associated with higher risk, and sufficient duration did not fully offset that association in irregular sleepers. (This is observational evidence, so it does not establish causality; it does show the strength of the association in a large cohort with device-based measurement.)

There is also a well-described failure mode when sleep tracking becomes an optimization contest rather than a measurement tool. The term “orthosomnia” was introduced in a 2017 Journal of Clinical Sleep Medicine paper describing patients whose pursuit of “better” tracker numbers corresponded with worse sleep and insomnia-like behavior. The research takeaway isn’t that sleep tracking should be avoided; it’s that sleep data is better used to stabilize patterns than to chase perfection.

Resting heart rate is the third signal that tends to behave like a measurement rather than a mood. It’s influenced by many factors—fitness, illness, sleep debt, hydration, medication, training load—but that breadth is part of its value as a background indicator. Importantly, research suggests the trend over time carries information beyond a single snapshot. In the ARIC cohort, a 2018 JAMA Cardiology analysis reported that higher resting heart rate and increases over time were associated with higher risk for outcomes including all-cause death and incident heart failure. Wearables aren’t clinical monitors, but in validation syntheses, heart-rate measurement is generally among the more defensible consumer outputs compared with more interpretive constructs like “stress” scores.

Put together, these three metrics share a practical property: they remain interpretable even when stripped of branding and scoring language. Steps describe volume. Sleep duration and timing describe rhythm. Resting heart rate describes baseline load and drift. They also map to actions that don’t require a new identity or a new app: move more across the week, stabilize sleep timing, and treat sustained resting heart rate changes as a cue to reassess recovery, illness, or training load—with medical evaluation when symptoms or alerts warrant it.

A wearable becomes more scientifically aligned when it’s used for weekly trends, not daily judgment.

Start with steps as the anchor, because it is the least interpretive metric and has strong outcomes data behind it. Then use sleep tracking primarily to monitor duration against the CDC’s ≥7 hours benchmark and to reduce large swings in timing across the week, rather than obsessing over stage breakdowns. Keep resting heart rate as a drift signal: a sustained rise above your usual baseline can coincide with illness, cumulative sleep loss, or excessive training load; a sustained decline can coincide with adaptation, but context matters.

Finally, align device data with public health fundamentals rather than competing with them. The CDC’s activity guidance for adults remains at least 150 minutes/week of moderate-intensity activity plus muscle-strengthening on 2 days/week; steps can support the aerobic volume, but they don’t replace strength work.