Personal baselines: why we compare you with you

Most health metrics are judged against population norms. A resting heart rate is "normal" if it falls within a broad range. Sleep is "enough" if it meets a general guideline. That's useful for screening a population. It's much less useful for spotting change in one person.

The problem with averages

People differ. Two people can have very different heart rate variability and both be perfectly healthy. A reading that is normal for one could be a warning sign for the other.

When you compare individuals against an average, two things go wrong:

• Real change gets missed. Someone whose HRV has dropped sharply may still sit inside the population range, so nothing is flagged.
• False alarms appear. Someone who has always had a lower reading gets flagged, even though nothing has changed for them.

Shaffer and Ginsberg's overview of heart rate variability metrics and norms makes the point that HRV varies widely between individuals and is influenced by age, fitness and many other factors. That variation is exactly why population norms are a weak benchmark for individuals.

What a personal baseline is

A personal baseline is each person's own normal, built from their own data over time. For wearable signals, that means a rolling window, for example their last 28 days, that updates as new data arrives.

Each new reading is then expressed relative to that baseline. Instead of asking "is this number high?", we ask "is this number unusual for this person?"

Why this comes from sport

This is standard practice in athlete monitoring. Coaches and sports scientists don't judge an athlete's recovery against other athletes. They track each athlete against their own history, because what matters is whether that person is adapting well or heading towards overreaching.

Our exercise physiologists work this way with endurance athletes every week. Everis applies the same discipline to the workplace.

Separating signal from noise

Day-to-day readings vary naturally. A late night, a hard workout or a long flight can shift a single reading. To avoid reacting to noise, we:

• use rolling and weighted averages that smooth single bad days,
• account for the typical measurement error of each signal,
• only flag change that exceeds a smallest meaningful change threshold and persists over time.

What this means for employees

Nobody is judged against anyone else. A person isn't flagged because their numbers look different from a colleague's. They're only flagged if their own pattern shifts in a meaningful, sustained way, and then only they see it first.

What this means for your organisation

Personal baselines make the data fairer and more accurate. Fewer false alarms mean physiologist time goes where it's needed. Fewer missed signals mean earlier support. When aggregated into de-identified team indices, baseline-adjusted data gives leaders a cleaner view of where strain is genuinely building.

Sources: Shaffer F, Ginsberg JP (2017). Full references on our Science page.