TL;DR: Sleep duration 7 to 9 hours, sleep efficiency above 85 percent, sleep latency 10 to 20 minutes, WASO under 30 minutes, deep sleep above 60 minutes, REM above 90 minutes, SpO2 above 95 percent. Those are the robust metrics. Sleep scores and single-night stage percentages are mostly marketing. Trends over 2 weeks beat any individual night.
This article does not replace a sleep physician. If you suspect sleep apnea or chronic insomnia, see a specialist.
Why Track Sleep at All
Your subjective feeling on waking is a poor indicator of sleep quality. Most people underestimate deep sleep deficits and overestimate how long they actually slept. A wearable delivers objective data: when did you fall asleep, how often were you awake, how stable was your heart rate? Those are numbers you would never see without measurement.
The real value is not in a single night. It is in weekly and monthly trends. You see your deep sleep drop by 40 minutes after three nights of beer. You notice your HRV takes three days to recover from long-haul travel. And you catch early signals when something changes — a rising resting heart rate over two weeks is often the first sign of an infection or an overtraining phase.
Combine sleep tracking with biomarker data and you see patterns that are invisible in isolation. Low ferritin and poor sleep? High morning cortisol and early waking? Those are the patterns that matter.
The Four Sleep Stages in Brief
A sleep cycle lasts 90 to 120 minutes. You run through 4 to 6 cycles per night. Each cycle contains four stages with distinct functions.
| Stage | Share | Function |
|---|---|---|
| N1 (falling asleep) | 2 to 5 % | Transition from wake |
| N2 (light sleep) | 45 to 55 % | Memory consolidation, muscle relaxation |
| N3 (deep sleep, SWS) | 13 to 23 % | Physical recovery, growth hormone peak |
| REM | 20 to 25 % | Cognitive consolidation, learning, emotion processing |
Deep sleep (N3) dominates the first half of the night. Going to bed early gives you more of it. Growth hormone release, immune repair and muscle recovery run here.
REM dominates the second half of the night. If you sleep only 6 hours you disproportionately lose REM. You feel it the next day as reduced cognition and worse emotional regulation.
A concrete example: you go to bed at 11 PM and get up at 6 AM — that is 7 hours in bed. At 85 percent efficiency that is about 6 hours of sleep split across 4 to 5 cycles. If you go to bed at 1 AM and get up at 7 AM, you miss an entire cycle with full deep sleep.
The Core Metrics That Matter
These eight numbers are the foundation. Everything else is decoration.
Sleep Duration
Adults need 7 to 9 hours. Under 6 hours raises the risk of cardiovascular disease, insulin resistance and cognitive decline. Regularly over 9 hours suggests poor sleep quality or an underlying condition. Find your personal optimum over 4 weeks of tracking — if you thrive on 7.5 hours, you do not need 9.
Sleep Efficiency
Sleep efficiency = actual sleep time / time in bed. Good: above 85 percent. Excellent: above 90 percent. If you spend 8 hours in bed but sleep only 6.5, your efficiency is 81 percent — time is being wasted. Typical causes: too much WASO, long sleep latency, going to bed too early.
Sleep Latency
The time from lights-out to falling asleep. Ideal: 10 to 20 minutes. Under 5 minutes usually means you are overtired. Above 30 minutes points to sleep-onset problems — typical causes are afternoon caffeine, too much blue light before bed, or nighttime rumination.
WASO (Wake After Sleep Onset)
Time you spent awake after falling asleep. Ideal: under 30 minutes. Brief wake-ups are normal and usually unnoticed. WASO above 60 minutes fragments sleep and makes it unrefreshing. Common causes: alcohol, warm bedroom, noise, full stomach.
Deep Sleep
Target: above 60 minutes per night. Individual wearable numbers are noisy, but your personal 2-week average is reliable. Alcohol measurably cuts deep sleep even if you fall asleep faster. Late exercise (after 8 PM) and warm bedroom temperature also reduce deep sleep.
REM Sleep
Target: above 90 minutes per night. REM concentrates in the second half of the night — early risers lose disproportionate REM. Alcohol is the strongest REM killer: two glasses of wine can cut REM by 20 to 30 percent. Antidepressants, THC and late meals have similar effects.
Breathing Rate
Healthy adults breathe 12 to 20 times per minute during sleep. Your personal baseline matters more than the normal range. A rise of 2 to 3 breaths per minute across several nights is often an early warning sign of infection — before you notice symptoms.
Nighttime Heart Rate and HRV
Nighttime heart rate typically sits 5 to 10 bpm below your daytime resting heart rate. HRV is higher at night than during the day and shows how well your parasympathetic nervous system is recovering. Read them together: falling heart rate plus rising HRV equals good recovery. More on interpretation in the wearable data quality guide.
Body Temperature and SpO2
Core body temperature drops about 0.5 degrees Celsius overnight — that is a sleep signal. A bedroom that is too warm blocks the drop and degrades deep sleep. SpO2 (blood oxygen) should stay above 95 percent. Repeated drops below 92 percent are a signal to get sleep apnea checked.
Overrated Metrics
Three numbers are heavily marketed by wearable brands but rarely help decisions.
Sleep Score. A single value between 0 and 100 that mixes several raw data points. Every manufacturer uses its own formula. The score is good for motivation but poor for root-cause analysis. When your score drops from 85 to 68 you do not know why — you still have to check the individual values.
Single-night deep or REM percentages. Wrist-worn wearables carry a 20 to 30 percent error rate for stage classification (de Zambotti 2019 and 2024 studies). Individual nights are noise. Only weekly trends become reliable.
Detailed dream-cycle views from wrist wearables. No wearable outside a sleep lab can precisely detect individual REM episodes or dream events. The polished charts with exact stage transitions are algorithmic interpolations, not measurements.
Device Comparison: Which Wearable for Whom
The right device depends on what you want to optimize.
| Device | Strengths | Weaknesses |
|---|---|---|
| Polysomnography (sleep lab) | Gold standard, EEG-based | One-time, expensive, only on suspected diagnosis |
| Oura Ring | ~79 % PSG agreement on stages, small form factor | Battery 5 to 7 days, subscription model |
| Whoop | Strong HRV and heart rate, recovery score | Limited stage detection, subscription required |
| Apple Watch / Garmin | Good trends, many extras | Coarser stage detection, must wear at night |
| Withings Sleep Analyzer | Mattress sensor, no wear required, apnea detection | Home only, one sensor per bed |
The de Zambotti (2019) Oura-PSG study remains the most-cited reference: 96 percent sensitivity for sleep vs. wake, but only 79 percent agreement for stage classification. That means duration and efficiency are accurate, while individual stages are error-prone.
If you already have a wearable, do not hop between devices. Each one has its own baseline and cross-brand comparisons are worthless.
How to Actually Use the Data
Most wearable users stare at single nights and draw wrong conclusions. Four rules make the difference.
Establish a 2-week baseline. Before testing any intervention you need averages for duration, efficiency, deep sleep, REM, HRV and resting heart rate. That baseline is your reference — not the average of other users.
Weekly trends, not individual nights. One bad night is normal. Three in a row is a signal. Always look at the 7-day rolling mean. It filters noise and shows real change.
Correlate with subjective feel. Log a 1 to 10 score every day for your perceived recovery. Compare it with wearable data. If the two diverge, either the sensor is off or you are ignoring a signal from your body.
Test interventions one at a time (N=1). Change only one variable at a time. 2 weeks without alcohol. 2 weeks with magnesium. 2 weeks with a cooler bedroom. Change three things at once and you never know what worked. The full methodology is in the insight-sprint playbook.
Spotting Common Sleep Problems in the Data
Your wearable can reveal patterns that point to concrete issues. Three examples with clear signatures.
Sleep apnea. Repeated SpO2 drops below 92 percent, elevated heart rate spikes at night, fragmented sleep despite adequate time in bed. If you wake up exhausted after 8 hours in bed, snore loudly and your partner notices breathing pauses, book a sleep lab. Apnea is treatable, but untreated it is a risk factor for heart attack and diabetes.
Alcohol effect. Two glasses of wine cut REM by 20 to 30 percent, raise WASO by 20 to 40 minutes and measurably drop HRV. Typical pattern: you fall asleep faster (short latency) but wake up between 3 and 5 AM and lose REM in the second half of the night. The effect is clearly visible in wearables and lasts 24 to 48 hours.
Cortisol dysregulation. Early waking between 3 and 5 AM, elevated heart rate in that window, inability to fall back asleep. The pattern shows up with chronic stress, late evening meals or elevated evening cortisol. If it persists for weeks, a salivary cortisol day profile is worth running.
Evidence-Based Interventions
Six levers with solid evidence. Implement in this order if you want to improve sleep.
1. Light hygiene. 10 to 30 minutes of morning light (outside, no window in between) within an hour of waking. Reduce blue light starting 2 hours before bed — dim lights, night mode or blue-blocker glasses. This synchronizes your circadian rhythm and improves sleep latency and deep sleep.
2. Bedroom 16 to 19 degrees Celsius. Core temperature must drop about 0.5 degrees overnight for optimal deep sleep. A warm bedroom (above 20 degrees) blocks that drop. Lowering the thermostat or changing blankets is the simplest lever with a measurable effect on deep sleep.
3. Caffeine cut-off 8 to 10 hours before bed. Caffeine has a half-life of 5 to 7 hours. An espresso at 3 PM is still 25 percent in your system at 11 PM and measurably cuts deep sleep. Sensitive types should cut off at noon.
4. Avoid (or strongly reduce) alcohol. The strongest REM killer. Even two glasses of wine cut REM by 20 to 30 percent and noticeably raise WASO. If you want to improve sleep, start here — the effect is visible in your wearable within a week.
5. Consistent timing (±30 minutes). Same bed and wake times, even on weekends. Shifts larger than 90 minutes act like a small jet lag and take 2 to 3 days to recover. The cyclic routine playbook helps build consistency.
6. Magnesium glycinate in the evening. 200 to 400 mg about 30 to 60 minutes before bed. Glycinate is the most tolerable form for sleep purposes. More on dosing and forms in the magnesium forms guide.
Integrating Sleep Data Into Your Dashboard
Wearable tracking on its own only gets you so far. The next step is integration with biomarkers and supplement routine. In Lab2go you see sleep data, lab values and supplement log on one timeline — and spot patterns you would never find in isolation.
Correlations worth watching:
- Low ferritin correlates with restless-leg symptoms and fragmented sleep
- Vitamin D deficiency often coincides with poor sleep quality
- High evening cortisol shows up as early waking between 3 and 5 AM
- Low HRV over multiple weeks correlates with elevated inflammation markers
The available plans cover how many data sources you want to sync. For a full stack with wearable import, biomarker trends and supplement log, the standard plan is enough.
Conclusion: Trends Beat Single Nights
Sleep tracking is not an end in itself. It is a tool to make decisions: does magnesium help? How strong is the alcohol effect? Is my new training too intense? You answer those questions with weekly trends, not single nights.
Three steps to start:
- Set a baseline. Track 14 nights without changing anything. Note averages for duration, efficiency, deep sleep, REM and HRV.
- Test one intervention. Caffeine cut-off, bedroom temperature or alcohol abstinence. 2 weeks strictly.
- Compare weekly values. Not night by night — week 1 vs. week 2 vs. week 3.
To go deeper, combine sleep tracking with long-term biomarker tracking and use the insight-sprint method to test interventions cleanly. Check the features and pricing if you want both on one platform.
This article does not replace medical advice. If you suspect sleep apnea, chronic insomnia or daytime fatigue despite enough time in bed, see a sleep physician.
Article FAQ
- Which sleep metric matters most?
- Sleep duration and sleep efficiency are the most robust markers. 7 to 9 hours in bed, with at least 85 percent of that actually sleeping, is the baseline. Individual deep sleep or REM values from wrist-worn wearables carry a 20 to 30 percent error rate. They become meaningful only when you look at consistent weekly trends.
- How much deep sleep per night is normal?
- Deep sleep (N3, slow-wave sleep) makes up 13 to 23 percent of sleep time, roughly 60 to 110 minutes for 7 to 8 hours of total sleep. It declines with age. What matters is your personal trend, not comparisons with others. Alcohol, late exercise and warm bedroom temperature all measurably reduce deep sleep.
- How accurate is my wearable at detecting sleep stages?
- The Oura Ring shows about 79 percent agreement with polysomnography for sleep stages (de Zambotti 2019). Whoop and Apple Watch land at 70 to 80 percent. Individual stage percentages are error-prone, but trends over 2 to 4 weeks are reliable. For medical diagnoses you always need a sleep lab.
- What does a Sleep Score actually mean?
- A Sleep Score is a marketing number that condenses several raw data points into a single value between 0 and 100. Every manufacturer uses its own formula. The score is good for motivation but poor for data-driven decisions. Look at the individual values instead: duration, efficiency, WASO, deep sleep, REM, HRV.
- Why do I wake up at night even without drinking anything?
- Brief wake-ups are normal. Healthy adults wake 4 to 6 times per night without noticing. WASO (Wake After Sleep Onset) under 30 minutes is ideal. Causes of higher WASO: alcohol in the last 6 hours, cortisol dysregulation (typically 3 to 5 AM), bedroom above 20 degrees Celsius, or a restless sleep partner.
- What does an SpO2 drop at night indicate?
- SpO2 (blood oxygen) should stay above 95 percent through the night. Repeated drops below 92 percent suggest sleep apnea, especially combined with snoring and daytime sleepiness. Oura, Apple Watch and Whoop measure SpO2 optically and accuracy is limited, but repeated patterns are a clear signal to book a sleep lab.
- How do I lower my nighttime heart rate?
- A healthy nighttime heart rate sits 5 to 10 bpm below your daytime resting heart rate. If it is higher, your body is working against something: alcohol, a late meal, an infection, stress or hard training. The three strongest levers are cutting alcohol, stopping food 3 hours before bed, and cooling the bedroom to 16 to 19 degrees Celsius.
- How long should I collect sleep data before drawing conclusions?
- At least 14 nights for a baseline before you test interventions. Single nights are too noisy. To evaluate an intervention (for example magnesium, bedroom temperature, caffeine cut-off) run 2 weeks with and 2 weeks without, then compare weekly averages. Single nights will send you down wrong tracks.
Discussion
Community comments coming soon. Until then, we welcome feedback and questions via email.
E-Mail anzeigen
