Sleep:
Ten things
that actually matter

The single highest-impact variable. The vast majority of adults need seven to nine hours of actual sleep, not time in bed, but sleep. The proportion of the population that can genuinely function on six or fewer hours with no performance or health impairment is estimated at under 3%. You are almost certainly not in that group, and the fact that you feel fine on six hours does not mean you are performing at your ceiling. It means you have adapted to a lower baseline and can no longer accurately judge the gap.

Your circadian rhythm is the master regulator of sleep quality, hormonal timing, and metabolic function. A consistent wake time, including weekends, is the single most powerful circadian anchor available to you. Social jet lag (sleeping in at weekends to compensate for weekday restriction) is associated with worse metabolic markers than shift work in some studies. It does not repay sleep debt; it compounds it by destabilising the rhythm.

Light is the primary zeitgeber: the external signal that sets your biological clock. Bright light exposure within 30–60 minutes of waking (ideally natural light, even on a cloudy day) advances your circadian phase and improves sleep onset that night. Conversely, bright artificial light in the two hours before bed, particularly blue-spectrum light from screens, suppresses melatonin by up to 50% and delays sleep onset by 60–90 minutes in controlled studies. The effect is dose and wavelength dependent; overhead LEDs are worse than lamps, screens are worse than ambient light.

Core body temperature must drop by approximately 1–1.5°C for sleep initiation and deep sleep maintenance. Room temperature is one of the most consistently underestimated variables in sleep quality. The optimal sleep environment is typically 16–18°C (60–65°F). Warmer rooms reduce slow-wave sleep (deep sleep) and REM duration. A warm bath or shower 1–2 hours before bed is counterintuitively useful: it draws blood to the periphery, which accelerates core temperature drop once you get out.

Caffeine works by blocking adenosine receptors: adenosine is the chemical that builds sleep pressure throughout the day. It does not reduce your adenosine load, it just masks it. The pressure accumulates and is waiting when the caffeine clears. With a half-life of five to seven hours (longer in some individuals, and extended by factors including oral contraceptives and liver enzyme variants), a 200mg coffee at 3pm leaves 100mg active at 8–10pm. This impairs sleep architecture even if you have no subjective difficulty falling asleep.

Alcohol is sedating, which is why people conflate it with sleep quality. It is not the same thing. Alcohol disrupts REM sleep in the second half of the night, increases sleep fragmentation, elevates heart rate during sleep, and impairs the glymphatic clearance process (the brain's overnight waste-removal system). Even moderate intake, two units, measurably degrades sleep architecture in studies using objective measurement. The sedation is real; the restorative value is not.

Exercise is one of the most effective long-term interventions for sleep quality. The timing caveat is real but often overstated. The primary concern is intense training within two to three hours of bed, which elevates core temperature, raises cortisol, and increases sympathetic nervous system activity, all of which oppose sleep initiation and deep sleep. For most people with a consistent training schedule and adequate fitness, a 7–8pm session will not be catastrophic. But if sleep onset is slow or sleep quality is poor, training timing is a reasonable variable to adjust before medication or supplements.

Sleep is not uniform. It cycles through stages, with slow-wave (deep) sleep dominating in the first half of the night and REM dominating in the second half. Deep sleep drives physical recovery, growth hormone release, immune function, and metabolic regulation. REM drives memory consolidation, emotional processing, and cognitive performance. Cutting sleep short, even by one hour, disproportionately removes REM because of its back-loaded distribution. This is why six hours feels roughly functional but is not the same as eight.

Sleep onset difficulty is rarely a physical problem. It is almost always a nervous system state problem. The autonomic nervous system needs to shift from sympathetic (activated) to parasympathetic (rest) mode before sleep is accessible, and that transition takes time. Checking email at 11pm, engaging with stimulating or anxiety-provoking content, or having difficult conversations close to bed all extend that transition. The research on pre-sleep cognitive arousal and sleep latency is consistent: a dedicated wind-down routine reduces the time to sleep onset and improves subjective sleep quality.

Subjective sleep quality and objective sleep quality are often poorly correlated. People routinely overestimate total sleep time and underestimate the number of awakenings. Consumer wearables (Garmin, Apple Watch, Oura Ring) are imperfect but useful for tracking trends in total sleep duration, sleep efficiency, resting heart rate, and HRV, all of which correlate with recovery status. The value is not in the absolute numbers, which can be unreliable, but in your personal baseline and deviations from it. A resting HR 8bpm above your baseline the night after drinking tells you something. Ignoring that data because it is uncomfortable is a choice.