Sleeping better: what actually works, in order

The “sleep” aisle is one of the most profitable in online retail, and one of the most misleading. It sells twelve-plant blends dosed at a thousandth of the clinical trials, “ergonomic” pillows with no data, and sensors claiming to score what sleep medicine still measures in a laboratory. Meanwhile, the two interventions with the best effect-to-effort ratio are not for sale: removing light, removing noise.

This comparison applies our usual grid: does the claimed result exist in humans, does the product’s dose match the trials, who must not take it, and what does a night of use cost. We applied it to a dozen common interventions. Seven survived: three that cost nothing, and four supplements.

The order matters more than the product

The biggest buying mistake in this aisle is not choosing the wrong supplement: it is starting with the supplement. Sleep obeys three systems, and two of them respond to the environment, not to a capsule.

• Sleep pressure. Adenosine accumulates in the brain during waking and creates the urge to sleep. Caffeine does not remove it: it blocks the perception of it, for hours.
• The circadian clock. It is set mainly by light. Morning light advances it, evening light delays it. It is the most powerful lever, and the cheapest: it is free.
• Arousal. A system that cuts sleep when something demands alertness: noise, heat, pain, rumination. That is where darkness, silence and temperature act.

Complete darkness: the most underrated lever

It is not an attractive subject, and it is the best supported by data on this whole page. Night-time light, even at low intensity, suppresses melatonin secretion and degrades sleep architecture: a few dozen lux — the order of magnitude of a night light, a street lamp behind a light shutter or a router LED — is enough to produce a measurable effect. Eyelids are not opaque: light passes through them and reaches the retina even with your eyes shut.

The good news is that it is not for sale. Walk round your bedroom once, at night, lights off, after five minutes of adaptation:

• The indicator lights. Router, charger, TV on standby, power strip, smoke detector. A piece of opaque tape on each, or the device unplugged. It is the action with the best effort-to-effect ratio on this whole page.
• The phone out of the bedroom. It settles the light and the temptation at the same time. If you use it as an alarm, that is the one purchase this page concedes — and an old clock will do.
• The window. A lined curtain you already own, folded double, blocks more light than a thin blind. A piece of cardboard cut to size behind the curtain works too, long enough to check whether darkness changes anything for you.
• Noise. Same logic: move the bed against an internal wall, close the doors in between, turn off notifications. A fan or a pink noise app masks irregular noise, which is the most disruptive kind.

Try it for a week. If complete darkness and silence change nothing for you, you will have learned that for zero euros — and the supplement question will then be asked on sound footing.

The best supplement: magnesium bisglycinate

If one supplement deserves a place in an evening routine, it is this one — with one clarification that changes everything: it does not make you sleep. It corrects an insufficient intake, very widespread in Western diets, which comes with cramps, muscular tension and more restless sleep. Randomised trials on sleep quality exist, but they are few and small: we therefore rate this evidence as emerging, not solid.

Two label traps, both systematic:

• Salt versus element. “500 mg of magnesium citrate” does not mean 500 mg of magnesium. Look for “elemental magnesium”: the useful target is around 200 to 350 mg a day, dietary intake included.
• The form. Oxide is the cheapest and most laxative, with poor absorption. Bisglycinate, chelated to glycine, is the most comfortable in the evening — and glycine has its own small body of sleep data.

Note: the European claims authorised for magnesium concern normal nervous system function, muscle function and the reduction of tiredness. None states that it “makes you sleep”. When a label promises that, it is outside the rules.

Melatonin, misunderstood and misdosed

It is the most badly used product in the category. Melatonin is not a sleeping pill: it is a clock signal. Its best established effect is shifting circadian phase and shortening time to fall asleep — at a low dose.

The documented effect is obtained between 0.5 and 1 mg; above 2 mg, sleep does not improve further and residual grogginess on waking increases. The market, meanwhile, sells 5 and 10 mg tablets in volume. In France, food supplements stay below the 2 mg threshold, which is rather good news. When you take it matters as much as the dose: 30 to 60 minutes before the target bedtime for falling asleep, 3 to 5 hours before to genuinely advance the clock — eastward jet lag, for instance.

The low-cost options that hold up

Glycine, 3 g before bed

A faintly sweet amino acid, at about 11 cents a dose. Small randomised trials report sleep judged more restorative and a fall in core temperature — precisely the physiological signal that accompanies falling asleep. Small numbers, so emerging evidence; but the cost and the tolerability make it a reasonable thing to try.

L-theanine, 200 mg

This is the amino acid in green tea. Its territory is not really sleep: it is a mind that will not stop. Trials show a reduction in situational anxiety without sedation, which makes it an interesting option when falling asleep is blocked by rumination rather than by a lack of sleep pressure. A bonus: 100 mg with your morning coffee smooths the caffeine peak.

Should you measure your sleep?

It is the question that comes up most, and the honest answer has two parts. A sleep sensor can do one genuinely useful thing: show you how irregular your bedtimes are, which is often the real problem. For everything else, it overstates what it can do.

Against polysomnography, these sensors estimate total sleep time reasonably well but remain imprecise on the stages — the “deep sleep” displayed is an estimate, not a measurement. And there is a documented side effect: orthosomnia, the performance anxiety where the score degrades the sleep it claims to assess.

Above all: what a sensor will end up telling you — go to bed at a more regular hour, make it dark, lower the temperature, stop the coffee earlier — is already written on this page, and you can apply it tonight. If you want to track your consistency, one line a day in your phone’s notes (bedtime, wake time) does the same job. This site does not compare devices and will not recommend any: that is not our subject.

The comparison table

Cost per night estimated over the product’s lifetime or number of doses. Prices change often: only the one shown on Amazon at the moment you buy counts.

What we ruled out, and why

• Eight-plant “sleep” blends. Every active ingredient sits far below the doses studied. A good active ingredient, underdosed, is still an ineffective product.
• 10 mg melatonin sprays. A counterproductive dose, an increased risk of residual grogginess, and outside the framework for supplements sold in France.
• “Sleep” CBD. Inconsistent data, real contents varying widely between batches, high price. We recommend nothing in this category for now.
• All the hardware in the sleep aisle. Masks, earplugs, “ergonomic” pillows and mattresses, trackers, dawn simulators. This site only compares supplements and other consumables: equipment is not our subject, and for darkness as for silence, the free version described above does the job.
• Blue-light glasses. Results too inconsistent against a free and more effective alternative: lower the light and stop earlier.

The seven-night protocol

1. Nights 1–2 — Darkness and silence. Tape over the indicator lights, phone out of the bedroom, curtain folded double, door closed. It is the only step whose effect shows immediately, and it costs nothing.
2. Night 3 — Caffeine. Last coffee before 1 pm if you go to bed around 11 pm. The half-life is around five to six hours: an espresso at 5 pm is still working at midnight.
3. Night 4 — Temperature. Bedroom between 18 and 19 °C. Falling asleep follows a drop in core temperature; a room that is too warm blocks it.
4. Night 5 — Morning light. Ten minutes outside within an hour of waking, without sunglasses. In winter, go out anyway: an overcast sky at 8 am far exceeds the brightest indoor lighting.
5. Night 6 — The fixed wake-up time. The same hour seven days a week, weekends included. It is the anchor of the clock, and the parameter most often sabotaged.
6. Night 7 — One supplement. If one specific point remains, add it: magnesium for muscular tension, theanine for rumination, melatonin for a shifted schedule. One at a time, two weeks, then judge.