Why the snooze button doesn't work

What the first ten minutes look like

Waking is not a switch. PET imaging of people woken from stage 2 sleep shows blood flow returning first to the brainstem and thalamus. That is enough to make you conscious. Blood flow to the front of the cortex keeps climbing for fifteen more minutes. That is the part that weighs consequences. The authors put the lag plainly. Consciousness comes back fast. Alertness follows twenty to thirty minutes later (Balkin et al., Brain, 2002).

A 2019 study scanned thirty-four people five minutes after waking them from a nap. Sleep-specific patterns were still present in their brain activity. In those woken from deep sleep, the networks that keep focused and idle states apart had stopped separating (Vallat et al., NeuroImage, 2019). That was an afternoon nap, not a morning alarm. The direction is clear. The numbers do not transfer.

Is snoozing actually bad for you

Nobody knows yet. The evidence is split. Most of what you have read on this is stated with more confidence than anyone has earned.

Thirty-one habitual snoozers were monitored overnight in a Stockholm lab. Half an hour of snoozing cost them about six minutes of sleep. Cognitive performance on rising was not impaired. The chance of being pulled straight out of deep sleep went down (Sundelin, Landry and Axelsson, Journal of Sleep Research, 2024). The survey arm of the same paper, with 1,732 people, found the opposite association. Snoozers reported more morning drowsiness and less sleep overall.

A Hiroshima lab study points the other way. An alarm every five minutes kept people in drowsy N1 sleep for longer. It prolonged sleep inertia compared with a single alarm (Ogawa, Kaizuma-Ueyama and Hayashi, 2022). Ten participants. Ten is not many.

A 450-person study found snoozers had a higher resting heart rate and lighter sleep before waking. They did not sleep less than non-snoozers. They did not feel sleepier, and they did not nap more (Mattingly et al., Sleep, 2022).

It is also completely normal. Sleep-app users logged more than three million nights. 55.6% of them ended with a snooze alarm, pressed on average 2.4 times. About eleven minutes a morning (Robbins et al., Scientific Reports, 2025).

What does not fix it

This part is worth knowing before you buy anything, including this app.

A 1999 study followed people through the first four hours after their usual wake time. One group got up, showered, ate breakfast and sat in ordinary room light. The other stayed in bed in near darkness eating small snacks. Sleep inertia faded at the same rate in both groups. Getting up made no measurable difference (Jewett et al., Journal of Sleep Research, 1999).

Exercise has been tested too. A thirty-second sprint immediately on waking made people feel less sleepy. It produced no improvement in cognitive performance at all (Kovac et al., Physiology & Behavior, 2021). Fifteen participants, woken from a two-hour nap at 2am.

Alarm sound has been reviewed systematically. Twelve studies were not enough to name a best kind of alarm (McFarlane et al., Clocks & Sleep, 2020).

Sleep inertia fades as time awake increases. It is worse after sleep loss (Hilditch and McHill, Nature and Science of Sleep, 2019). Little else about it is settled. It takes the time it takes.

What is left

One variable. Whether you spend those twenty minutes upright or asleep.

Snoozing does not buy rest. It buys a second attempt at a decision, made by the same brain that got it wrong two minutes ago. Then a third attempt. Robbins found the average snoozer takes 2.4 of them.

Ember removes the decision. Silence the alarm and a wake-up check starts. Scan a code you left in the kitchen. Finish a tile puzzle. Walk four hundred steps. The alarm keeps coming back until the check is done.

Ember does not shorten sleep inertia. Nothing does. The claim is narrower. By the time the check is finished you are out of bed. The twenty minutes pass on your feet.