There is enormous comfort in a printed schedule. Nap at 9:00, nap at 12:30, nap at 3:30, bed at 7:00. It promises that the day has a shape, that the chaos can be tamed with a grid, that if you just hold the line the baby will fall into place. Sleep-deprived parents reach for these schedules the way a drowning person reaches for anything that floats. And for a while, with the right baby on the right week, it might even work.
Then it stops. The baby who napped beautifully at 9:00 last month now lies in the crib babbling and kicking, wide awake, until 9:25 — at which point they are cranky, and the nap is short and fitful. Or they fall apart at 8:30, an hour before the scheduled nap, and you are left bouncing a screaming baby while the clock insists it is not yet time. The schedule, which felt like an ally, has become a thing you are failing to live up to. The problem is not your discipline. The problem is the premise.
A schedule fixes the wrong variable
A by-the-clock nap schedule treats the time of day as the thing that determines when a baby is ready to sleep. But time of day is not what drives an infant toward sleep. Two internal systems do. The first is the homeostatic sleep drive — the build-up of sleep pressure, governed by adenosine, that accumulates with every minute the baby is awake. The second is the circadian rhythm, the internal clock that, once it matures, biases the body toward sleep at particular phases of the day.
A clock schedule is a crude proxy for both, and it is most accurate only when the day before was identical to the template — same wake-up time, same nap lengths, same quality of sleep. The moment reality deviates, the proxy drifts out of true. If the baby woke forty minutes early, then by the time 9:00 arrives they have been awake far longer than the schedule assumes, and they are already overtired. If the previous nap ran long, they reach 9:00 undertired, with not enough sleep pressure to drop off. The grid says one thing; the baby's biology says another; and the baby's biology always wins.
Wake windows track the thing that actually matters
The more reliable anchor is the wake window — the elapsed time since the baby last woke. Anchoring to wake windows rather than clock times means you are measuring the variable that genuinely governs readiness for sleep: how much sleep pressure has built since the last sleep cleared it. A baby who wakes at 6:20 and one who wakes at 7:10 are not ready for their first nap at the same wall-clock time; they are ready after roughly the same amount of awake time. Tracking the window absorbs the day's inevitable variation instead of being broken by it.
This is why two babies of the same age, in the same house, can be on wildly different "schedules" and both be perfectly well-rested. They are running the same wake windows, just offset by their different morning starts. The schedule that matters is relative, not absolute.
The schedule that is always one size too small
There is a second, slower way fixed schedules fail: they go stale. Wake windows lengthen steadily as a baby's brain matures and its capacity to tolerate wakefulness grows. A schedule that was perfectly calibrated at three months is, by five months, asking the baby to sleep before enough pressure has built — and the parent, watching their formerly good napper resist, often misreads it as a regression or a problem rather than a baby who has simply outgrown the spacing.
Because the change is gradual, a printed schedule never tells you when it has expired. It just slowly stops fitting, and the friction it produces gets blamed on the baby. A wake-window approach updates itself: as the windows you can comfortably sustain stretch, the day naturally spreads out, and the transitions from more naps to fewer happen when the baby is ready rather than when a chart says so.
What a flexible day still gives you
None of this means the day should be formless. Babies thrive on predictability, and there is real value in a stable rhythm — a consistent morning wake-up that anchors the whole day, a familiar order of events, a reliable bedtime range. The point is to make that structure responsive rather than rigid. Think of it as a rhythm rather than a timetable: the sequence stays the same, but the exact timing of each nap flexes to meet the baby where they are that day.
In practice this looks like holding the morning wake-up reasonably steady (because the first window of the day sets up everything after it), then letting each subsequent nap be timed off the previous wake-up plus an age-appropriate window, adjusted by what you see in the baby. A short nap means the next window should probably be a touch shorter, because less sleep pressure was cleared. A long, restorative nap can stretch the next window a little. The day breathes instead of snapping.
The exception that proves the rule
By the toddler years, once naps have consolidated down to a single midday sleep and the circadian rhythm is strong and stable, a clock-based nap does become reasonable — a roughly fixed early-afternoon nap and a consistent bedtime. That is precisely because, at that age, the underlying biology has finally become regular enough that the clock is a decent proxy for it. The fixed schedule isn't wrong forever; it is just wrong for the months when a baby's sleep is still developing fastest and varying most. For most of the first year, the clock is following the baby, not leading.
Running the day off wake windows is obviously more demanding than reading numbers off a printed chart — you have to know the right window for your baby's age, track time since the last wake-up, and adjust as they grow. That arithmetic is exactly what Drowsy does in the background. It holds an age-calibrated window, counts the time since your last one-tap log, and shows a live countdown to the next likely nap — then tightens those estimates to your baby's own pattern after a couple of weeks of data, so the day spreads out on its own as the windows lengthen. No stale chart to outgrow. If you'd rather follow your baby than a grid, Drowsy is at drowsy.lumenlabs.works.