The 90-Minute Wave: Working With Your Ultradian Rhythm

Attention is not a tap. It runs in waves — the ultradian rhythm, roughly ninety minutes of rising and then falling alertness, repeating through the day. Working with that shape, rather than against it, is the idea behind planning in cycles instead of in hours.

What a cycle-planned day looks like

Give the ultradian cycle planner a standard eight-hour window starting at nine, with the default 90-minute focus and 20-minute rest:

• 09:00–10:30 focus, rest until 10:50
• 10:50–12:20 focus, rest until 12:40
• 12:40–14:10 focus, rest until 14:30
• 14:30–16:00 focus, rest until 16:20

Four cycles: 360 minutes of focus, 80 minutes of rest, 40 minutes left over. Six hours of actual focused work out of an eight-hour day — which is a good deal more than most people manage, and the reason is the rest, not the focus.

The rest is the mechanism

The instinct on seeing 80 minutes of scheduled rest is to reclaim it. That is the error the whole approach exists to prevent.

The premise is that alertness declines through a cycle and recovers during a break, so the break is what makes the next cycle possible at full strength. Delete it and the following block is not a fresh 90 minutes; it is a continuation of a declining one. The output looks similar on a timesheet and is not similar in quality.

The rest also has to be genuine. Answering email is not rest — it is a different kind of cognitive work at the same desk. What tends to work is a change of mode and posture: walking, standing, something without a screen, ideally something that does not require decisions.

Ninety minutes is a starting point, not a law

The figure is an average and the planner takes it as an input rather than a constant, because individual cycles vary and so does the work.

Run the same eight-hour window at 60 minutes of focus and 15 of rest and you get 6 cycles, still 360 minutes of focus, with only 30 minutes left over. Identical focused time, arranged differently — shorter blocks, more of them, less unallocated slack.

Which is better depends on the work, not on the theory. Deep problems with a long warm-up favour longer blocks; work that arrives in discrete pieces favours shorter ones. The planner's usefulness is that it makes the trade visible before you commit a day to it.

The leftover figure is doing real work

Notice that the eight-hour day leaves 40 minutes unallocated. That is not waste — it is the honest remainder after whole cycles are placed, and it is exactly the slack that real days consume: the meeting that runs long, the conversation in the corridor, the twenty minutes lost to a badly-timed interruption.

Plans that allocate every minute fail on contact with an ordinary Tuesday. A plan that says "four cycles and forty minutes spare" survives one.

Where this fits, and where it does not

Cycle planning suits work you control the shape of: writing, analysis, coding, study, design. It suits it much less well when your day is structured by other people's meetings, by customers arriving, or by anything that cannot be told to wait ninety minutes.

For a fragmented day, the shorter cadence of the focus session planner fits the available gaps better. For deciding how much of your week can realistically be this kind of work at all, the deep work budget planner is the blunter and more useful instrument.

And a caveat on the theory: the ultradian rhythm is a real and well-described phenomenon, but the idea that it should govern a work schedule is a practical heuristic rather than a settled finding. The planner is a scheduling aid, not a claim about your neurology, and it is not medical advice.

Trying it without rearranging your life

The cheapest experiment is one cycle. Take a single 90-minute block tomorrow morning, protect it properly — notifications off, door shut, one task — and take the 20 minutes afterwards away from the desk.

Then notice two things: whether the block held up to the end, and whether the work after the break started better than it usually does. If the block faded at sixty minutes, shorten the cycle. If the break was hard to take, that is worth knowing too, because it is the part people quietly delete and then wonder why the second half of the day is worse than the first.

What actually breaks a cycle

The plan is easy to write and harder to hold, and the failures are fairly predictable:

• Notifications. A single interruption does not cost you the seconds it takes to read — it costs the time to get back to where you were, which is substantially longer and is why a "quick look" is not quick.
• Starting without a defined task. A block that opens with "work on the report" spends its first twenty minutes deciding what that means. Decide the night before; the cycle should start with the first real action already chosen.
• Meetings landing mid-block. A 45-minute meeting at 11:00 does not remove 45 minutes from an 09:00 cycle; it removes the whole cycle. Cluster meetings at cycle boundaries if you have any say in it.
• Skipping the rest because the block went well. The most tempting failure. Momentum feels like a reason to continue and is usually the last of the wave rather than the middle of it.

Fitting cycles to a real calendar

Almost nobody has an unbroken eight-hour window. What most people have is a morning that is relatively free and an afternoon that is not, or the reverse.

That is worth planning around directly. Feed the planner the window you actually control rather than the whole day: a 240-minute morning yields two clean cycles finishing at 12:40, with 20 minutes spare. Two protected cycles is a genuinely good day's focused work, and it is achievable in a way that "four cycles" often is not.

It is also worth checking what an extra hour buys before you fight for it. A 300-minute window — a full hour longer — still fits only two cycles, because the third needs another 110 minutes. That hour is worth having for other reasons, but it does not add a cycle, and knowing that in advance saves a pointless negotiation.

Matching the block to the task

One last practical point. Cycles are a container, and not everything fits the same container.

Work with a long warm-up — anything where you have to reload a lot of context before you can be useful — wastes proportionally more of a short block, so it belongs in the longer cycles. Work that is naturally chunked, like reviewing a set of items or clearing a queue, survives interruption better and can go in the leftover time or the shorter cadence.

Putting the wrong work in a protected cycle is a quieter kind of waste than not protecting it at all, and a good deal more common.