How detailed should a goal plan be?

Granularity is a dial, not a virtue. What under-planning costs, what over-planning costs, and a resolution rule for deciding where to stop cutting.

Most advice about planning tells you to break things down. Very little of it tells you when to stop. That omission matters, because detail is not a virtue that increases monotonically — it is a dial with a cost at both ends, and the sensible setting moves depending on how much of the work you have already done.

This page is about the dial itself. If you want the mechanics of cutting a goal into layers, that is a separate question and covered elsewhere. Here the question is narrower: given that you are breaking something down, how far down should you go before the breaking-down becomes the problem?

The cost of too little detail

The symptom of under-planning is not confusion. It is a small, repeated hesitation that you probably do not notice as a cost.

You look at your list. It says "work on the presentation". You know roughly what that means, so you have not failed to plan in any obvious sense. But before you can start, you have to decide something: which section, what format, whether to gather figures first or draft the argument first. That decision is small. It is also, reliably, the exact point at which people open a different tab.

The friction in doing things is very often the deciding, not the doing. A step that still contains a decision is a step that has to be planned again at the moment of execution — usually in the morning, usually tired, usually with less patience than you had when you wrote it. The entire practical value of writing a step in advance is that you were calmer then than you are now. A step that defers the decision throws that advantage away.

So the first test is not "is this small" but "is this decided". "Draft the three bullet points for the second section" is decided. "Work on the presentation" is not, and it will cost you a moment of negotiation every time you look at it.

The cost of too much detail

The other end of the dial is more comfortable, which is why more people end up there and fewer people recognise it.

Planning as a substitute for work

Building a plan feels like progress. It produces an artefact, it gives an immediate sense of control, and it requires none of the discomfort of the thing itself. An afternoon spent producing a beautifully structured hundred-item breakdown is an afternoon in which nothing happened, but it does not feel that way, and the feeling is the trap. Nobody procrastinates by doing something that feels like procrastination.

Maintenance that costs more than the task

Every item in a plan has an upkeep cost. It has to be read, re-read, reordered when things change, and eventually deleted. A plan with forty live items spends your attention on administration at a rate that can genuinely exceed the work it describes. The test is simple and slightly brutal: if you spend more time arranging the structure than executing it, the structure has become the activity.

Brittleness

This is the cost people miss entirely. A chain of fifteen dependent steps assumes that steps one through fifteen were all correct when written. If step four turns out to be wrong — the tool does not work the way you thought, the person you needed is unavailable, the approach is a dead end — then steps five to fifteen are not merely delayed. They are, in many cases, void, because they were built on an assumption that has just failed.

The longer the chain, the more likely it contains at least one wrong link, and the more work is destroyed when it breaks. Detail multiplies your assumptions, and assumptions are what fail.

The resolution rule

Here is the rule that resolves both ends: plan to the resolution at which your uncertainty starts, and not further.

Detail what you already know how to do. Leave a placeholder where you would be guessing. That is the whole of it, and it explains why the correct depth is inversely proportional to how much of the work you have already done. At the start of an unfamiliar project you know almost nothing, so almost everything should be a placeholder. Three weeks in, the next fortnight can be specified precisely, because you have learned what the work actually consists of.

In practice this produces a plan with an uneven texture, which looks untidy and is correct. The near part is fine-grained and decided. The far part is a handful of named headings with nothing under them. The temptation to fill in the far part is strong, because a plan with gaps looks unfinished — but the gaps are honest, and filling them with guesses does not make the guesses better. It just makes them harder to abandon.

Northstar is arranged around this shape. Phases sit as titles until you reach them, and you write steps only for the one you are in. Steps carry a frequency — once, daily or weekly — and a rough estimated time, so a repeating step does not need re-typing and a one-off does not need to be dressed up as a project.

Time estimates: a calibration tool, not a commitment

Attaching a time estimate to a step is worth doing, but not for the reason people assume. It is not a promise, and it is not a scheduling input. It is a measuring instrument pointed at yourself.

People are consistently over-optimistic about how long their own future tasks will take. This is well documented under the heading of the planning fallacy, and it is one of the more robust findings in the area — though it is worth saying that its size is debated, that a good deal of the effect may be motivational rather than purely cognitive, and that the classic demonstrations use student projects and short horizons that do not obviously generalise to a six-month piece of work with a family and a job around it.

What survives all that hedging is modest and useful: your first estimate is probably low, and you can find out by how much. Write twenty minutes, do the thing, notice that it has never once been twenty minutes. After a fortnight you have a personal multiplier, which is worth more than any published figure because it is about you and this kind of work.

Treat a missed estimate as data, not as a failure. The estimate was an experiment; it returned a result.

How to tell over-planning from preparation

Some planning is real work. Drawing up a research protocol, writing an outline before a chapter, listing the parts you need before you start a build — these are not procrastination, and telling people to "stop planning and start doing" is bad advice when the planning is the doing.

The tell is a single question: does the artefact you just produced change what you do tomorrow?

• An outline that determines which paragraph you write first: preparation.
• A colour-coded schedule for the next six months, with no change to tomorrow: over-planning.
• A list of three unknowns you need to resolve before proceeding: preparation, and probably the most valuable hour of the week.
• A restructured tagging system for a project you have not started: a different hobby.
• A decision made in advance that removes a choice from Tuesday morning: preparation, by definition.

Apply the question at the moment you finish, while the satisfaction is still fresh. That satisfaction is not evidence of anything — it is generated equally by useful and useless plans, which is precisely why it is unreliable.

The honest note on implementation intentions

The strongest evidence for planning in detail comes from research on implementation intentions — specifying in advance when, where and how you will act. The meta-analytic support is decent, and it is one of the better-established effects in behavioural research.

But it is narrower than the popular summary suggests. The effect is strongest for simple, single, well-defined actions: take the pill, attend the appointment, do the one exercise. It is smaller and considerably less reliable for complex multi-stage projects, where the number of things that can intervene grows and any one plan covers a smaller share of the work. Much of the literature is also short-horizon laboratory work with self-reported outcomes and modest sample sizes, and publication bias tends to inflate the averages you see quoted.

So "plan in detail, it is scientifically supported" does not extrapolate cleanly to a six-month project. What the evidence supports is deciding one specific action in advance. What it does not support is a fifteen-step chain covering next spring. Those are different claims that borrow the same citation.

A working setting for the dial

1. Name the whole shape of the goal once — a few phases, headings only. This takes a minute and is not planning in the costly sense.
2. Detail only the current phase, to the point where each step is decided rather than merely small.
3. Stop at the first thing you would be guessing about. Write the guess down as an open question instead of as a step.
4. Give steps a rough time estimate, and treat every one as a hypothesis about yourself.
5. When something turns out to be wrong, rewrite from that point rather than repairing the chain. A short plan is cheap to rewrite, which is the main argument for keeping it short.
6. Review by comparing what you intended against what was actually completed. Northstar's seven- and thirty-day views do this from your own entries — completion average, days logged, and how much of your recorded effort was attached to a goal at all.

That last comparison is the only real calibration you get. A plan that is consistently over-specified shows up as a large gap between intended and completed, week after week, with the same items sliding forward. A plan that is under-specified shows up differently: as days where you logged something but it was not attached to any goal, because when the step did not tell you what to do, you did something else instead.

What the right level of detail cannot do

It will not make an uncertain project predictable. Some work is genuinely exploratory, and the correct plan for it is short by necessity — three moves and a review point. Adding detail to uncertainty does not reduce the uncertainty; it just hides it behind something that looks like a schedule.

It will also not tell you whether the goal is worth doing. That question sits above the plan, and a well-calibrated breakdown of something you do not want is simply a clearer view of a thing you are not going to do. Finding that out is a legitimate result, and it is usually cheaper to discover after two weeks of real work than after an afternoon of planning.

Frequently asked questions

Related guides

• How to break a big goal into steps you can do this week
• If-then plans that actually work
• Why annual goals collapse by February

All guides