If-then plans that actually work
Implementation intentions are the best-supported technique in this field, and the popular version of them is still wrong. The format, the honest evidence, the craft of choosing a cue, and what they cannot do.
If there is one technique in the whole of goal research that has earned its reputation, it is this one. It is also routinely oversold, so this page tries to do both things at once: explain why it works, and be straight about how much smaller the effect is in your life than in the abstracts.
The format is: if situation X arises, then I will do Y. It comes from Peter Gollwitzer's work on implementation intentions in the 1990s, and the distinction it draws is the important part.
Goal intentions and implementation intentions
A goal intention states an outcome: I intend to exercise more. I intend to read more. I intend to be better at keeping in touch with people.
An implementation intention states a situation and a response: if it is Tuesday and I have finished dinner, then I will put on my running shoes. The difference looks trivially small. It is not, and the reason has nothing to do with motivation.
A goal intention leaves the decision — when, where, whether now — to be made in the moment. And the moment is always the worst possible place to make it. You are tired, the alternative is comfortable, and the goal is competing against something immediately pleasant. An implementation intention makes the decision once, in advance, when none of that is true, and then hands the moment a cue rather than a choice.
The proposed mechanism is that the cue becomes linked to the action closely enough that encountering the cue brings the action to mind without deliberation. Whether that is best described as automaticity is still argued over. The practical claim is more modest and holds regardless: you have moved the decision from a bad time to a good one.
What the evidence actually says
Gollwitzer and Sheeran's 2006 meta-analysis pooled a large number of studies and reported an average effect in the medium-to-large range. That is a genuinely strong result for this kind of research, and it is why the technique appears in almost every serious summary of goal pursuit.
Now the qualifications, which are not decoration.
- Many of the studies are small, and small studies produce noisy estimates that are individually unreliable even when the pooled average is real.
- A great deal of the outcome data is self-reported behaviour rather than measured behaviour, and people over-report doing the thing they told a researcher they intended to do.
- Publication bias inflates averages in this literature as in every other. Studies that found nothing are less likely to have been published, so the pooled figure is an upper bound rather than a best guess.
- Follow-ups are often short — weeks, not years. Whether an if-then plan is still working in month nine is largely unstudied.
- The effect is much weaker for behaviours that are complex, unpleasant, or competing against a strong existing habit. Attending a scheduled screening appointment is a very different proposition from not reaching for your phone.
The honest summary: this is a real effect, it is probably smaller in your life than the abstracts imply, and it is still the best-evidenced single thing you can do. Both halves of that sentence are true and neither cancels the other.
The craft: choosing a cue
Most if-then plans that fail do so because the cue was badly chosen, not because the technique does not work. Four rules cover nearly all of it.
The cue must actually occur
This sounds obvious and is broken constantly. A cue has to be an event that reliably happens, that you will notice when it happens, and that you can identify unambiguously. A completed daily event is ideal: after I put the kettle on in the morning, after I close my laptop at the end of work, after I get off the bus at my stop.
Feelings make terrible cues. "If I feel motivated, then I will write" is not a plan; it is a description of the problem. Clock times are weaker than they look, because a time is only a cue if you are reliably somewhere that shows you the time and reliably interruptible at it. If you already ignore your 7am alarm on four days out of seven, seven o'clock is not functioning as a cue in your life.
One cue, one action
If-then plans degrade quickly when overloaded. "If it is morning, then I will meditate, journal and stretch" is three actions behind one cue, and what happens in practice is that the cue triggers a small negotiation about which of the three you have time for — which is exactly the deliberation the format was meant to remove.
One cue, one action. If you want three things, you need three cues, and you should probably start with one.
Anchor to an existing routine
The most reliable cues are things you already do without deciding: the kettle, the shower, locking the front door, sitting down at your desk, brushing your teeth. Anchoring borrows the reliability of something already stable, rather than requiring you to build two new things at once.
This is also why "when I have time" fails by construction rather than by bad luck. It names no situation. There is no moment at which it becomes true, no observable event that fires it, and so there is nothing for the plan to attach to. It is a goal intention wearing an if-then costume.
Write the coping variant too
The single most useful addition is a plan for the failure, not just the success. "If I have missed two days, then I do the smallest version" is a coping implementation intention, and it addresses the thing that actually ends most practices — not the missed day, but the collapse that follows it.
The smallest version has to be genuinely small. One page, not one chapter. Five minutes, not the full session. Its function is not progress; it is to keep the cue alive so the practice is still there when your circumstances improve.
What this looks like written down
- Take one goal. Not three.
- Name one action that would move it — small enough that starting requires no further decision.
- Find an event in your existing day that reliably precedes a moment when that action is possible.
- Write the sentence in full: if [event], then I will [action]. Write the "then" as a physical thing you do, not as an intention to try.
- Write the coping line: if I have missed two days, then I will [smallest version].
- Leave it for a fortnight before judging it. If it never fired, the cue was wrong, not you — change the cue rather than the amount of resolve.
In Northstar this maps onto steps you write yourself, each marked once, daily or weekly, with a flag for what belongs to today. The nextAction view names one action at a time rather than presenting a list, which is the same principle applied to the app: a single thing, decided in advance.
It is worth being clear about a limitation here that is also the point. Northstar sends no notifications — there is no push mechanism at all — so nothing in your day will buzz to remind you. That means the cue genuinely has to be something in your life: the kettle, the commute, the closed laptop. An app that reminds you can attach a habit to itself, and when you eventually mute it, the habit goes with it. An app that cannot remind you forces the cue to live where it will still exist next year. That is a real inconvenience, and it is also the more durable arrangement.
What if-then plans cannot do
They do not create wanting
An implementation intention converts an existing intention into action more reliably. It does not manufacture the intention. If you do not actually want the goal — if it belongs to someone else's expectations, or to a version of you from four years ago — the plan will fire, you will notice you have no interest in complying, and the useful discovery is about the goal rather than the plan.
They break when the environment changes
This is the failure mode nobody warns you about. Your cues are made of your circumstances. A new job destroys every cue you had: the commute is gone, the desk is different, the kettle is in another room, the end of the working day is at a different hour. Move house, change shifts, have a child, and the same thing happens.
What follows is usually misread. The practice stops, and you conclude you have lost your discipline. In fact your discipline is unchanged and your cues have been deleted. The two look identical from the inside, which is why the diagnosis matters.
They can make you rigid
A well-drilled cue-response pair does what it is told, including when the situation has changed and it should not. People carry on with a plan that has stopped serving the goal because the plan is running smoothly and running smoothly feels like success. Periodically ask whether the action still leads where you wanted, rather than only whether it fired.
Rewriting cues after a life change
Because cues are environmental, they need explicit maintenance after any disruption. This takes about ten minutes and is almost never done.
- List the plans you had. Most people find that half of them stopped firing weeks ago without being noticed.
- For each, identify which part broke — the cue no longer occurs, occurs at a useless time, or occurs somewhere the action is impossible.
- Find the nearest equivalent event in the new arrangement. It rarely maps cleanly, and the honest answer is sometimes that a plan has no home in the new life and should be dropped.
- Cut the number of plans while the new routine is still forming. Two working cues beat six aspirational ones.
- Give it a fortnight, then check which ones actually fired. A review over seven or thirty days will show whether anything was logged at all, which is the ground truth here.
Doing this deliberately after a disruption is the difference between a practice that survives a job change and one that quietly ends at it. The technique is not fragile in itself; it is only as stable as the day it is attached to, and days change.
Frequently asked questions
What is an implementation intention?
A plan in the form "if situation X arises, then I will do Y" — it specifies a cue and a response rather than an outcome. The point is that the decision about when and whether to act is made in advance, calmly, rather than in the moment when you are tired and the alternative is more appealing.
Do if-then plans really work?
Yes, with qualifications. Gollwitzer and Sheeran's 2006 meta-analysis reported a medium-to-large average effect, which is strong for this field. But many studies are small, outcomes are often self-reported, publication bias inflates the average, follow-ups are short, and the effect is much weaker for complex or unpleasant behaviours. It is a real effect that is probably smaller in your life than in the abstracts.
Why does "I will do it when I have time" not work?
Because it names no situation. A cue has to be an event that actually occurs and that you can identify when it occurs. "When I have time" has no such moment, so there is nothing for the plan to attach to — it is a goal intention in an if-then costume.
What should I do when a life change breaks all my cues?
Rewrite them deliberately rather than trying harder. Cues are made of your circumstances, so a new job or a house move deletes them, and the resulting collapse gets misread as lost discipline. List the plans, find which part broke, anchor to events in the new routine, and keep fewer of them while things are still settling.
Try it on a real goal
Northstar cuts a goal into phases and phases into steps you write yourself, then names one next action a day. Its statistics come from your own entries, and it shows an empty state rather than a placeholder number. Everything stays on the device — no account, no sync, no analytics, no notifications. Free, with no in-app purchases.