The habit loop: cue, routine, reward

Draw three boxes. Label them cue, routine, reward. You have the most famous cartoon in popular habit writing. Charles Duhigg put that cartoon in The Power of Habit and a lot of people, including people who build apps, started talking as if the cartoon were a protocol.

The cartoon is useful. It is also incomplete. A loop drawn on a napkin does not tell you how long automaticity takes, what to do when the reward is vague, or why the same routine fires in the kitchen and dies in a hotel. For those questions you need the less viral papers: Wood and Neal on context, Lally on time, Gollwitzer on if-then plans, Webb and Sheeran on the gap between intending and doing.

A diagram that travelled further than the data

Duhigg is a journalist who synthesised other people's work into a story the public could carry. That is a legitimate job. It is not the same job as running a trial. When a product page says "scientifically designed around the habit loop," ask whether the science is the cartoon or a cited result.

What the three boxes get right

Behaviours that repeat tend to have a starting signal, an action, and some consequence that makes the action worth selecting again. A phone buzz is a cue. Opening the feed is a routine. Novelty or social contact is a reward. You do not need a laboratory to see that pattern in your own evening. The box model stops you from treating a habit as a mysterious trait and starts you treating it as a sequence you can interrupt.

Fogg's Tiny Habits model is a cousin, not a clone. He talks about a prompt, an action sized to current ability, and a tiny celebration. The celebration is a constructed reward: you are teaching the brain that the repetition counts. That is closer to coaching than to Duhigg's reported loops, and it is honest about how small the first routine should be.

What the boxes hide

They hide instability. If the cue is "when I feel motivated," you do not have a cue. You have a mood. Wood and Neal argue that habits are tied to recurring contexts: places, prior actions, other people, times of day. The cue is often the room, not a feeling. The boxes also hide time. Lally and colleagues (2010, European Journal of Social Psychology) found a median near 66 days to an automaticity plateau, with a range from 18 to 254. A loop you ran for a fortnight is a practice loop, not a finished habit.

They hide the intention-behaviour gap. Webb and Sheeran showed that even when interventions move intention, behaviour moves less. Drawing a prettier loop does not close that gap. Specifying the cue in an if-then sentence, Gollwitzer's method, does more.

Context does more work than the reward sticker

Popular writing overweights the third box. If the reward is big enough, the story goes, the routine will stick. That is a decent account of why you might choose a behaviour once. It is a weaker account of why you keep doing it when you are tired. Wood's line of work says the context starts retrieving the action. The reward can be real and still not be the daily engine.

Rewards that are too far away

"I will be fit in June" is not a reward the loop can use on a wet Wednesday. Outcome goals sit too far from the routine. A closer consequence is needed: the feeling of having done the minimum, the check mark, the kettle boiling because you already stood up. Fogg's celebration is an attempt to put a consequence inside the minute of action. It is slightly theatrical. It is also more mechanically honest than a vision board.

Duhigg's reporting on companies and societies is full of large rewards and dramatic keystone stories. Read those as narratives about organisations. Do not import them as a personal protocol that says "find a bigger treat and the 66-day problem goes away."

Cues that are already in the building

The cheapest cue is one you already obey. Shoes by the door. A glass on the sink. The end of a commute. Gollwitzer would write: if the commute ends, then I change before sitting down. That sentence is the loop with the cue nailed down. Without it, the routine waits for a decision, and Webb and Sheeran's gap opens.

Lally's participants were asked to repeat a behaviour in a consistent context. The paper is, among other things, a warning against loops that migrate around the house. A routine with no address takes longer to become automatic, or never does.

Building a loop you can actually run

Start with the routine sized for a bad day, not a photogenic one. Name the cue as a situation, not a virtue. Let the reward be immediate and small. Then leave the cartoon on the wall and measure repetitions in that context, not vibes.

Box	Popular reading	More careful reading
Cue	A reminder notification	A stable context: place, prior action, time
Routine	The impressive version of the behaviour	The version that still happens on a low day
Reward	A treat or a future identity	A close consequence, including Fogg's tiny celebration
Time	Implied to be short once the loop is "right"	Lally: median ~66 days, range 18-254

If you only remember one correction to the cartoon, remember this: Duhigg popularised a sequence; Wood and Neal described why the first box is often the room; Gollwitzer told you to write the sequence as if-then; Lally told you not to expect the sequence to feel automatic on a challenge calendar; Webb and Sheeran told you that meaning to run the sequence is not running it.

A home-screen widget in Habit AI is a designed cue, closer to Wood's context than to a motivational quote sitting in a notes app you never open.

Read next

How long does it actually take to form a habit puts a clock on automaticity. The 21-day habit myth is what happens when the cartoon gets a fake duration. More on the product: the iPhone tracker.