Temptation bundling

Temptation bundling is Katy Milkman's name for a simple trade: you only get the want if you are doing the should. The audiobook only plays at the gym. The favourite show only plays on the treadmill. The fancy coffee only happens after the walk. The tempting thing becomes a cue, or a close reward, instead of a rival.

It is a clever use of motivation. It is not, by itself, a habit. Wendy Wood and David Neal still want a stable context. Phillippa Lally still wants weeks of repetition. Peter Gollwitzer still wants an if-then that can fire when the want is unavailable (headphones broken, show finished, cafe closed). Bundling is a booster rocket. It is a poor satellite.

Pairing want with should

The pairing works when the should is otherwise easy to skip and the want is otherwise easy to overconsume. You are rationing a pleasure so that it pulls a behaviour you endorse. Charles Duhigg's popular cue-routine-reward loop can describe this: the want is a vivid third box. Duhigg popularised the loop; Milkman and colleagues tested bundling as a behavioural strategy. Keep the credits separate. Do not treat a TED-friendly story as the trial.

Initiation is the job it is good at

Thomas Webb and Paschal Sheeran showed that intention often fails to become behaviour. A bundle attacks that gap with something more visceral than a goal statement: you are slightly deprived of a pleasure until you start. Fogg would say you raised motivation for a behaviour whose ability might still be low. If the gym is 40 minutes away, a novel may not be enough. If the bike is in the spare room, a podcast might be.

This is not medical advice. Bundling a show with a walk can help a walk happen. It does not prescribe a training plan.

When the want is the wrong shape

If the want is a phone feed, bundling it with a stretch often produces a stretch that lasts one notification. The feed is a rival routine with its own cues, as Wood would emphasise. Bundle pleasures that do not steal the motor system you need. Audio is a better gym companion than a social app for that reason. The should has to remain possible while the want is on.

What bundling cannot buy

It cannot buy automaticity on a timetable. Lally and colleagues (2010, European Journal of Social Psychology) found a median near 66 days to a plateau of automaticity, with a range from 18 to 254. When the series ends, you will find out whether the context can retrieve the behaviour without the treat. If it cannot, you had a promotion, not a habit.

The cue should outlive the treat

Write Gollwitzer's sentence underneath the bundle: if I enter the gym, then I start the first machine before I press play. The gym is the cue. The audiobook is the sweetener. If you reverse them (if I want to listen, then I might go), the want remains the boss. Bosses are flaky.

Bundles can become permission structures

"I only eat biscuits when I read" can become a lot of biscuits and very little reading, or a lot of both. The rule needs a ceiling on the want, not only a lock on the should. Environment design still wins: do not keep a warehouse of the tempting thing next to the sofa and hope the rule holds at 10 p.m.

Making the pair a cue, not a treat

Put the want behind a physical threshold that is the should. Headphones live in the gym bag, not on the sofa. The show is logged out on the television and logged in on a laptop that only comes out in the training space. That is Wood's context plus a bundle. Fogg would still shrink the first visit: put on shoes, walk to the corner, podcast allowed. Two minutes of should can earn two minutes of want while Lally's curve is still a rumour.

Bundle	Helps when	Breaks when
Audiobook only at the gym	The gym is reachable; audio does not steal form	The book ends; the bag stays in the car
Coffee after a walk	The walk is short; coffee is already a ritual	The cafe is closed and there is no backup if
Favourite show on the bike	The bike is in a dedicated corner	The show migrates to the sofa "just this once"
Social app after writing	Writing has a clear start cue	The app is opened "to check" and the writing dies

Webb and Sheeran's gap is not closed by a cleverer treat if the situation is unspecified. Name the place. Size the should. Let the want be smaller than you would like. Repeat until the place does some of the pulling. Then you can retire the bundle, or keep it as a pleasure, without needing it as a crutch.

Bundling also fails when the want is a variable ratio reward, the kind of feed that never ends. A novel ends. A podcast episode ends. A feed does not, so the should never gets to be the thing that finished. Gollwitzer's then needs an end state. "Until the episode ends" is an end state. "Until I have had enough scrolling" is not. Wood's context cannot save a bundle whose treat has no off switch. Pick wants that stop.

A home-screen widget can sit next to the bundle as a second cue: Habit AI puts today's should in the visual field, so the want is not the only thing that can start the sequence when the headphones are missing.

Read next

Habit stacking that does not collapse is another way to borrow a cue, without a treat. Tiny habits and the two-minute rule is how small the should should be. the iPhone tracker can hold the should when the series ends.