Temptation Bundling: Milkman's Research That Boosted Gym Visits by 51% (And How to Apply It)

Katherine Milkman had a problem: she knew she should go to the gym, but she couldn't make herself go consistently. She was also addicted to audiobooks of novels she was slightly embarrassed to admit enjoying.

The solution she devised — and then studied rigorously — became one of the most practically useful findings in behavioral economics: temptation bundling.

The concept: pair something you want to do with something you should do. The immediate pleasure of the "want" behavior counteracts the delayed or abstract reward of the "should" behavior, making initiation easier and follow-through more likely.

Milkman, now a professor at Wharton, published the foundational research in 2014 with colleagues Julia Minson and Kevin Volpp. The study design was elegant: participants were given access to audiobooks of popular novels only during gym visits. The result was a 51% increase in gym visits compared to the control group.

The Want-Should Conflict

The psychological root of the problem temptation bundling solves: want behaviors offer immediate rewards; should behaviors offer delayed or abstract rewards.

The brain's reward circuitry is heavily present-biased — a well-documented phenomenon called temporal discounting. A reward available now is experienced as significantly more valuable than the same reward available in the future. The farther in the future a reward, the more it's discounted.

Exercise produces health benefits over years. Deep work produces professional results over months. Studying produces knowledge that pays off over a career. In each case, the primary reward is so temporally distant that the immediate cost of initiation reliably outweighs it in the brain's value calculation — even for people who genuinely care about the long-term outcome.

The competing "want" behaviors — social media, entertainment, comfort food — offer immediate rewards that win the present-moment comparison easily.

Milkman's insight: you can't eliminate temporal discounting, but you can engineer an immediate reward for the "should" behavior. Bundling the want with the should makes the combined activity immediately rewarding, changing the cost-benefit calculation at the point of initiation.

The Research Results

In the gym study, participants were randomly assigned to three conditions:

• Full bundling: Access to audiobooks only during gym visits
• Intermediary bundling: Given audiobooks but encouraged (not required) to use them only at the gym
• Control: No audiobooks provided

The full bundling group visited the gym 51% more often than the control group during the intervention period. Even after the intervention ended and the restriction was removed, gym visits remained elevated — suggesting the bundling had created associations between gym attendance and positive experience that outlasted the immediate incentive.

The intermediary group showed smaller effects, pointing to an important nuance: the bundling needs to create a genuine linkage, not just a loose suggestion. When the want behavior is available regardless of whether the should behavior is performed, the motivational mechanism weakens.

Milkman followed this research with "Healthy Habits" and "Fresh Start" studies that built on the same principles, and later synthesized the full body of research in her book How to Change. The temptation bundling finding has replicated across contexts from exercise to retirement savings contributions to medical screenings.

Designing a Temptation Bundle

For a temptation bundle to work, the pairing needs to satisfy several conditions:

The want behavior should be truly immediate and enjoyable. Vague pleasures don't work well. The more specific and genuinely craved the want behavior, the stronger the bundle. "Nice background music" is a weak want. "The podcast I'm completely hooked on" is a strong one.

The want behavior should require the should behavior to be present. The critical design principle: the want behavior is only available when the should behavior is being performed. If you can access the want behavior any time, the linkage breaks.

The want behavior should not impair the should behavior's core mechanism. For focus work, this introduces a constraint: a podcast you actively follow impairs writing (both use language processing). The temptation has to be compatible with the type of attention the should behavior requires.

Compatible bundles for focus work:

• Instrumental music (no lyrics) you love → only during deep work sessions
• A premium coffee or tea → only prepared and consumed at the start of a focus session
• A specific comfortable setting (favorite café, cozy chair) → only visited for focus sessions
• A small treat after session completion → only when the session actually runs

Incompatible bundles for focus work:

• Narrative podcasts during writing (language interference)
• Social media during sessions (obvious)
• Watching video during cognitively demanding work

The Post-Session Reward Bundle

A particularly effective variant for deep work: bundling the session itself with something genuinely enjoyable that follows it immediately.

The mechanism here is slightly different — instead of making the during-session experience more rewarding, you engineer an immediate post-session reward that the brain begins to associate with session completion.

This draws on Fogg's celebration mechanism and Graybiel's reward-encoding research: immediate positive experience after a behavior accelerates the basal ganglia encoding process. The brain doesn't just remember the reward — it begins to anticipate it at the point of initiation, creating a craving that drives behavior toward the session rather than away from it.

The post-session reward needs to be:

• Genuinely enjoyable (not instrumental — not another "should")
• Accessible only after session completion
• Immediate (within minutes, not hours)

Examples: a favorite food only eaten after sessions, a specific leisure activity that serves as the session's "reward period," even a specific notification sound or visual that reliably produces a small dopamine response.

The Bundling Trap

Milkman's research also identifies when temptation bundling fails: when the want behavior becomes more salient than the should behavior, and the should becomes a means to access the want.

This reframes the relationship in a way that can undermine the should behavior's intrinsic value. If you become entirely dependent on the audiobook to tolerate the gym, and the audiobook is unavailable one day, gym attendance collapses.

The solution: design the bundle for the initiation problem (getting started), not as the permanent structure of the habit. Once the should behavior has its own established rewards — the intrinsic satisfaction of completion, the visible progress, the competence development — the want bundle can be gradually relaxed without the behavior collapsing.

Application to the Focus Practice Specifically

For building consistent deep work sessions, temptation bundling addresses the hardest part: the moment of initiation when competing alternatives are available.

The bundle doesn't need to be elaborate:

"I only make my favorite pour-over coffee when I'm about to start a focus session."

This works through multiple mechanisms simultaneously:

• The coffee ritual becomes the anchor habit (Fogg's habit stacking)
• The coffee itself becomes the immediate reward for session initiation
• The sensory experience (aroma, taste, warmth) creates a positive physical state associated with focus work
• Over time, the coffee ritual activates the focus state automatically

The bundle creates what Milkman calls "strategic self-control" — using the structure of the environment and the pairing of immediate rewards to make the desired behavior the path of least resistance, rather than requiring raw willpower to override competing demands.

The Bottom Line

Temptation bundling — pairing a "want" behavior with a "should" behavior, with the want accessible only when the should is being performed — increases follow-through by engineering immediate rewards for behaviors whose primary rewards are delayed or abstract.

Milkman's 2014 research found a 51% increase in gym visits through bundling alone. The mechanism: temporal discounting makes present-moment costs feel more significant than future rewards; bundling creates an immediate reward that changes the initiation calculus.

For focus work: pair something genuinely enjoyed with session initiation. The enjoyable element counteracts the present-moment friction of starting.

Frequently Asked Questions

Does temptation bundling work for everyone?

Milkman's study showed significant average effects, but individual response varied. The key variable is whether the individual genuinely craves the "want" behavior enough for it to function as a meaningful immediate reward. Choose wants that you actually find compelling, not ones you think you should enjoy.

What if I start enjoying the bundled behaviors so much I do them without the should behavior?

This is the bundling trap Milkman identifies. The solution is strict linkage: the want behavior is genuinely not accessed outside the bundle. This requires commitment in advance, when motivation is higher, to enforce the restriction when motivation is lower.

Can you bundle should behaviors with other should behaviors?

The mechanism requires the bundled behavior to be genuinely immediately rewarding, not just valuable in the abstract. Pairing two "should" behaviors may produce some motivation through sense of efficiency, but it doesn't engage the present-bias correction mechanism that makes temptation bundling effective.

How do I find the right "want" to bundle?

Milkman suggests auditing your current leisure time: what are the things you look forward to, feel slightly guilty about enjoying, or find yourself returning to repeatedly without external prompting? Those are your highest-value want behaviors for bundling purposes.