All articles
6 min readfocus / productivity

Say when and where: the 2-line plan that closes the intention gap

I run about 5,000 km a year, which means starting a run roughly 300 times. On the evenings I decide the hour and the door I will leave from, I go. On the evenings I settle for "long run tomorrow", the run happens maybe half the time. Field experiments with hundreds of thousands of people have measured that same gap, and the format that closes part of it fits in two lines.

The gap has a number.

Paschal Sheeran ran a meta-analysis of meta-analyses on the link between what people intend and what they do: 422 studies, 82,107 participants, a sample-weighted intention-behavior correlation of .53 1. Intentions explain 28% of the variance in later behavior. Strong intentions predict action, and they leave most of the outcome undecided.

Peter Gollwitzer's answer to that gap is a sentence with a trigger in it. An implementation intention names the situation and the response in advance: if situation Y happens, then I do X. Gollwitzer and Sheeran pooled 94 independent tests and found a medium-to-large effect on goal attainment, d = .65 2. The plans in those studies cost people under a minute to write.

What the field experiments show.

A large firm offered free on-site flu clinics and mailed every eligible employee the times and locations. Katherine Milkman's team randomized what else the mailing contained. One group got a blank line for the date they planned to go. Another got lines for the date and the time. Employees who received no prompt were vaccinated at 33.1%. The date-only prompt added 1.5 percentage points, a difference too small to separate from noise. The date-and-time prompt added 4.2 points, a statistically reliable gap 3. Same clinic, same mailing, one extra blank for the hour.

David Nickerson and Todd Rogers tested the same mechanism on 287,228 voters before the 2008 US presidential election. A standard get-out-the-vote call moved nothing. Neither did asking people to predict whether they would vote. The script that worked asked three questions: what time will you vote, where will you be coming from, and what will you be doing beforehand. Turnout among people reached by that call rose 4.1 percentage points, and 9.1 points in households with a single eligible voter 4. In households with several eligible voters the script did nothing, since those people already build the plan out loud with each other.

Angela Duckworth added the obstacle step and tested it on teenagers studying for the PSAT over a summer. Students trained to picture the outcome, name the obstacle that would get in the way, and write an if-then plan for that obstacle completed 140 practice questions. The control group completed 84 5.

The effect shrinks outside the one-shot decision.

Getting a shot and casting a vote are single actions with a deadline. Repeated behavior is harder, and the numbers say so. Ariane Bélanger-Gravel's meta-analysis of implementation intentions for physical activity found g = 0.31 right after the intervention and g = 0.24 at follow-up 6, well under the 0.65 of the pooled literature. A 2021 meta-analysis of the obstacle-plus-plan version across 21 studies and 15,907 participants reported g = 0.336, and its authors flagged publication bias, so read that as a ceiling 7.

The clearest limit comes from the largest test. Milkman's team ran 54 four-week digital programs on 61,293 members of an American gym chain. Weekly visits rose in 45% of the programs, by 9% to 27%. After the four weeks ended, 8% of the programs still showed a measurable effect 8. A written plan starts a behavior. Sustaining it depends on whether the cue outlives the prompt.

Phillippa Lally's team followed 96 volunteers doing one chosen action daily in a consistent context for 12 weeks. The median time to reach 95% of maximum automaticity was 66 days, with individuals spread between 18 and 254 days. Missing one opportunity did not measurably damage the curve 9. Consistency of the context matters more than perfection of the streak.

The protocol.

  1. Write the plan as an if-then sentence with a clock time and a physical place. "At 7:00, in the clothes laid out the night before, I leave by the front door." The date-only version of that sentence bought Milkman's employees nothing measurable.
  2. Anchor the trigger to something that already happens at a fixed point in your day: the end of the 9:00 stand-up, closing the laptop, the kids leaving for school. A floating "in the morning" has no trigger to fire on.
  3. Name one obstacle and write its counter-move in the same sentence. Duckworth's students planned for the specific thing that would stop them, then wrote what they would do instead.
  4. Keep one plan running at a time. The studies above each tested a single behavior; nobody has shown that six plans at once work.
  5. Put the written plan where the cue happens: a calendar block at that clock time, a note on the door, the bag by the exit. The plan has to be visible at the moment it fires, not filed in a notes app.
  6. Hold the same time and place for two months before you judge it. Lally's median was 66 days, and one missed day was harmless.
  7. Rewrite the plan when the context breaks. Travel, a new schedule or a move deletes the trigger, and the behavior goes with it unless you name a new one.

I apply step 7 in the trainings I run. People used to leave my sessions with a list of techniques and good intentions. Now the last 10 minutes are spent writing one sentence: which case from their own job they will run with AI, on what day, at what hour, and what they will do when the first output comes back wrong. Ask for the hour. The date on its own is worth 1.5 points.

Sources.

  1. Sheeran, P. (2002). Intention-Behavior Relations: A Conceptual and Empirical Review. European Review of Social Psychology, 12(1), 1-36. doi.org/10.1080/14792772143000003
  2. Gollwitzer, P. M., & Sheeran, P. (2006). Implementation Intentions and Goal Achievement: A Meta-analysis of Effects and Processes. Advances in Experimental Social Psychology, 38, 69-119. sciencedirect.com
  3. Milkman, K. L., Beshears, J., Choi, J. J., Laibson, D., & Madrian, B. C. (2011). Using implementation intentions prompts to enhance influenza vaccination rates. PNAS, 108(26), 10415-10420. doi.org/10.1073/pnas.1103170108
  4. Nickerson, D. W., & Rogers, T. (2010). Do you have a voting plan? Implementation intentions, voter turnout, and organic plan making. Psychological Science, 21(2), 194-199. doi.org/10.1177/0956797609359326
  5. Duckworth, A. L., Grant, H., Loew, B., Oettingen, G., & Gollwitzer, P. M. (2011). Self-regulation strategies improve self-discipline in adolescents: benefits of mental contrasting and implementation intentions. Educational Psychology, 31(1), 17-26. doi.org/10.1080/01443410.2010.506003
  6. Bélanger-Gravel, A., Godin, G., & Amireault, S. (2013). A meta-analytic review of the effect of implementation intentions on physical activity. Health Psychology Review, 7(1), 23-54. doi.org/10.1080/17437199.2011.560095
  7. Wang, G., Wang, Y., & Gai, X. (2021). A Meta-Analysis of the Effects of Mental Contrasting With Implementation Intentions on Goal Attainment. Frontiers in Psychology, 12, 565202. doi.org/10.3389/fpsyg.2021.565202
  8. Milkman, K. L., Gromet, D., Ho, H., Kay, J. S., Lee, T. W., Pandiloski, P., et al. (2021). Megastudies improve the impact of applied behavioural science. Nature, 600, 478-483. doi.org/10.1038/s41586-021-04128-4
  9. Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. (2010). How are habits formed: Modelling habit formation in the real world. European Journal of Social Psychology, 40(6), 998-1009. doi.org/10.1002/ejsp.674

Keep reading