Goal Setting

Implementation Intentions: If-Then Plans That Work

By iDel Published

An implementation intention is a plan in the form “If situation X happens, then I will do Y.” The concept comes from Peter Gollwitzer, and the main meta-analysis of it, Gollwitzer and Sheeran (2006), pooled 94 independent tests and found a medium-to-large effect on goal attainment (d = .65). Here d is a standardized mean difference: on average, people who formed if-then plans scored about two-thirds of a standard deviation higher on goal-attainment measures than people in the comparison conditions. It describes the gap between two group averages, so it does not mean that 65 percent of planners reached their goal.

A 2024 meta-analysis of 642 tests found effects between d = .27 and d = .66 across cognitive, affective and behavioral outcomes, with larger effects when plans used an if-then format, when people were highly motivated to pursue the goal and when plans were rehearsed. In two large field experiments the gains were a few percentage points: 4.2 points more flu vaccinations among employees prompted to write down a date and time, and 4.1 points higher turnout among voters reached by a plan-making phone call. If-then plans suit goals a person already wants, where the trouble is turning the intention into action.

Quick facts

ItemDetail
Format”If (cue), then I will (action)“
ConceptPeter M. Gollwitzer; overview in his 1999 paper in American Psychologist
Main meta-analysisGollwitzer and Sheeran 2006: 94 independent tests, d = .65 on goal attainment
Larger reviewSheeran, Listrom and Gollwitzer 2024: 642 independent tests, d = .27 to .66 across cognitive, affective and behavioral outcomes
Flu-vaccination field study3,272 employees; control group 33.1% vaccinated; date-and-time prompt +4.2 points
Voting field study287,228 registered Pennsylvania Democrats, 2008 presidential primary; plan-making call +4.1 points among people reached
Related methodWOOP (Wish, Outcome, Obstacle, Plan), developed by Gabriele Oettingen

What an implementation intention is

Gollwitzer’s 1999 paper in American Psychologist, Implementation intentions: Strong effects of simple plans, describes plans that link an anticipated critical situation to a goal-directed response: “Whenever situation x arises, I will initiate the goal-directed response y!” According to the paper, such plans delegate control of the response to the anticipated cue, which, when actually encountered, elicits the response automatically. Gollwitzer presents this as a way to call on automatic processes when people struggle to translate goals into action: failing to get started, becoming distracted or falling into bad habits.

The paper starts from a documented gap: intentions account for only 20% to 30% of the variance in behavior, and the weak link between intention and action is largely due to people who have good intentions and fail to act on them, a finding the paper credits to Orbell and Sheeran (1998). A goal intention (“I intend to exercise more”) names an aim. An implementation intention adds when, where and how. In the 2006 meta-analysis abstract, the authors describe the plan as spelling out “the when, where, and how of goal striving in advance.”

What the 2006 meta-analysis found

Gollwitzer and Sheeran (2006), published in Advances in Experimental Social Psychology (volume 38, pages 69 to 119), reports that “94 independent tests showed that implementation intentions had a positive effect of medium-to-large magnitude (d = .65) on goal attainment.” The abstract reports that implementation intentions were effective for four things:

  • starting goal-directed action
  • shielding ongoing goal pursuit from unwanted influences
  • disengaging from failing courses of action
  • conserving capability for later goal striving

It also reports support for two processes: planning made the specified opportunities more mentally accessible and automated the goal-directed response.

Three cautions belong next to the headline figure. First, d = .65 averages very different goals, populations and measures; the effect on a given reader’s goal may be smaller or larger. Second, a meta-analysis from 2006 rests on the literature available then, and the field has since grown much larger. Third, averages pooled from published studies can run high when studies with weak or null results go unpublished, a problem known as publication bias.

The 2024 update: 642 tests

Sheeran, Listrom and Gollwitzer (2024), in the European Review of Social Psychology, analyzed 642 independent tests. Forming implementation intentions “proved effective for cognitive, affective, and behavioural outcomes,” with effect sizes between d = .27 and d = .66. Effect sizes were larger when plans had a contingent (if-then) format, when participants were highly motivated to pursue the goal, and when plans were rehearsed. The authors also developed a taxonomy of the cues (such as time-and-place or a task juncture) and responses (such as cognitive procedures, “ignore” responses or inner speech) that plans can specify, and tested how well each promoted outcomes.

For a reader writing plans, those three moderators are the practical content: write the plan as an if-then sentence, use it for something you already want, and rehearse it.

Field studies: what happened outside the lab

Two large published field experiments show what the effect looks like outside the laboratory.

Flu shots at a large employer (Milkman and colleagues, 2011)

Milkman, Beshears, Choi, Laibson and Madrian (2011), in Proceedings of the National Academy of Sciences, ran a randomized experiment at free on-site influenza clinics at a large Midwestern utility firm in fall 2009. The headline rates appear in the published abstract; the sample details, group sizes and confidence interval below come from the NBER working-paper version of the same study.

  • Sample: 3,272 employees with a vaccination indication (age 50 or older, or a chronic condition raising the risk of flu complications).
  • Design: every employee got a mailing listing clinic times and locations. The treatment mailings added one line, “Many people find it helpful to make a plan for getting their shot. You can write yours here:”, followed by spaces for either the planned date or the planned date and time. Group sizes were 1,268 control, 1,270 date-and-time and 734 date-only, because employees at locations with one-day clinics were randomized only between control and date-and-time.
  • Result: the control group’s vaccination rate was 33.1%. In regression-adjusted estimates, the date-only prompt added 1.5 percentage points, which was not statistically significant, and the date-and-time prompt added 4.2 percentage points (95% confidence interval 0.5 to 7.8, P = .03). Before adjustment, the date-only and date-and-time groups were vaccinated at 35.6% and 37.1%.

Only the more specific prompt produced a statistically significant gain over the control group. The authors list limitations, including a predominantly male sample at a single firm and unusual conditions in fall 2009, when news about H1N1 and shortages of a separate H1N1 vaccine were widespread. As they note, everyone in the experiment faced the same background circumstances, so those conditions limit how far the result generalizes without contaminating the comparison itself.

Voting plans (Nickerson and Rogers, 2010)

Nickerson and Rogers (2010), “Do You Have a Voting Plan?” in Psychological Science, tested get-out-the-vote phone calls on 287,228 registered Democrats in Pennsylvania who were eligible to vote in the 2008 presidential primary, one person per household. Households were randomly assigned to a no-call control group or to one of three types of call: a standard call reminding people of the election and their duty to vote; the same call plus a question about whether they intended to vote; or that call plus three plan-making questions (what time they would vote, where they would be coming from and what they would be doing beforehand).

Among people actually reached, the plan-making call raised turnout by 4.1 percentage points, while the standard encouragement call and the self-prediction question had no significant impact. Because many targets were never reached, the effect across everyone assigned to the plan-making call was 0.9 points.

In a post hoc analysis, the result depended on household. Among single-eligible-voter households, plan formation raised turnout among people contacted by 9.1 percentage points; those in multiple-eligible-voter households were unaffected by all scripts. The authors present data suggesting why: people in one-voter households were more likely to answer “do not know” to the plan-making questions, which fits the authors’ view that people who live with other voters were more likely to have made a voting plan already (“organic plan making”).

What the two studies share

Both interventions were cheap, both kept a standard reminder (a mailing or a call) and added a prompt to make a concrete plan, and both produced gains of a few percentage points. Across thousands of people that is meaningful; for any one person it is modest.

Mental contrasting and WOOP

Implementation intentions address the “how” of a goal already chosen. Gabriele Oettingen’s work on mental contrasting addresses the earlier question of whether to commit at all.

Oettingen, Pak and Schnetter (2001) in the Journal of Personality and Social Psychology describe fantasy realization theory: when people contrast fantasies about a desired future with reflections on present reality, “a necessity to act is induced” and goal commitment follows expectations. Strong commitment arises when expectations are favorable and weak commitment when they are unfavorable. People who only fantasize, or only dwell on reality, show moderate commitment regardless of expectations. The paper reports four experiments across life domains.

The official WOOP site describes WOOP (Wish, Outcome, Obstacle, Plan) as a science-based mental strategy that people can use to find and fulfill their wishes, set preferences and change their habits. Its practice page says a WOOP takes only a few minutes but needs focus and the four steps in the right order. Research combining mental contrasting with implementation intentions (often abbreviated MCII) includes:

  • Stadler, Oettingen and Gollwitzer (2009): a 4-month randomized trial with 256 women aged 30 to 50 in Germany. Women who learned mental contrasting with implementation intentions in addition to a health-information session were twice as physically active as the information-only group, nearly an hour more per week of moderate-to-vigorous activity, from the first week after the intervention through the 4 months. Activity was self-reported.
  • Christiansen, Oettingen, Dahme and Klinger (2010): 60 chronic back pain patients randomly assigned to intervention or control while both groups followed a 3-week outpatient back pain program. The group given a one-hour goal-pursuit intervention (mental contrasting, a short problem-solving component and implementation intentions) had increased its physical capacity significantly more than the control group at both follow-ups, 3 weeks after discharge and 3 months after returning home, on behavioral measures (ergometer, lifting) and subjective ratings.

Both trials are modest in size, Oettingen is a co-author of each, and each compared MCII with information or routine treatment rather than with another planning technique. They are encouraging results, not evidence that WOOP beats other planning approaches.

Where if-then plans work less well

The 2024 review’s moderators already imply the limits: weakly motivated people gain less, and unrehearsed plans work less well. Other documented limits:

  • Context-specific effects. In the Nickerson and Rogers study, people in multi-voter households did not benefit, and the authors’ data suggest many of them had already made a plan.
  • Vague plans. In the flu study the date-only prompt did not significantly beat the control.
  • Small effects on some outcomes. A 2026 meta-analysis by Melum and Martiny-Huenger pooled randomized trials that tested if-then plans for fruit and vegetable intake in adults against active control groups (10 articles, 12 comparisons, N = 2,399). Planners reported eating about a quarter of a portion more per day than controls (mean difference 0.29, 95% CI 0.11 to 0.48). The authors call the effect small, note that every included study relied on self-reported intake, and point to the method’s low resource requirements.

The 2006 figure of d = .65 sits at the top of the .27 to .66 range that the 2024 review reports across outcome types. Different reviews report different averages depending on outcome, population and design, and plans do not fix goals the person does not want.

How to write an if-then plan

The guidance, tables and examples below are this site’s own synthesis of the research above; no paper prescribes this exact protocol.

The parts of a good plan

PartTestWeakStrong
CueCan you notice it happening?”If I have time""If I close my laptop at 5:30”
ResponseIs it one concrete action?”I will be healthier""I will change into running shoes”
LinkDoes the cue naturally precede the response?A cue that occurs at the wrong placeA cue at the place and moment the action is possible

Time-and-place cues and task-juncture cues are both categories in the 2024 taxonomy. A plan only runs when its cue occurs, so pick a cue that reliably happens on the days the plan is meant for.

Worked examples

GoalObstacleIf-then plan
Morning workoutSnoozing the alarmIf my alarm rings, then I will put my feet on the floor and put on my workout shirt before I touch my phone
Weekly reviewLetting Sunday evening driftIf I finish washing up on Sunday, then I will open my task list and run the weekly review
Deep workChecking messages firstIf I sit down at my desk, then I will open the draft and write one sentence before opening email
Healthy lunchDefaulting to the vending machineIf it is 12:30 on a workday, then I will go to my bag and eat the lunch I packed
Handling a setbackSkipping after one missed dayIf I miss a planned session, then I will do the shortest version the next day at the same cue

The last row applies the if-then format to a setback rather than a start; the 2006 meta-analysis found plans effective beyond getting started, including for shielding ongoing goal pursuit from unwanted influences.

Combining with a WOOP pass

For a goal you are not yet sure about, run the WOOP order first: name the wish, picture the best outcome, name the main inner obstacle, then write the if-then plan for that obstacle. Example: wish, finish the first draft by the end of the month; obstacle, I check social media when a paragraph gets hard; plan, if I feel the urge to open social media, then I will write the next sentence badly and keep going.

Fitting plans into the rest of your system

If-then plans are the smallest unit of planning, and they sit well under larger goals. A goal written the way SMART goals recommend names what, how much and by when; the plan supplies the cue. Plans also pair naturally with process goals, because a process goal (“write for 30 minutes”) is exactly the type of action that benefits from a trigger. For the daily layer, a ten-minute daily planning method is a natural place to choose that day’s cues, and the Sunday weekly review is where plans that did not fire get rewritten.

Common reasons plans fail

  • The cue never occurs (for example “when I get home” on days spent elsewhere).
  • Two plans compete for the same cue.
  • The response is too large for a low-energy day; shrinking it keeps the cue-response link alive.
  • The goal is not wanted; the 2024 review’s motivation moderator applies.

Sources