Psychology · 7 min read
What Implementation-Intention Studies Show About Following Through on Goals
A calibrated review of when if–then planning improves goal follow-through and where the evidence is weaker.

What Implementation-Intention Studies Show About Following Through on Goals
The intention–action gap
Deciding to do something and doing it are different events. A person may sincerely intend to take a walk after work, mail a form, or attend a screening appointment, yet fail when the relevant moment arrives. The obstacle can be forgetting, noticing the opportunity too late, facing a competing demand, or having to decide again while tired. Psychologists call this broad pattern the intention–behavior gap. It does not imply that intentions are useless; stronger intentions are generally associated with action. It means that an intention stated only as a desired end often leaves the occasion and response unspecified.
An implementation intention adds that missing link. Instead of “I intend to exercise,” it takes an if–then form: “If I finish work on Monday, Wednesday, or Friday, then I will change clothes and walk for twenty minutes.” Peter Gollwitzer introduced the construct in the 1990s as a supplement to goal intentions, not as a substitute for motivation, resources, or skill. The person first chooses a goal and then connects a recognizable cue to a feasible action.
The central empirical question is modest: when people already have a goal, does making this cue–response plan increase the chance that they act? Controlled studies and quantitative reviews generally find that it can. They do not show that a sentence overcomes poverty, illness, coercion, or an impossible schedule. The size and reliability of the benefit depend on what is being attempted, how outcomes are measured, and whether the plan identifies a useful cue.
What an implementation intention changes
The proposed mechanism has two parts. Specifying a situation may make that cue easier to notice. Linking it to a response may reduce the need for a fresh deliberation at the critical moment. In a 2007 experiment, Thomas Webb and Paschal Sheeran tested these component processes in a verbal task. Participants who formed implementation intentions showed greater accessibility of the cue and stronger cue–response links; those measures statistically mediated better goal attainment. This supports, but does not conclusively establish, the claim that planning partly transfers control from repeated conscious choice to a prepared response.
“Automatic” is easy to overstate here. It does not mean involuntary, instantaneous, or immune to context. A person can abandon a plan. The cue may never occur, may be too vague to recognize, or may call for a response that is no longer appropriate. A useful plan is specific enough to be detected and flexible enough not to produce absurd behavior. “If it is 7 p.m., then I will take the medicine prescribed for that time” is a clearer link than “If I have time, I will be healthier.”
The device also differs from a reminder. A calendar notification supplies an external cue. An implementation intention is a mental or written relation between a cue and an action. The two can be combined, and many real interventions bundle planning with reminders, information, feedback, or social support. Such bundles complicate attribution: an improved outcome cannot automatically be assigned to the if–then sentence alone.
What controlled studies and meta-analyses find
Gollwitzer and Sheeran’s 2006 review assembled 94 independent tests and reported a medium-to-large average effect on goal attainment, standardized as d = 0.65. The tests ranged across laboratory tasks and everyday aims, and the review found effects on getting started, persisting, and responding to difficulties. That result made implementation intentions one of the better-known brief techniques in self-regulation research.
The headline estimate needs context. The review combined heterogeneous designs, outcomes, populations, and comparison groups. Some studies measured a concrete behavior; others used self-report or a proximal laboratory outcome. Many were conducted by researchers close to the theory. An average across such studies describes the collected literature, not a guaranteed gain for a particular person or goal.
Later work has expanded the evidence base and often found smaller effects under narrower conditions. A 2021 meta-analysis by Doris Wang and colleagues examined “mental contrasting with implementation intentions,” a related intervention that asks people to contrast a desired future with present obstacles before planning. It found a small-to-moderate benefit and also detected signs of publication bias, warning that the underlying effect might be smaller. Because this intervention contains an added component, its estimate is not interchangeable with the 2006 estimate. It is nevertheless evidence that brief planning effects are sensitive to study selection and analytic method.
A more recent synthesis by Sheeran, Olivia Listrom, and Gollwitzer collected hundreds of tests to examine when planning works and which components matter. Its much larger scope reinforces the broad conclusion that if–then planning can improve goal attainment while directing attention to variation rather than a single universal effect. Stronger inferences come from patterns that recur across research teams, outcomes, and designs, not from the largest number printed in one review.
Where the effect becomes smaller or uncertain
Implementation intentions cannot create opportunity. A plan to attend a clinic will not supply transport, paid leave, an appointment, or accessible care. In health research especially, an intervention may improve a near-term behavior without demonstrating a clinical benefit. Taking more walks is not the same outcome as preventing cardiovascular disease; returning a screening kit is not the same as reducing mortality. Each link requires separate evidence.
The quality of the cue matters. A cue defined by an observable event—finishing breakfast, entering the office, hearing an alarm—is easier to identify than a mood such as “feeling motivated.” The response must also be under the planner’s control. Plans can fail when they are too ambitious, conflict with other aims, or depend on another person. Repetition may help a response become habitual, but implementation intentions and habits are not identical: habits are learned cue–response associations built through experience, whereas an implementation intention can be formed before the behavior has been repeated.
There is also a measurement problem. Participants know whether they received a planning exercise, so behavioral interventions are rarely blinded in the way a drug trial may be. Self-reported action can reflect demand characteristics or faulty recall. Attrition may remove the people who found the intervention least workable. Preregistered trials, objective outcomes, active comparison groups, and independent replication reduce these risks, but the literature does not use them uniformly.
Poorly chosen plans may have costs. A rigid response can crowd out adaptation when circumstances change. Planning one route to a goal can also leave a person unprepared when the specified cue fails to appear. Researchers disagree about how much flexibility to build in before the link loses its useful specificity. “If obstacle A occurs, I will use response B” is testable; a long menu of exceptions may become another deliberation problem.
What the evidence supports in practice
The defensible practical conclusion is narrower than “if–then plans make goals happen.” For a goal a person values and can realistically pursue, specifying a recognizable occasion and a concrete response is a low-cost technique with a positive average record. It is most plausible when failure comes from forgetting, delayed initiation, or an avoidable moment of indecision. It is less plausible when the binding constraint is money, authority, safety, time, health, or knowledge.
A reasonable use follows four checks. State the goal separately. Choose a cue likely to occur. Attach a small action that can begin at that moment. Decide when the plan should be reviewed rather than treating it as permanent. The result might be: “If I put my lunch plate in the sink on weekdays, then I will walk outside for ten minutes; after two weeks I will assess whether this fits.” That statement is a planning aid, not a prediction.
What is established is that implementation intentions, averaged across many controlled comparisons, can improve follow-through. What is contested is the magnitude after publication bias, intervention bundles, and study heterogeneity are considered. What remains unknown for any individual is whether a chosen cue, action, and setting match the mechanism well enough to help. The evidence supports trying a precise plan; it does not support blaming a person when conditions make the plan fail.
Sources
4 sources, 2 of them with a link you can follow.
- Peter M. Gollwitzer and Paschal Sheeran, “Implementation Intentions and Goal Achievement: A Meta-Analysis of Effects and Processes,” Advances in Experimental Social Psychology 38 (2006), doi:10.1016/S0065-2601(06)38002-1.
- Thomas L. Webb and Paschal Sheeran, “How Do Implementation Intentions Promote Goal Attainment? A Test of Component Processes,” Journal of Experimental Social Psychology 43 (2007), doi:10.1016/j.jesp.2006.02.001.
- Doris Wang and colleagues, “A Meta-Analysis of the Effects of Mental Contrasting With Implementation Intentions on Goal Attainment,” Frontiers in Psychology 12 (2021), PMID 34054628.
- Paschal Sheeran, Olivia Listrom, and Peter M. Gollwitzer, “The When and How of Planning: Meta-Analysis of the Scope and Components of Implementation Intentions in 642 Tests,” European Review of Social Psychology (accepted manuscript).
