Zeigarnik Effect: The Power of the Unfinished
The Zeigarnik effect suggests we remember unfinished tasks better. What the 1927 study found, why replication fails, and how to use progress cues fairly.
Where does it come from?
Bluma Zeigarnik published a study in 1927 in the journal Psychologische Forschung. Her starting point was an observation by Kurt Lewin, who noticed that a waiter remembered unpaid orders while closed checks slipped from his memory (Wikipedia: Zeigarnik effect). Lewin’s explanation was that an unfinished task creates a lasting tension that keeps its content accessible in memory. It is a theoretical explanation, not a proven law.
Replications have not been kind to it. Wikipedia reports that Van Bergen (1968) found no significant difference in recall between finished and unfinished tasks, and that several later replications also failed. That is why the effect is still debated.
What does recent evidence say?
A 2025 meta-analysis by Ghibellini and Meier reviewed the literature on the Zeigarnik and Ovsiankina effects. According to its abstract, the authors found no memory advantage for unfinished tasks, but a general tendency to resume them (RePEc: Ghibellini & Meier, Interruption, recall and resumption).
The authors conclude that the Ovsiankina effect represents a general tendency, while the Zeigarnik effect lacks universal validity. They propose that the memory advantage depends on situational influences and individual differences, such as the authority of the experimenter, the demands of the task, and how involved the person is. Those conditions were more common in historical studies and are rarer today, which may explain why replication is difficult.
For design, the lesson is clear: what has better support is that people tend to resume interrupted tasks. What has less support is that a pending task is reliably remembered better.
That tendency to resume interrupted tasks is known as the Ovsiankina effect, after Maria Ovsiankina, a colleague of Zeigarnik. It is distinct from the Zeigarnik effect, which is about memory. For an interface the distinction matters: a progress indicator does not make someone remember the task better, but it may make it easier to resume it if they can see clearly what is left.
How is it used in interfaces?
The most cited applications are progress indicators and to-do lists. A typical example is the profile-completion meter, which shows that a profile is 64% complete and offers to fill in the missing fields. Wikipedia mentions that the effect is used in some software products for onboarding, although the article marks that claim “citation needed”.
Two uses are reasonable. First, show what is missing without hiding what has already been achieved: an indicator that says “Two steps left” guides better than one that only marks a percentage without context. Second, keep a to-do list visible so the person decides when to complete it, rather than hiding it after the first day.
When does it become manipulation?
The limit matters. A progress indicator that never completes, or that is filled artificially to pressure people, does not guide: it deceives. The same goes for lists that reappear every time the person tries to work, or for guilt messages such as “You haven’t finished your profile” repeated without end. Those patterns sit close to dark patterns, and the Zeigarnik effect does not justify them.
A simple test: if the person would finish the task anyway when they need it, the indicator helps. If it only pushes them back into the app to comply, the indicator is doing the work of an aggressive notification. That test is more useful than any figure in the literature, because it measures the person’s real benefit.
Illustrative example
Illustrative example: a personal finance app has a five-step onboarding: connect the account, choose categories, set a budget, configure alerts, and review the first month. The earlier version showed only a bar at 20%, without saying what was missing. The team swaps the indicator for a list of three pending items, with the first one highlighted: “Choose your categories: takes 1 minute.” The list can be dismissed and reappears only on the app’s home screen, not on every screen. Then the team measures whether the change improves the rate of defined budgets, and treats the result as a test of its design, not as confirmation of the Zeigarnik effect.
To decide which steps come first, review onboarding patterns. And if the critical moment is the end of the flow, the peak-end rule explains why the close carries so much weight.
Common mistakes
- Presenting the Zeigarnik effect as proven.
- Using progress bars that never complete.
- Showing to-do lists on every screen until the person stops seeing them.
- Filling the indicator artificially to pressure people.
- Confusing the tendency to resume tasks with the ability to remember them.
- Testing a progress change without measuring whether the real task improved.
What you can do tomorrow
- Review your progress indicator and check that each step it shows corresponds to something real the person does.
- Replace one percentage with a short list of concrete steps, with the first one highlighted.
- Define a task completion metric and compare the current version with the new one for two weeks.