The Zeigarnik effect (the unfinished task effect)

What is the Zeigarnik effect?

The Zeigarnik effect says that interrupted or unfinished tasks are remembered better than finished ones.

In interface design, the idea is often mentioned together with progress indicators. The percentage on a profile card and the setup checklist are based on it. The person sees the unfinished task and wants to complete it.

The effect itself is disputed. Below I describe both the 1927 study and the 2025 meta-analysis that questions it.

Bluma Zeigarnik and the 1927 study

Bluma Zeigarnik (1900–1988) studied and earned her doctorate at the University of Berlin. She wrote her thesis under the Gestalt psychologist Kurt Lewin. It appeared in 1927 in the journal Psychologische Forschung.

Its German title is “Das Behalten erledigter und unerledigter Handlungen” (the retention of completed and uncompleted actions). It is in volume 9 of the journal, pages 1–85.

The starting point was an observation in a restaurant. According to Wikipedia’s account, Lewin noticed that a waiter remembered unpaid orders better. Once the bill was paid, the waiter no longer remembered the details of the order.

Lewin’s explanation rests on the idea of tension. A task that has been started creates a tension that keeps the related information in mind. When the task is done, the tension is released; if it stays unfinished, the tension persists.

Is the effect real?

The reliability of the effect is disputed. According to Wikipedia, some later studies in other countries found no clear difference in recall of finished and unfinished tasks.

In 2025, Romain Ghibellini and Beat Meier of the University of Bern combined the studies on this topic in a meta-analysis. The study appeared in the journal Humanities and Social Sciences Communications. They looked at the Zeigarnik effect and the Ovsiankina effect together.

The Ovsiankina effect is named after Maria Ovsiankina, a colleague of Zeigarnik’s. This effect isn’t about memory but about the urge to resume an interrupted task.

Comparison of the Zeigarnik and Ovsiankina effects

Effect	What it says	2025 meta-analysis
Zeigarnik	Unfinished tasks are remembered better.	No memory advantage found.
Ovsiankina	There is an urge to resume an interrupted task.	Found as a general tendency.

Source: Ghibellini and Meier (2025), abstract.

According to the authors, the Zeigarnik effect may depend on situational influences and individual differences. As examples they name the experimenter’s authority, the demands of the task and how involved the person is in the task. These circumstances were more common in older studies.

In my view, that may be why the effect showed up more easily in old experiments. Don’t use the sentence “unfinished tasks stay on the mind” like a fixed law. The finding that helps interfaces is the tendency to resume unfinished tasks.

The Zeigarnik effect in the interface

In the interface, you meet the idea in the following patterns. In all of them, the goal is the same: show the person that a task is unfinished and how much of it is left.

Progress bar

A counter such as “2 of 4 steps” in a multi-step form tells the person how much work is still ahead. The number must be tied to real steps. If a step opens or closes depending on a condition, the counter must change too.

Profile completion

The percentage on the profile card shows the unfinished task. It must be clear what the percentage counts: which step adds how much, and what is left?

Checklist

The setup checklist lists what to do on first use. If each item is tied to a single task, the person knows where to continue. Details are in the onboarding pattern.

Picking up where you left off

The strong finding of the meta-analysis was the tendency to return. The simplest way to support it: don’t let unfinished work get lost. Save a long form as a draft, and greet the person on their return at the step where they stopped.

A line such as “You’re on step 3 of your application, continue” reminds the person what is unfinished. They don’t have to click through from the start to find their place.

Endowed progress

There is a related finding. Joseph Nunes and Xavier Drèze called it the “endowed progress effect” in 2006. They presented a task of eight steps as ten steps with two already done, and people persisted more.

According to the authors, the reason given for this head start, among other things, changes the effect. If you already entered your email when signing up, counting that as the first completed step is honest. Showing a filled bar without anything having been done isn’t.

Live example: three profile cards

Below are three states of the same fictional profile card. Choose a state, then tick or untick steps. The box next to it describes what the card is doing; on narrow screens it moves below the card.

Three profile cards

This demo works with JavaScript. Without JavaScript, you see the honest state of the card: 2 of 5 steps done.

Your profile

The profile card of a fictional app

2 of 5 steps done40%

• Add a profile photo
• Confirm your phone number
• Write a short bio
• Choose your interests
• Set up notifications

What happens in this state?

The indicator really counts the boxes. Each tick moves the bar on by a fifth; untick one, and the bar goes back. Once all five steps are done, the card says “complete” and asks for nothing more.

A fictional profile card; the steps are made up. In the honest state, the bar counts the ticked boxes. The manipulative state deliberately shows a false number and adds steps as soon as the list is done. Nothing is stored.

In the honest state, the bar counts the ticked boxes. In the manipulative state, the indicator stays stuck at 90 percent, and the list gets longer every time it is done. Both cards show the same boxes; the only difference is whether the number is true.

Common mistakes

With progress indicators, I first check whether it is unclear what the number counts. The percentage rises, but nothing says which step adds how much. The person can’t judge what they have to do for the task to be done.

Then there is the indicator not counting what is already done. Someone who has already confirmed their email is confused when it says “Confirm your email”. The indicator must follow the actual state of the data.

Don’t show optional steps like required ones. Once the required steps are done, the card must say 100 percent.

The ethical limit: fake incompleteness

The shortest way to abuse the Zeigarnik effect is fake incompleteness. Telling a profile that is actually complete “one last step”, or adding a new step after every completed one, belongs in this class. It is an example of a dark pattern.

When I look at an indicator, I ask myself:

• Does the indicator count a real state, and is the task done when the list is done?
• Does the missing step benefit the person, or only the product?

If either answer is “no”, change the indicator. Inviting friends or trying a paid plan isn’t a step that has to be completed; it is an optional suggestion.

Design the end too. The peak-end rule says people largely judge an experience by how it felt at its peak and at its end. A list that never ends is an experience with a bad ending.

My note

I think of a progress indicator as a reminder, not a trap. I show the person honestly what is unfinished, and they decide whether to continue.

In audits, I look at where the number comes from. If nobody on the team can say which steps make up the percentage, the person won’t understand it either. First I tie the number to real steps, then I design the bar.

If you want the indicators in your own setup flow checked from outside, the UX Audit looks at a flow from start to finish and delivers a prioritised list of fixes.