Cognitive biases
The planning fallacy: why our own tasks take longer than we expect
On Friday evening Jenna decides: “Tomorrow morning I’ll clear the pile on my desk in an hour, and then I’ll go for a calm walk.” A stack of papers, bills, cables, books that are long overdue to be returned. By lunchtime on Saturday, half of it is sorted. Under the papers she finds a letter that needs a reply. The charger turns out to belong to a different phone. Jenna moves the walk to Sunday. And she catches herself thinking that it was exactly the same last time.
This is the planning fallacy: we expect our own task to take less time than it really does. Even if similar tasks ran long before. In big projects, money often runs short too.
Below, we look at what research has shown, when this mistake happens and when it doesn’t, and how to plan more accurately. In the middle of the article you will clear a small desk on the screen and compare your estimate with a stopwatch.
Contents
In short: how the planning fallacy works
What the planning fallacy is
The name was coined by the psychologists Daniel Kahneman and Amos Tversky in the late 1970s. They studied how people make forecasts and proposed distinguishing two kinds of information. The first is what we know about the task itself: what our plan is and how many steps it has. The second is how similar tasks usually end. Later Kahneman and Dan Lovallo called these the “inside view” and the “outside view.” Their conclusion: people tend to consider their own task special and do not look at how things went in the past. A forecast from the inside clings to the plan and so turns out too optimistic.
The most famous series of experiments was run by Roger Buehler, Dale Griffin, and Michael Ross in 1994. Participants said when they would finish their own tasks, academic and everyday ones. Then the researchers checked when the tasks were actually finished. The forecasts proved too optimistic for a wide range of tasks.
The most interesting part was what was going on in people’s heads. Participants were asked to think aloud while they estimated a deadline. For the most part, people talked about the future: what I will do first, what next. Past similar tasks stayed off to the side. And their own past misses with deadlines they explained by outside, one-off causes, like “I was sick then” or “that was an unusual week.” As if those did not apply this time.
Two more results from the same work. When participants were asked outright to link the forecast to past experience, the optimistic tilt disappeared. And when people estimated a deadline for someone else, they more often relied on that person’s past tasks and even gave them more time than was needed.
Do we always underestimate? No, and this matters. In a large review of studies, underestimation was more common in engineering and project-management work, while in psychological experiments it was no more common than overestimation. With very short tasks it can go the other way: in an experiment by Michael Roy and Nicholas Christenfeld, people were more likely to overestimate tasks shorter than about two minutes and to underestimate long ones. Part of this pattern may come simply from measurement error. The takeaway: the planning fallacy is more noticeable in big tasks of your own with many steps than in tasks of a couple of minutes.
Try it yourself: clearing the desk
This is a small version of Jenna’s Saturday. It takes a couple of minutes.
Step 1. You will see a desk with 12 items on it. Each has its place: everything you write with goes in the cup, papers in the tray, the rest in the drawer. With a slider you estimate how many seconds the cleanup will take you.
Step 2. You clear the desk. The stopwatch starts by itself. The numbers are hidden while you work, so that you do not race against your own estimate.
Step 3. We compare your estimate with how much time actually passed.
Why a cleanup? It is a real task, not a problem on paper: the time is measured honestly. And the desk is the same for all players. But we warn you right away: the task is short, and all the items are in view from the start. If you finish in time, that is a normal result, and at the end of the game we explain why that happens.
What may happen. With a small desk, people often finish on time or even faster: short tasks are more often overestimated. The planning fallacy is stronger in big tasks that stretch over days and cannot be seen all at once.
If you did not finish in time, look at what the time went on. Usually it is small things that do not get counted in an estimate: pulling an item out from under the papers, moving something that was put in the wrong place. On Jenna’s Saturday there were far more of those small things.
Why we expect the best again
Picture a route in a navigation app. On the map the road looks like a straight line: twenty minutes, and you are there. Traffic lights, traffic jams, and the search for parking are not drawn on it. This is an image, but it conveys the “inside view” well: we look at the plan, and a plan is always cleaner than the road.
A plan is the best-case scenario. When we estimate a deadline, we walk through the steps of the task in our heads, in order. A study by Ian Newby-Clark and colleagues showed that such plans usually resemble the most successful course of events. People were asked to imagine a bad scenario too. But such scenarios seemed less likely to them and barely changed the forecast for their own task. For someone else’s task, people did take them into account.
Past delays look like flukes. Each past delay looks special: “the internet went down then,” “relatives came to visit then.” Taken one at a time, that is true. But together, these “special cases” are just how things normally go. In the experiments of Buehler and colleagues, people explained their past misses exactly this way: by outside, one-off causes.
We want to be done sooner. In another study by Buehler, Griffin, and Heather MacDonald, people who were expecting a tax refund gave more optimistic forecasts for when they would fill out their tax return than those who were not expecting one.
Perhaps memory lets us down, too. There is another explanation, and scientists are discussing it. A review by Roy and colleagues collected evidence that we also remember the duration of past tasks as shorter than it was. If so, we underestimate the future because we underestimate the past. These explanations do not rule each other out.
Where it gets in the way
Home. Clearing a desk, sorting a closet, a small repair “over the weekend.” The sneakiest part is what you cannot see in advance: a letter under the papers, a broken shelf behind the closet, the store does not have the right paint.
Work. “The report will be ready by Wednesday.” We name a deadline while picturing the work without interruptions. And on Wednesday there will be two more meetings and an urgent question from a colleague. With big projects it is the same thing, only costlier: Bent Flyvbjerg’s work on large projects, for example transportation ones, shows that forecasts of their cost are often inaccurate. There are two reasons: optimism and, sometimes, deliberate understating of the figures.
School and studying. A term paper “a week before it’s due,” exam prep “in three evenings.”
How to plan more accurately
The planning fallacy is one of the biases on the cognitive bias map, in the section on decisions. Next to it is the anchoring effect: the first deadline named easily becomes the starting point for all the ones that follow. And the sunk cost fallacy, which makes a task that has dragged on hard to stop in time.
Frequently asked questions
Why do we give other people more cautious deadlines than ourselves?
Probably because we do not see the detailed plan of someone else’s task, but we do remember how that person coped before. In the experiments of Buehler and colleagues, people who estimated other people’s tasks relied more on past experience and gave even more time than was needed.
Does it help to imagine that everything goes wrong?
Not much, at least for your own tasks. In the study by Newby-Clark and colleagues, people described a bad scenario but considered it unlikely, and the forecast barely changed. It is more reliable to lean on facts: how long similar tasks took before.
So should I always allow twice as much time?
There is no universal multiplier: for some people tasks run over more, for others less, and short tasks sometimes take less time than they seem to. The best multiplier is your own, calculated from your past similar tasks.
Do we underestimate every task?
No. The planning fallacy is more noticeable in big tasks of your own with many steps. With tasks shorter than a couple of minutes, people more often overestimate, though part of that pattern may be a statistical artifact. And in psychology experiments, underestimation was no more common than overestimation.
