If your tasks regularly take longer than you expect, the problem may not be poor discipline. People commonly underestimate how long future tasks will take, even when they have completed similar tasks before. This pattern is known as the planning fallacy. A useful way to counter it is to estimate in two passes: first, list what the task actually involves; then compare your estimate with the time similar tasks really took.
The aim is not to predict every minute perfectly. It is to make plans that fit the work and the time available, while learning from the difference between what you expected and what happened.

Why your first estimate is often too optimistic
When you picture a task, you may focus on the work itself and imagine it going smoothly: the report gets written, the room gets sorted, the form gets completed. That is the “inside view”—a forecast built around the particular task in front of you and the steps you can readily picture.
The trouble is that this view can leave out what comparable tasks were like in practice. Perhaps the last report needed another round of revisions, or the previous form took longer because you had to find information first. Research on the planning fallacy has found that people’s completion predictions can be overly optimistic even when they have relevant past experience. In their original studies, Buehler, Griffin and Ross found that people focused on the future task while giving too little weight to what had happened before. Their research on the planning fallacy helps explain why simply knowing that you tend to run late may not be enough to change your next estimate. Use what you learn from completed tasks when you prepare the following workday before leaving.
There may be more than one reason this happens. A review by Roy and colleagues discusses the possibility that people remember past events as shorter than they actually were, alongside explanations focused on how future tasks are imagined. That means it is better to treat the bias as a useful description of a common forecasting problem, not as a single flaw with one guaranteed fix.
Underestimating a task is not proof that you are lazy or careless. It is a reason to build a more grounded estimate: include the work you might otherwise overlook, and make use of your own record of similar tasks where you have one.
Define what “how long” means
Before estimating, decide what you are trying to predict. “Finish the presentation by Thursday” could refer to the total time you will actively work on it, or the number of calendar days until it is complete. Those are different measures.
Active working time counts time spent doing the task. Calendar time also reflects when you can work, time between work sessions, other people’s availability, dependencies, interruptions and waiting. A task that requires an answer from someone else might take only a short time to complete once the answer arrives, but several days to finish in calendar terms.
For a realistic plan, name the measure and track it consistently. If you are deciding whether a task fits into an afternoon, active time may be the most useful estimate. If you are promising a delivery date, calendar time matters too. When both affect the plan, estimate both rather than treating them as interchangeable.
Also define what “done” includes. Is the task finished when the first draft exists, or when it has been checked and sent? Does “organise the room” include putting away the items you find along the way? A clear endpoint makes estimates easier to compare later.
Break the task into the work you might miss
A vague task invites a vague estimate. Instead of asking, “How long will the application take?” write down the main components. Depending on the task, these might include preparation, finding information, doing the central work, reviewing it, making revisions, coordinating with someone or waiting for a response.
Then estimate each component and add them up. For example, “prepare a short presentation” might involve gathering material, choosing a structure, making the slides, checking the content and revising after feedback. The point is not to create an elaborate project plan for every small job. It is to expose steps that disappear when a task is treated as one solid block.
Experiments by Kruger and Evans found that asking people to enumerate the elements of a task increased their predicted completion times and reduced the planning fallacy. Their work on unpacking tasks supports a practical habit: spell out the parts before settling on an estimate.
A breakdown is not a guarantee of accuracy. You can still underestimate each step, forget an important dependency or assume that everything will go smoothly. Use the list as a way to make the forecast more complete, not as proof that it is correct.
For a quick estimate, a few meaningful steps may be enough. Add detail when the task is unfamiliar, has several handoffs or carries a deadline that depends on other people. If a step is uncertain, mark the assumption instead of hiding it inside a confident-looking total.
Use the outside view: compare with what happened before
After listing the steps, ask a different question: how long did similar tasks actually take? This is the outside view. It shifts attention from the task as you hope it will unfold to the pattern in comparable work you have already completed.
Choose comparisons carefully. A short routine email may not be a useful guide to a sensitive message that requires several reviews. A small room reset may not resemble a larger clear-out. Think about what makes tasks comparable: their size, uncertainty, number of people involved and the amount of review or waiting they require.
A lightweight record can make this comparison easier. For each task, note:
- The type of task and what “done” meant.
- Your initial estimate.
- Actual active working time.
- Elapsed calendar time, if timing to completion matters.
- Major interruptions, delays, revisions or unexpected steps.
You do not need a detailed log of every minute. A brief note after finishing can be more useful than relying on memory weeks later. Keep different types of time separate: if you worked for two hours across three days, that is not the same as two hours from start to finish on one afternoon.
When you plan a similar task, compare your new breakdown with those records. If previous tasks of that kind regularly needed a review round or time to collect information, include those parts in the forecast. Adjust by task type rather than applying the same percentage to everything. The evidence here does not establish a universal rule such as “add 50%” or “double every estimate.” Your own relevant history is a better starting point, while still leaving room for uncertainty.
Use a range when you have little history
Some work is genuinely new, unusual or hard to compare with anything you have done before. In that case, a single precise number can suggest more certainty than you have. Estimate a plausible range instead, and state what would make the task fall near the shorter or longer end.
For example, you might expect the central work to take a few hours if the information is readily available, but longer if you need to find and verify it. The useful part is not choosing a perfect range; it is identifying what the estimate depends on. That makes it easier to revise the forecast when those assumptions change. If the backlog keeps growing, estimates are only one part of understanding why the task list stays full.
For a task with several distinct steps, ranges can be assigned to the uncertain steps rather than making the whole estimate opaque. You might know that drafting is fairly predictable but that feedback could add an unknown amount of calendar time. Keep those uncertainties visible. A single total can conceal where the risk lies.
As new information arrives, revisit the estimate. If you discover that the source material is incomplete, the task has changed. Updating the plan is not a failure of forecasting; it is a response to better information.
Plan the start without treating it as a time-estimate fix
A realistic estimate only helps if there is a workable opportunity to begin. Turn the first step into a specific plan: “At 10 a.m. at my desk, I’ll open the notes and list the sources I need.” A plan that names when, where and what to do next is more actionable than “work on the report later.”
In an experiment on an assignment, Koole and van ’t Spijker found that participants who specified where and when they would work showed less optimistic bias in their completion predictions. Fewer interruptions partly explained the effect. The study examined a specific implementation-intention intervention; it does not show that a start plan will correct every estimate or work equally well for every task.
Broader research also supports the usefulness of specific if–then plans for helping people move from intention to action. A 2024 meta-analysis covered 642 tests across cognitive, affective and behavioural outcomes. That evidence is relevant to planning and follow-through, but it is not a direct estimate of how long everyday tasks will take. Use a concrete start plan to support action, not as a substitute for breaking down work and checking past results.
If your estimate assumes uninterrupted focus, be honest about whether that is realistic. You might need to protect a work period, reduce avoidable interruptions or choose a time when the first step is feasible. That improves the plan’s fit with your actual schedule; it does not make the task itself take less time.
Turn each finished task into a better next estimate
After completing a task, compare the outcome with the original forecast. Look for the source of the gap rather than labelling the estimate simply “wrong.” Did you leave out preparation? Did review take longer than expected? Was active work close to the estimate, but a dependency stretched the calendar time? Did interruptions break up the available work periods?
Use the answer to adjust the next estimate for a similar task. If you repeatedly forget the time needed to gather materials, add that step to the breakdown. If a certain kind of task often expands after review, account for that history before setting a deadline. If the delay came from an unusual event, record it without assuming it will always happen.
Keep the process proportionate. For a familiar, low-stakes task, a quick mental comparison may be sufficient. For work with a firm deadline or several uncertain parts, write down the steps, assumptions and relevant past examples. The more costly a missed estimate would be, the more useful it is to make the reasoning visible.
Over time, the goal is not to become certain about the future. It is to replace a best-case guess with a forecast informed by the task’s components, your own experience and the conditions under which you will work. When a task surprises you, that result is information for the next estimate—not a reason to demand perfect predictions from yourself.