Why a familiar task takes longer again
The planning fallacy is a tendency to underestimate task duration even when past experience shows similar work took longer. To refine an estimate, compare several comparable completed tasks, use actual work time as a reference, and separately account for waiting for data and acceptance. Justify a buffer through specific uncertainty rather than a universal percentage.
When imagining future work, it is easy to see a smooth sequence: open data, write conclusions, send the report. Checking a disputed number or obtaining additional approval may fall outside that scenario. A past delay can conveniently be explained as an exception: ‘I was simply unlucky that time.’ If similar causes recur, include them in the next estimate.
One missed deadline does not prove a persistent planning fallacy. Scope may have changed or a new dependency appeared. Start by clarifying the promised result and which past cases genuinely resemble the current one.
First define exactly what must be ready
Estimation needs a task with a clear boundary. ‘Make a report’ can mean a draft, checked document, or approved version. These results take different amounts of time.
- Result: what the recipient gets, in which scope and format.
- Starting conditions: which data, permissions, and decisions must already be available.
- End of your work: which checks are included before handoff.
- Acceptance: who checks the document, when they reply, and which corrections are agreed.
For example: ‘Prepare a report from two finished tables, write three conclusions, check numbers, and give the document to the manager.’ Managerial approval is a separate event. Until the reply deadline is agreed, final acceptance time cannot confidently be promised.
Compare actual time from similar completions
Find past tasks with similar scope, quality requirements, and inputs. Include available comparable cases even when inconvenient for the initial forecast. A quick report from a ready template is a poor reference for a report requiring new calculations.
Separate active time from calendar waiting. Active time means minutes preparing, checking, and correcting the result. Waiting is the interval before data or a reviewer’s reply arrives. An hour of waiting does not necessarily consume an hour of your work but may move completion to a later date.
| Case | Initial estimate | Active work | Waiting for acceptance | Conditions |
|---|---|---|---|---|
| Report A | 60 minutes | 100 minutes | Until the next workday | Two tables, three conclusions, number checking |
| Report B | 70 minutes | 125 minutes | Two workdays | Same scope; a discrepancy needed clarification |
| Report C | 60 minutes | 110 minutes | Until the next workday | Same scope; one wording correction |
In this example, a new 60-minute estimate conflicts with all three completions. Actual active work took 100–125 minutes; the middle case is 110 minutes. Those 110 minutes can form a working baseline while retaining the full range visibly. Three observations give a preliminary reference rather than a statistically reliable probability of meeting a deadline.
If no records exist, reconstruct available information from the calendar and document history, noting imprecision. Start recording actual time during the next performance. A short tracking method is described in ‘Time tracking’: https://sinqly.ru/blog/khronometrazh-vremeni-prostoy-uchet
Add a buffer with a clear basis
A buffer covers the uncertain part of work. Include a known stage in the main estimate: mandatory number checking must not be hidden in reserve. If actual times from past cases already include checking and ordinary corrections, do not add them again.
In the fictional example, Irina chooses a 110-minute baseline and a 15-minute buffer to allow a case like the longest of the three comparable ones. This produces 125 minutes of active time. It is the author’s decision for this example: it does not cover an unknown serious problem or guarantee completion.
- Name the risk, such as a possible source-data discrepancy.
- Check whether the chosen baseline already reflects it.
- Record which scenario the buffer covers and which past case supports it.
- Specify a review trigger, such as new requirements or missing data.
If a new task is much more complex than past ones, increasing the buffer does not make the comparison appropriate. Estimate the new stage separately or first examine a small part of the unknown work. When delay has substantial consequences, discuss the range and conditions with the recipient before promising timing.
Separate promised handoff from promised acceptance
Now place the required work into available time. 125 minutes of active work do not mean the result will be ready 125 minutes after the day starts: meetings and other commitments may separate work blocks.
Continue the fictional example. Irina has Monday blocks from 10:00 to 11:00 and 14:00 to 15:05, totaling 125 minutes. The baseline already includes preparation, checking, ordinary correction, and handoff. The manager separately agreed to review by Tuesday at 12:00. Irina reserved Tuesday 14:00–14:30 for new comments.
| Event | Deadline | Condition | What to review |
|---|---|---|---|
| Hand over the checked report | Monday, by 15:05 | All source data ready; scope unchanged | If data changes, refine the estimate before handoff |
| Receive comments or confirmation | Tuesday, by 12:00 | The manager agreed on review timing | If the reply is late, agree on a new acceptance date |
| Hand over a corrected version if needed | Tuesday, by 14:30 | Comments arrive on time and fit within 30 minutes | Estimate a larger correction scope separately |
| Obtain final acceptance | After another review | If corrections were made, repeat-review timing is agreed separately | Do not declare a corrected version automatically accepted |
The interval before the manager replies affects calendar timing but is not added in full to Irina’s active work. New corrections are estimated separately from ordinary fixes already included. If another review is needed, agree on its date too.
To place this estimate among commitments, use ‘How to plan your day’: https://sinqly.ru/blog/kak-planirovat-den-effektivno. For work spanning several days, distribute the load across the week next: https://sinqly.ru/blog/planirovanie-nedeli-nagruzka-i-rezerv. Related approaches are collected in ‘Productivity’: https://sinqly.ru/topics/produktivnost
What the research examined and does not promise
Buehler, Griffin, and Ross’s 1994 work studied completion forecasts, attention to future scenarios, and use of past experience. In Study 4, participants were asked to connect relevant past experience with a new forecast. Under those experimental conditions, the observed optimistic bias disappeared.
This is a limited finding from one experiment. It does not establish universal work-estimate accuracy, a buffer percentage, or guaranteed benefits from any tracking method. This article’s process and tables are the author’s practical application of comparison with past experience; their effectiveness was not separately tested.
A single-task estimation template
Copy these lines onto paper or into any editor and fill them in before the next deadline promise.
- Result and boundary: exactly what I deliver and which checks are included.
- Comparable cases: scope, conditions, and actual active time for each.
- Estimate baseline: chosen reference and why.
- Buffer: duration, risk covered, and what is already included.
- Calendar: available work blocks and result handoff time.
- Acceptance: who replies, when they review, and how new corrections are agreed.
- Review: which change triggers warning the recipient and refining timing.
After completion, record actual active time, waiting, and the reason for any difference from the estimate. Mark scope changes separately: the next forecast must rely on comparable work. This turns a specific completion into material for the next decision.
Sources
Frequently asked questions
Should I always double the initial estimate?
There is no universal multiplier here. Compare actual times from similar tasks, check work components, and justify reserve through specific uncertainty. A multiplier without that examination can underestimate or overestimate.
Can I exclude the longest past case?
Yes, if it is not comparable for a clear reason, such as a separate new stage. Record the exclusion reason. If a risk that could recur caused the delay, retain the case in the comparison.
How do I estimate a task I have never done?
Split it into familiar and unknown parts. Use past cases for familiar work; clarify requirements or run a small trial for unknown work first. State that the estimate is preliminary and when it will be reviewed.
What if the recipient gives no review deadline?
Agree on handoff timing and explicitly state that acceptance timing remains unknown. Propose a review time or follow-up. A buffer in your own work cannot eliminate waiting for another person’s reply.
Turn an idea into action
Was this helpful?
Rate the article and tell us what was missing. Your feedback goes to the Sinqly team and will not be published on the site.
Bullet Journal: quick notes and deliberate task migration without decoration
A minimal Bullet Journal: tasks, events, notes and deliberate migration. A monthly review example without decorative requirements.
The 1-3-5 method for planning your day
The 1-3-5 method: a simple daily planning system. 1 main task, 3 medium tasks, 5 small tasks. How to use it, examples, and integration with trackers and AI planners.
GTD: a planning system explained simply, with a diagram and example
How the GTD planning system works: a five-step diagram, processing your inbox, and a project example. Where to start, where to record the next action, and when you need a calendar.