How to Estimate Task Durations: 5 Practical Methods (With Examples)

November 12, 2026

GanttMaster

Plan Smarter. Execute Faster.

Interactive Project Management Tool, real-time collaboration.

⭐ 4.8 Rating • Trusted by modern teams worldwide

Every schedule is built on guesses about how long things will take. When those guesses are too optimistic, the whole plan slides one task at a time. When they are padded everywhere, the project looks slower and more expensive than it needs to be. Good duration estimates sit in between: realistic, explained, and easy to update when you learn more.

This guide covers what a task duration estimate is, five practical estimating methods (including the three-point and PERT formulas), a worked example, and the mistakes that most often throw estimates off. It is written for project managers, construction and field teams, healthcare operations staff, and small business owners who plan work on a timeline.

What is a task duration estimate?

A duration estimate is the number of working days (or hours) a task will take from start to finish. It is different from two related numbers that people often mix up:

  • Effort (work): the total person-hours needed. Painting a room might take 16 person-hours.
  • Duration: the calendar time it takes, given who is doing it. With two painters working 8-hour days, those 16 person-hours become 1 day.
  • Elapsed time: calendar days including weekends, holidays, and waiting time, such as plaster drying or a lab result coming back.

Your schedule needs durations, but you usually work them out from effort and the people available. Keeping the three apart is the first step to better estimates.

Before you estimate: break the work down

Large, vague tasks are hard to estimate. "Renovate the clinic waiting room" could mean two weeks or two months. "Remove old flooring", "Patch and prime walls", and "Install new seating" are much easier to size. As a rule of thumb, break work down until each task can be finished by one person or crew in a short period, often a few days, and has a clear finished state.

For each task, write down three things before you put a number on it: what "done" means, who will do it, and what it depends on. Estimates without those details are guesses about the wrong thing.

5 methods to estimate task durations

1. Expert judgment

Ask the person who will do the work, or someone who has done it before. A site foreman knows how long it takes his crew to frame a partition wall; a nurse manager knows how long it takes to train staff on a new intake process.

Best for: familiar work and teams with experience. Watch out for: optimism, and estimates given under pressure to please. Ask "what would make this take longer?" before accepting the number.

2. Analogous estimating

Use the actual duration of a similar past task and adjust for differences. If wiring the last shop fit-out took 6 days and this unit is roughly a third larger, start at about 8 days and adjust for anything else that differs.

Best for: early planning, when details are thin. Watch out for: tasks that only look similar. A past project with a different crew, site, or approval process may not be a fair comparison.

3. Parametric estimating

Multiply a known rate by the quantity of work. If a crew typically tiles about 20 square meters a day and the job is 120 square meters, the estimate is 6 days. If a team processes about 40 supplier invoices a day and there are 200 to review, plan for 5 days.

Best for: repetitive, measurable work. Watch out for: rates taken from ideal conditions. Use your own team's real rates where you have them, and adjust for site access, weather, or interruptions.

4. Three-point estimating (and PERT)

Instead of one number, give three:

  • Optimistic (O): how long it takes if everything goes well.
  • Most likely (M): the realistic expectation.
  • Pessimistic (P): how long it takes if reasonable things go wrong (not a disaster).

Then combine them. There are two common formulas:

  • Triangular (simple average): (O + M + P) / 3
  • PERT (beta) estimate: (O + 4M + P) / 6, which gives the most likely value more weight.

PERT also gives a rough measure of uncertainty, the standard deviation: (P minus O) / 6. A wide spread tells you the task is risky and worth a closer look.

Example: installing a medical equipment unit. O = 2 days, M = 4 days, P = 9 days.

  • Triangular: (2 + 4 + 9) / 3 = 5 days
  • PERT: (2 + 16 + 9) / 6 = 4.5 days
  • Standard deviation: (9 minus 2) / 6, roughly 1.2 days

Best for: tasks with real uncertainty, such as first-time work, outside vendors, or inspections. Watch out for: treating the result as precise. It is a better-reasoned estimate, not a guarantee.

5. Bottom-up estimating

Break a task into subtasks, estimate each one, and add them up, taking into account which steps can overlap. "Set up new point-of-sale system" might become: order hardware (wait 5 days), install and configure (1 day), load products and prices (2 days), train staff (1 day), and a test day. Some of these overlap; the ordering wait can run while products are loaded.

Best for: important or complex tasks where accuracy matters. Watch out for: the time it takes, and adding a buffer to every subtask, which stacks up fast.

A worked example: small office fit-out

Here is an illustrative example of how the methods work together. The numbers are for demonstration only, not benchmarks.

  • Strip out old partitions: expert judgment from the foreman, 2 days.
  • Build new partitions: parametric, 30 meters of wall at about 10 meters a day, 3 days.
  • Electrical rough-in: analogous to a similar past unit, 4 days.
  • Inspection: three-point, O = 1, M = 3, P = 8 working days of waiting, PERT about 3.5, so plan 4 days.
  • Drywall, tape and finish: bottom-up from hang (2 days), tape (2 days), finish and sand (2 days), with taping starting once hanging is underway, 5 days.
  • Paint and flooring: parametric from area, 4 days.

Once each task has a duration, link them with dependencies. Inspection can only start when rough-in finishes (Finish-to-Start). Taping can start shortly after hanging starts (Start-to-Start). Final cleaning cannot finish until painting finishes (Finish-to-Finish). The durations plus the links give you the end date, and they show which tasks push it out if they run long.

How to add contingency without padding everything

Most people quietly add a bit of extra time to each task. The trouble is that work tends to expand to fill the time available, so hidden padding gets used up even when it is not needed. A cleaner approach:

  • Estimate each task honestly, using the most likely or PERT value.
  • Identify the tasks with the widest spread between optimistic and pessimistic.
  • Add a visible buffer at the end of a phase, or right before a fixed date, instead of inside every task.
  • Record the reasons for the buffer (an inspection, a long-lead delivery) in your risk register so it can be released when the risk passes.

Common duration estimating mistakes

  • Confusing effort with duration. Forty hours of work is not a week if the person is only on the task half the time.
  • Forgetting waiting time. Curing, drying, approvals, deliveries, and inspections take calendar time even when nobody is working.
  • Estimating for someone else. The person doing the work should at least review the number.
  • Assuming full availability. People have meetings, other projects, shifts, and leave. A healthcare team covering patient care cannot spend whole days on a project.
  • Ignoring dependencies. A task can be estimated perfectly and still start late because its predecessor slipped.
  • Never comparing estimates to actuals. Without looking back, the same misjudgments repeat on every project.

Task duration estimating checklist

  • Work broken into tasks with a clear "done" state.
  • Owner identified for each task, and their availability checked.
  • Effort and duration kept separate.
  • Waiting time (approvals, deliveries, curing, inspections) included.
  • A method chosen that fits the task: judgment, analogous, parametric, three-point, or bottom-up.
  • Three-point estimates used for uncertain tasks.
  • Dependencies linked (FS, SS, FF) so the timeline reflects real sequence.
  • Contingency added visibly, not hidden in every task.
  • Estimates reviewed with the people doing the work.
  • Actual durations compared to estimates as the project runs.

Putting your estimates on a timeline in GanttMaster

Estimates are only useful when they turn into a schedule your team can see and update. GanttMaster, a project management app for iPhone and iPad, helps you take durations from a list to a working plan:

  • Start and end dates on every task. Enter the dates that follow from your estimate, and assign the task to a user, a team, or both.
  • Tasks and subtasks. Use subtasks for bottom-up estimates, so the parts and the whole stay together.
  • Dependencies. Link tasks with Finish-to-Start, Start-to-Start, and Finish-to-Finish dependencies so the sequence is clear.
GanttMaster task details on iPhone showing the task title, description, start date, end date, and assigned team.
Example project: each task carries its own start date, end date, and assignment.
GanttMaster Add Dependency screen on iPhone showing Finish-to-Start, Start-to-Start, Finish-to-Finish and Start-to-Finish options.
Example project: choosing a dependency type when linking two tasks.
  • Interactive Gantt chart. See every duration as a bar on the timeline, with bars colored by task status, so tasks running behind stand out.
  • Today view with reminders. Team members see the tasks due today and can set reminders for things like booking an inspection.
  • Risk tracking. Log the uncertain tasks behind your buffers with likelihood, impact, mitigation, an owner, and a due date.
  • Project templates. Reuse a past project's structure, so your next estimate starts from real work you have already planned.
GanttMaster Gantt chart on iPad showing task bars across daily columns with dependency arrows.
Example project: task durations shown as bars on the Gantt chart, linked by dependencies.

There is no per-user pricing, so the people who give you the estimates can see and update the schedule too. GanttMaster is free to download with a 1-week trial, then $5.99 a month or $34.99 a year.

Better estimates do not come from a perfect formula. They come from breaking work down, asking the people who do it, being honest about uncertainty, and checking your numbers against what really happened. Do that on each project and your schedules get more reliable every time.

GanttMaster logo

Plan it on your iPhone or iPad

Full disclosure: we make GanttMaster, an iPhone and iPad app with FS/SS/FF dependencies, tasks and subtasks, checkpoint tasks, a Today view with reminders, team assignment, risk and budget tracking, and no per-user pricing. It is free to download with a 1-week trial, then $5.99/month or $34.99/year.

Download on the App Store