Decompose scope, then sequence the work
Gantt Chart vs PERT Chart: When to Use Each
One draws time, the other draws logic. How Gantt and PERT differ, the three-point estimate maths, and which to bring to which meeting.
On this page
A Gantt chart is a calendar view: bars on a timeline. A PERT chart is a network view: boxes and arrows showing what must precede what, with probabilistic durations. They answer different questions and most schedules need both at different moments.
Head to head
| Gantt chart | PERT chart | |
|---|---|---|
| Shape | Horizontal bars against dates | Nodes and arrows — a dependency network |
| Shows time to scale | Yes | No — geometry is logical, not proportional |
| Duration treatment | Single fixed estimate | Three-point: optimistic, most likely, pessimistic |
| Best question | What is happening in week six? | What genuinely depends on what, and how uncertain is it? |
| Audience | Sponsors, clients, delivery teams | Planners, estimators, risk reviewers |
| Weakness | Hides logic and uncertainty behind neat bars | Unreadable for stakeholders who want a date |
| When to build it | After the network is agreed | First, while sequencing the work |
The PERT estimate, worked
PERT weights the most likely case four times, then divides by six: Expected = (Optimistic + 4 × Most likely + Pessimistic) ÷ 6. Standard deviation = (Pessimistic − Optimistic) ÷ 6.
| Activity | Optimistic | Most likely | Pessimistic | PERT expected | Std dev |
|---|---|---|---|---|---|
| Visual design | 6d | 8d | 16d | 9.0d | 1.67d |
| Front-end build | 7d | 9d | 17d | 10.0d | 1.67d |
| Integration & QA | 3d | 5d | 13d | 6.0d | 1.67d |
| Cutover | 1d | 2d | 3d | 2.0d | 0.33d |
Notice what happened: every single-point estimate grew. Visual design was carried as 8 days on the Gantt chart; weighted for its long tail it is 9. Across a chain those additions are the difference between a plan that survives and one that needs a rescue.
Turning deviations into a confidence range
Add the expected durations along the critical path for the mean. Combine deviations as the square root of the sum of squares, then apply the normal rule: roughly 68% of outcomes fall within one deviation, 95% within two.
- Chain mean: 9.0 + 10.0 + 6.0 + 2.0 = 27 days.
- Chain deviation: √(1.67² + 1.67² + 1.67² + 0.33²) ≈ 2.9 days.
- About 68% confidence: 24–30 days. About 95% confidence: 21–33 days.
- Commit externally near the upper bound, plan internally to the mean, and hold the gap as schedule reserve you can name.
Which one to bring to the meeting
| Meeting | Bring | Because |
|---|---|---|
| Planning workshop | PERT network | You are debating logic, not dates |
| Estimation review | PERT three-point table | Uncertainty is the subject |
| Steering committee | Gantt chart with the critical path highlighted | They need the date and the risk in one picture |
| Daily delivery | Gantt chart filtered to the next two weeks | Bars beat networks for near-term coordination |
| Recovery planning | Both | Network to find the compression options, bars to sell them |
A practical hybrid workflow
- 1List activities with no dates attached.
- 2Draw the network by hand or on a whiteboard — arrows only where the work genuinely cannot start earlier.
- 3Collect three-point estimates per activity from the people who will do the work.
- 4Compute PERT expected durations and the chain deviation for the critical path.
- 5Load the expected durations into your Gantt tool and highlight the critical path.
- 6Publish the confidence range alongside the committed date, and re-run after every actuals update.
The PM Delivery Toolkit turns everything in this guide into fill-in templates — $25, one payment.
Get five PM habits that quietly save projects
Baselines, critical path discipline, RAID logs that people read, range estimating and status packs sponsors act on — sent straight to your inbox, free.
One email, no spam, unsubscribe any time.
Discussion
Share how you applied this in practice, ask a question, or add a tip for other readers.
Loading comments…