Learn how critical path method (CPM) works with Gantt charts. Calculate float, identify critical tasks, and use CPM to protect your project schedule from delays.
Most project managers use the phrase "critical path" without applying it rigorously. They mean "important tasks" when critical path has a precise meaning: the longest sequence of dependent tasks through a project, which determines the earliest possible completion date. Any delay on this path delays the project. Tasks off this path have float — they can slip without affecting the end date.
Understanding this distinction changes how you manage a project. When you know exactly which tasks are critical, you focus your risk management, your contingency time, and your daily attention on the right work. When you treat every task as critical, you spread your attention thin and discover delays too late to recover.
This guide explains how critical path method (CPM) works, how to calculate it in a Gantt chart, and how to use it to run tighter, more predictable projects.
CPM was developed in the late 1950s independently by DuPont (for industrial plant maintenance) and the U.S. Navy (for the Polaris submarine program). The fundamental insight is simple: not all tasks affect the project end date equally. Some tasks have dependencies that create a chain from start to finish. The longest such chain is the critical path.
Three terms to understand before calculating anything:
Float (also called slack): The amount of time a task can be delayed without delaying the project. Tasks on the critical path have zero float. Tasks off the critical path have positive float.
Early start / Early finish: The earliest a task can start or finish, given its predecessors.
Late start / Late finish: The latest a task can start or finish without delaying the project end date.
Float = Late Start − Early Start (or equivalently, Late Finish − Early Finish).
Tasks where float equals zero are on the critical path.
To find the critical path, you need two things: a complete task list with durations and a complete set of predecessor relationships. Both live in your Gantt chart.
Start with every task in your project. Assign a realistic duration to each one in working days. For this example, a software release:
| Task | ID | Duration |
|---|---|---|
| Requirements finalized | A | 5 |
| Architecture design | B | 8 |
| Database schema | C | 4 |
| Backend API development | D | 15 |
| Frontend development | E | 12 |
| Unit testing — backend | F | 5 |
| Unit testing — frontend | G | 5 |
| Integration testing | H | 7 |
| User acceptance testing | I | 5 |
| Staging deployment | J | 2 |
| Production deployment | K | 1 |
| Task | Predecessor(s) |
|---|---|
| A | — (project start) |
| B | A |
| C | B |
| D | C |
| E | B |
| F | D |
| G | E |
| H | F, G (both must complete) |
| I | H |
| J | I |
| K | J |
Work from project start to project end, calculating early start and early finish for each task.
Project duration: 52 working days.
Notice that Integration Testing (H) must wait for both F (Day 37) and G (Day 30) to finish. It waits for whichever finishes last — Task F on Day 37.
Work from project end back to start, calculating the latest each task can start and finish without pushing the end date beyond Day 52.
Float = Late Start − Early Start:
| Task | Early Start | Late Start | Float | Critical? |
|---|---|---|---|---|
| A | 0 | 0 | 0 | Yes |
| B | 5 | 5 | 0 | Yes |
| C | 13 | 13 | 0 | Yes |
| D | 17 | 17 | 0 | Yes |
| E | 13 | 20 | 7 | No |
| F | 32 | 32 | 0 | Yes |
| G | 25 | 32 | 7 | No |
| H | 37 | 37 | 0 | Yes |
| I | 44 | 44 | 0 | Yes |
| J | 49 | 49 | 0 | Yes |
| K | 51 | 51 | 0 | Yes |
Critical path: A → B → C → D → F → H → I → J → K
Frontend development (E) and frontend unit testing (G) have 7 days of float each. The frontend track can slip up to 7 days without affecting the project end date. If the frontend developer calls in sick for a week, the project doesn't slip. If the backend developer calls in sick for a week, the project slips one week.
This is the operational insight CPM provides.
On a Gantt chart, the critical path is typically displayed in a distinct color — red is common — or with bolded bars. Gantt-chart.io highlights critical path tasks automatically when you enter dependency relationships.
What you see in a Gantt chart with CPM applied:
When a task on the critical path slips, the chart recalculates: all downstream tasks shift right, the project end date moves right by the same number of days, and the new critical path is recalculated if float relationships have changed.
When a non-critical task slips within its float, nothing else moves. The chart shows the slip consuming float, and the project end date is unchanged.
Once you have the critical path, focus your risk management on critical tasks. Where are the uncertainties? Which tasks have assumptions that might not hold? Which are dependent on external parties (vendors, regulatory approvals, client decisions)?
Non-critical tasks carry risk too, but float is your buffer. If a non-critical task with 7 days of float has a 3-day uncertainty, you have 4 days of remaining buffer. If a critical task has a 3-day uncertainty, you have zero buffer.
"Crashing" means compressing the critical path by adding resources to critical tasks. Key principle: you can only compress the project end date by compressing critical tasks. Adding resources to non-critical tasks doesn't help.
If you need to deliver 5 days earlier, look at critical tasks that can be shortened with additional resources — overtime, additional staff, parallel work within the task. The backend development task (15 days) might compress to 10 days with two developers instead of one. That would compress the critical path from 52 to 47 days.
Crashing costs money (more resources) or creates risk (parallel work that could conflict). Use it deliberately, on critical tasks only.
Fast tracking means overlapping tasks that were planned sequentially. Instead of waiting for backend API development to completely finish before integration testing starts, you begin integration testing on the completed portions of the API.
Fast tracking compresses the schedule without necessarily adding cost, but it increases coordination complexity and rework risk. It works best when the overlapped tasks have a natural split — the first 60% of backend development is stable enough to test against while the final 40% is still in development.
During project execution, monitor float on non-critical tasks. When a non-critical task starts consuming its float, you get early warning:
A Gantt chart with live float tracking gives you this early warning automatically as task status is updated.
Construction: Permit approval is almost always on the critical path. Procurement of long-lead materials (structural steel, custom windows, specialized equipment) is frequently on the critical path. Interior finishes often have float because multiple trades can work simultaneously once the building is enclosed.
Software development: The critical path typically runs through the most complex feature dependency chain — not through every feature, but through whichever sequence gates the release. Testing phases are frequently critical because they're downstream of everything and cannot be paralleled away.
Marketing campaigns: The critical path often runs through legal review and creative approval cycles because these are sequential (legal must review final copy, not drafts) and their durations are outside the marketing team's control.
Event production: The critical path typically runs through venue contracting (everything else depends on venue confirmation), A/V production, and the final run-of-show approval. Sponsor deliverables often have float because they're not on the path to the event itself — the event happens whether or not every sponsor's logo is perfect.
Ignoring external dependencies. A task that depends on a client review, a vendor delivery, or a regulatory approval is on the critical path if its timing is constrained. Leaving these off the chart makes the critical path look shorter than it is.
Over-optimistic duration estimates. CPM gives you precise answers based on your input. If your duration estimates are wishful, your critical path calculation is wrong. Use historical data from similar projects, not first-principles optimism.
Not updating the plan. CPM requires current data. If task A is 3 days late and you don't update the Gantt chart, all your float calculations are wrong. Update task completion status every week; recalculate the critical path when significant changes occur.
Treating all near-critical tasks as critical. Tasks with 1–2 days of float are nearly critical and deserve attention. Tasks with 10+ days of float are not critical even if they feel important. The math is the guide, not the instinct.
Open gantt-chart.io and build your project:
CPM turns project management from intuition to analysis. When you can answer "which tasks determine my project end date?" with a Gantt chart that shows the evidence, you make better decisions about where to spend your time and where to accept risk.
The critical path is not a constraint. It's information — and information is how you protect your project's end date.