Skip to content

Plan and Control the Project Schedule

Step 34 of 57

PMP Schedule Planning: Estimation, Critical Path, and Float

Introduction

Project schedule planning connects activity estimates with the dependencies and calendars that determine the finish date. For example, a supplier delivery can set the earliest start for testing even when the team is ready sooner. This lesson covers schedule planning in Process Task 8 of the July 2026 PMP Examination Content Outline (ECO). Process accounts for 41% of exam coverage. You will estimate durations, find the critical path and float, and forecast releases from delivery data.

Learning objectives

  • Select a schedule approach that matches the development approach (predictive, agile, hybrid).
  • Estimate durations with analogous, parametric, bottom-up, three-point, and relative techniques.
  • Use benchmarks, historical data, and expert input to improve estimates.
  • Identify task dependencies (finish-to-start and others, leads and lags) and the types of dependency.
  • Compute early and late dates, total float, free float, and the critical path.
  • Coordinate the schedule with other projects, operations, and resource calendars.
  • Plan releases, iterations, and the roadmap for adaptive work.

1. Choose the schedule approach

The schedule approach follows the development approach selected during planning.

Approach How the schedule is built Typical artifacts
Predictive Build and approve a schedule from the activities and estimates available. Refine later work under the planning and change rules. Work breakdown structure (WBS), activity list, network diagram, Gantt chart, milestone chart
Iterative / incremental Plan by phases or increments; detail the near term, outline the rest Phase plan, feature roadmap
Agile Time-boxed iterations pulling from a prioritized backlog; or flow-based delivery Product roadmap, release plan, iteration plan, kanban board
Hybrid Predictive frame for contracts, regulation, or hardware; agile inside for development Master milestone schedule with iteration plans inside

Rolling wave planning is common: work in the near term is planned in detail, and later work stays at a higher level until more is known. It is progressive elaboration applied to the schedule.

Example: two schedules in one project

A company builds a smart thermostat. The hardware team has a supplier lead time of 14 weeks for chips and a safety certification slot booked for week 30, so it uses a predictive network with fixed milestones. The firmware and app teams work in two-week sprints with a roadmap. The project manager keeps one integrated milestone schedule: chips arrive week 14, the firmware feature-complete release is week 24, certification starts week 30. Inside those milestones the software teams plan by sprint. Check the dependencies between the streams. For example, verify that the firmware build needed for certification can be tested on the first hardware batch.

2. Define activities and sequence them

In a predictive project, work packages from the WBS are decomposed into activities that can be estimated, assigned, and tracked. Adaptive teams can split larger backlog items into smaller items and plan the tasks needed for near-term delivery.

Dependency relationships

Type Meaning Example
Finish-to-start (FS) Successor cannot start before predecessor finishes Pour concrete, then build frame
Start-to-start (SS) Successor cannot start before predecessor starts Start an integration test setup after the build work begins
Finish-to-finish (FF) Successor cannot finish before predecessor finishes Documentation finishes when testing finishes
Start-to-finish (SF) Successor cannot finish before predecessor starts The old system shuts down only after the new one starts

Lead advances the successor (an overlap): start draft documentation two days before the design review finishes, if rework is acceptable. Lag delays the successor: wait 3 days for the concrete to cure.

Dependency types by cause

  • Mandatory (hard logic): required by physics or contract. A test on a physical unit requires the unit to be available.
  • Discretionary (soft logic): preferred practice or best practice; can be changed to compress the schedule.
  • External: relies on something outside the project, such as a regulator's approval or a supplier delivery.
  • Internal: relies on another project activity or team.

Understanding the cause matters for compression: check which activities can overlap safely and who can agree to any change. External and mandatory are separate classifications.

3. Estimating

An estimate is a quantitative assessment of the likely amount or outcome. Estimates should state the basis of estimate (assumptions, constraints, range, and the basis for any claimed confidence level).

Estimating techniques

Technique How it works Strengths Limits
Analogous Uses actual duration or cost of a similar past project Fast, low cost Less accurate; depends on similarity
Parametric Applies a rate to a measure, such as hours per unit Accurate when data are reliable and scalable Requires validated rates
Bottom-up Estimates smaller pieces of work, then aggregates them Makes assumptions visible at a detailed level Requires enough detail and time
Three-point Uses optimistic, most likely, pessimistic values Reflects uncertainty Needs judgement for the three values
Expert judgement Experts estimate from experience Useful with unique work Biased by optimism
Wideband Delphi / planning poker People estimate independently, compare assumptions, and may estimate again Reduces anchoring, builds consensus Needs a facilitator and time
Relative (story points, t-shirt sizes) Compare effort to reference items Good for adaptive work Not directly in hours

Three-point estimates

  • Triangular: E = (O + M + P) / 3
  • Conventional PERT: weighted estimate = (O + 4M + P) / 6, spread approximation = (P − O) / 6, and approximate variance = spread squared.

Worked example: PERT estimate

Scenario. An activity "configure the data pipeline" has optimistic 8 days, most likely 12 days, pessimistic 22 days. What are the conventional PERT weighted duration and spread approximation?

Step 1: PERT expected value. E = (8 + 4 x 12 + 22) / 6 = (8 + 48 + 22) / 6 = 78 / 6 = 13 days.

Step 2. Estimate the spread. (22 − 8) / 6 = 14 / 6 ≈ 2.33 days.

Step 3. Interpret the estimate. A range of plus or minus one or two spread estimates has no stated confidence level here. To estimate the chance of meeting a date, use a justified distribution and schedule risk model.

Step 4: compare with the triangular estimate. (8 + 12 + 22) / 3 = 14 days. PERT weights the most likely value more heavily, so it is lower here.

Why other choices are weaker. Using only the most likely value (12) ignores risk of delay. The pessimistic input describes one end of the range. Record uncertainty and allow for risk explicitly when preparing the schedule.

Estimating in agile teams

Some adaptive teams use story points and estimate with techniques such as planning poker or affinity grouping. The velocity (points completed per iteration) turns relative estimates into a forecast: if the backlog holds 240 points and the stable velocity is 30 points per iteration, the forecast is 240 / 30 = 8 iterations. At 26–34 points per iteration, the same backlog would take about 8–10 whole iterations if delivery and scope stay comparable. Split large items where useful or use a short investigation to reduce uncertainty.

Benchmarks and historical data

Organizational process assets such as past project schedules, productivity rates, lessons learned, and industry benchmarks improve estimates. Compare your estimate against benchmarks: if a similar mobile app took 24 to 30 weeks in four previous projects and your plan shows 12 weeks, either the scope differs or the estimate is optimistic. Document the reasoning. Beware of using data from teams with different skills or technology as if they were identical.

Cognitive traps

  • Anchoring: the first number mentioned dominates.
  • Optimism bias and planning fallacy: people underestimate time for their own work.
  • Student syndrome: starting only at the last moment.
  • Padding: each level adds a hidden buffer.

Counter them with ranges, reference-class data, independent review, and explicit schedule reserves.

4. Build the schedule: network diagram and critical path

The critical path method (CPM) uses dependencies, durations, and calendars to identify the path driving completion and calculate float. In the simple unconstrained network below, the longest path determines the earliest finish.

Definitions

  • Early start (ES) and early finish (EF): earliest dates, found in the forward pass.
  • Late start (LS) and late finish (LF): latest dates without delaying the project, found in the backward pass.
  • Total float (slack): how much an activity can slip before it affects the applicable finish constraint. In the simple model below, it is LS − ES or LF − EF.
  • Free float: how much an activity can be delayed without delaying the early start of any successor.
  • Critical path: the longest path; activities on it have zero total float (with an unconstrained end date).
  • Near-critical path: a path with very little float, which could become critical.

Forward and backward pass

For finish-to-start links with zero lag and a common calendar, use elapsed working days, with the first activity starting at day 0.

  1. Forward pass: ES of an activity = the largest EF among its predecessors (all FS). EF = ES + duration.
  2. Backward pass: start from the project end, where LF = EF of the final activity. LS = LF - duration. LF of an activity = the smallest LS among its successors.
  3. Float: LS - ES.

Worked example: critical path and float

Scenario. A project has these FS activities.

Activity Duration (days) Predecessors
A 4 -
B 6 A
C 3 A
D 5 B
E 7 C
F 2 D, E

Step 1: list the paths.

  • A-B-D-F: 4 + 6 + 5 + 2 = 17
  • A-C-E-F: 4 + 3 + 7 + 2 = 16

Step 2: forward pass. A: 0-4. B: 4-10. C: 4-7. D: 10-15. E: 7-14. F: ES = max(15, 14) = 15, EF = 17.

Step 3: backward pass. F: LF = 17, LS = 15. D: LF = 15, LS = 10. E: LF = 15, LS = 8. B: LF = 10, LS = 4. C: LF = 8, LS = 5. A: LF = min(LS of B = 4, LS of C = 5) = 4, LS = 0.

Step 4: float. A: 0. B: 4 - 4 = 0. D: 10 - 10 = 0. F: 0. C: 5 - 4 = 1. E: 8 - 7 = 1.

Step 5: answer. The critical path is A-B-D-F, duration 17 days. C and E each have one day of total float. C has zero free float because E is due to start as soon as C finishes. E has one day of free float before F.

Why the other readings are wrong. The longest path is not the one with the most activities (A-C-E-F has the same count but is shorter). Activities C and E have float, but they share it: delaying C by 1 day consumes E's float too, since they are on the same path. The sum of durations of all activities (27) is not the project duration.

Gantt charts, milestones, and schedule reserve

A bar chart (Gantt) shows activities against time; a milestone chart shows only key events; a network diagram shows logic. Document time contingency for identified risks and show how it relates to the approved schedule. Follow the project’s governance for any time allowance managed separately. Buffers in critical chain scheduling protect the chain's end rather than each task: resource constraints are addressed, safety margins are pooled, and a project buffer is placed at the end of the critical chain.

Resource-constrained schedules

A dependency-based CPM calculation alone does not resolve competition for a shared resource. Resource leveling adjusts start and finish dates to resolve overallocation and can lengthen the schedule and change the critical path. Resource smoothing adjusts activities within their float without changing the end date. See planning resource capacity and calendars for how to test levelling and smoothing with actual availability. The schedule must be checked against real resource calendars (holidays, leave, shared specialists).

5. Coordinate with other projects and operations

Few projects are isolated. The project manager identifies dependencies with other projects, programs, and daily operations, and agrees dates with their owners.

  • Shared resources: a specialist split across projects may be only 40% available. Plan with actual availability.
  • Operational blackout periods: a retailer will not accept a system cutover in November and December. Add these constraints as calendars.
  • Interproject dependencies: record them in the schedule and the risk register; agree interface milestones and escalation points.
  • Change windows: production changes may be restricted to Sunday nights; plan deployments accordingly.
  • Governance: align with the portfolio or program calendars, and raise conflicts to the steering committee or PMO.

Worked example: a blackout constraint

Scenario. A bank's ledger migration needs 9 working days in the production environment. The bank forbids production changes during the last 5 business days of every quarter. The quarter ends in 15 business days, and the migration cannot start until testing finishes in 10 business days.

Step 1: compute the window. Testing finishes at the end of day 10. Production changes are barred on days 11 through 15. With working days counted consecutively across the quarter boundary, migration starts on day 16 and takes days 16 through 24 inclusive.

Step 2: options. Schedule the migration after the blackout (a delay of 5 working days), compress testing to start migration earlier, or split the migration if the team can preserve data consistency and approvals. Show the revised dates when baselining and controlling the schedule.

Step 3: decide with stakeholders. Splitting risks data inconsistency; compressing testing raises quality risk. The project manager recommends scheduling after the blackout and shows the sponsor the delay, the risk comparison, and the effect on downstream milestones.

Why other choices are weaker. Starting anyway breaches an operational rule. Ignoring the blackout in the baseline would create a schedule that cannot be executed.

6. Plan releases and iterations (adaptive)

Product roadmap and release plan

The roadmap shows themes or outcomes across quarters. The release plan indicates which features are planned for which release, based on priority, value, dependencies, and team velocity. A release forecast should reflect recent evidence. Identify any contractual date or approved commitment separately.

Iteration planning

In Scrum, the Scrum Team defines the Sprint Goal and Developers select work in discussion with the Product Owner. Their forecast draws on capacity, past performance, and the Definition of Done. Capacity accounts for holidays and non-project work: a team of 6 with 10-day iterations, with 20% of time on support, has about 6 x 10 x 0.8 = 48 person-days.

Example: release forecast

A product owner has 180 story points of must-have items. The last six iterations delivered 28, 32, 30, 26, 34, 30 points (average 30). Forecast: 180 / 30 = 6 iterations, At the observed low and high rates, 180 points take six to seven whole iterations. A seven-iteration window leaves one iteration at the average rate and no whole-iteration margin at the slowest rate. With five iterations, compare a smaller release with a date or capacity change. Review commitments and which work can be deferred with the Product Owner.

Dependencies and cross-team planning

Large-scale agile uses joint planning events to surface dependencies. Teams visualize them on a board and agree interface milestones. Dependencies are limited by structuring teams around products or value streams where possible.

7. Common pitfalls and exam tips

  • Pitfall: confusing float with reserve. Float belongs to activities; reserve is added for risk.
  • Pitfall: a single-point estimate with no range. The best answer usually shows uncertainty.
  • Pitfall: ignoring resource availability. A path-based schedule can be impossible if the people are unavailable.
  • Pitfall: critical path with leads and lags. Include them in the pass calculations.
  • Pitfall: treating the release plan as a promise. It is updated as velocity data change.

Exam tip: For simple zero-lag finish-to-start networks, compare path durations. In other schedules, calculate using the actual relationships, lags, calendars, and constraints. Check that you have used the right formula: PERT divides by 6, triangular by 3.

Exam tip: If a question gives you a conflict with another project, the best answer is early coordination with the other project manager or the PMO, with options presented to the sponsor.

Exam tip: For agile estimates, prefer relative sizing by the whole team (planning poker) over a manager assigning hours.

Exam tip: When estimating accuracy is the issue, bottom-up is more accurate but costly; analogous is quick but rough. Pick based on the stage of the project and the information available.

8. Mini case study: the regulatory launch date

Context. A project to launch a new insurance product has a fixed regulatory deadline of 31 March. The network has a critical path of 84 working days from the start date, which leaves 6 working days before the deadline. The pricing team is available at 50% for the first 20 working days and at full capacity after that. Its critical-path work is scheduled to start now. A previous similar project took 98 working days. The sponsor asks whether the date is safe.

Question. What should the project manager answer and do?

Solution (commented).

  1. Compare with history. The benchmark of 98 days suggests the 84-day estimate may be optimistic. Review the estimate basis, scope differences, and assumptions before replying.
  2. Re-check the critical path with real availability. If pricing needs 20 full-capacity days and work scales proportionally, the first 20 days deliver ten of those days of effort. The remaining ten take ten working days at full capacity. With no other changes, the path grows to 94 days and exceeds the 90-day window by four working days.
  3. Use three-point estimates for the riskiest activities (regulatory review, pricing validation) to represent uncertainty. Use a schedule risk analysis if the decision requires a probability of meeting the date.
  4. Prepare options. Increase pricing capacity (borrow a specialist), fast track discretionary dependencies (start the documentation while pricing completes), descope non-mandatory features, or negotiate an earlier regulatory review slot. Each has cost and risk.
  5. Coordinate with the other project that shares the pricing team, through the PMO.
  6. Reply honestly. The date is at risk at current assumptions; here are the options, their effects, and a recommendation. Update the schedule baseline only after the change request is approved.
  7. Do not promise the date based on the 84-day figure alone, or pad every task secretly.

Key terms

  • Critical path: longest path, zero total float.
  • Total float / free float: delay possible without delaying the end / the next activity.
  • Lead / lag: overlap / delay between linked activities.
  • Rolling wave planning: detailed near-term planning, high-level far-term.
  • PERT estimate: (O + 4M + P) / 6.
  • Analogous / parametric / bottom-up estimating: estimating from past projects / from rates / from details.
  • Velocity: story points completed per iteration.
  • Roadmap / release plan: high-level direction / planned feature delivery by release.
  • Schedule reserve: time added for identified risks.
  • Mandatory / discretionary dependency: hard logic / preferred logic.

Summary checklist

  • I can match the schedule approach to the development approach.
  • I can list estimating techniques and their trade-offs.
  • I can compute a conventional PERT estimate and explain its spread approximation.
  • I can do forward and backward passes and find the critical path and float.
  • I know the difference between total float, free float, and reserve.
  • I coordinate schedule dependencies with other projects and operations.
  • I can forecast a release from velocity and backlog size.
  • I understand common estimating biases and how to counter them.

Official sources