Playbook

How to Plan a Project That Finishes on Time

Six questions, asked in order, for a project with a date it can't miss.

A project planning playbook is a six-stage sequence of planning tools that defines all the work, finds which work sets the finish date, and keeps the plan honest until the finish.

Project planning playbook: six stages in order, from the Work Breakdown Structure and the Critical Path Method through PERT and the Resource Allocation Matrix to risk management and the Gantt chart.

The route

Six questions, in order

The route runs in the order a plan has to be built: the work, its order, the odds, the people, the risks, then the schedule everyone follows.

On timing: stages 1 to 5 are planning, about two weeks for a project of a few months. Stage 6 starts when the work does and runs to the finish.

Is the scope still changing week to week? Plan in short cycles with Agile Sprint Planning instead.

Free tools run each stage on your own project. The Project Planning Tool lays out all six stages and starts you off; or start at stage 1 with the Work Breakdown Structure Tool, or take the 15-minute guided tour first.

  1. What is all the work?Stage 1 · Work Breakdown Structure

    Passes on: every work package, each with one owner

    Scope changing every week?Plan in short sprints instead.

  2. Which work sets the finish date?Stage 2 · Critical Path Method

    Passes on: the critical path, and the float on everything else

    Critical path longer than the time you have?Shorten it on the path, nowhere else.

  3. How sure is the date?Stage 3 · PERT

    Passes on: an expected finish, and its odds

    Odds too low?Buy time on the critical path now.

  4. Is anyone booked twice?Stage 4 · Resource Allocation Matrix

    Passes on: a load each person can carry

    Anyone over 100% in a busy week?Move work that has float.

  5. What could knock it off course?Stage 5 · Risk Management Framework

    Passes on: risks with owners, and one shared buffer

  6. Is it still on track?Stage 6 · Gantt Chart

    Passes on: a schedule updated every week

    Finish date never moves?Check for dates fixed by hand.

  7. Did it finish on time?Then · Compare actual with estimated

The example

One company, followed all the way through

This playbook follows one company from start to finish. It is an illustrative composite, and the details are simplified.

A furniture and homeware retailer in Lima, Peru, has one showroom and is opening a second in a new shopping center. The center opens in 22 weeks, and every tenant is expected to open with it. A store that opens late pays the center PEN 3,000 (Peruvian soles) a day, and misses the opening crowds in the weeks before Christmas.

Most of its stock comes from local furniture makers, its suppliers; its best-selling sofas are imported. The operations manager has a date and a task list from the kickoff meeting. Planning takes two weeks, leaving 20 for the work.

Each stage below ends with what the company produced at that step, so you can watch one stage's output become the next stage's input.

Stage 1 of 6

What, exactly, is all the work?

Tool: Work Breakdown Structure · Time: 3 to 5 days

Start with what the project has to produce, not with a task list. A work breakdown structure (WBS) divides the whole scope, level by level, into work packages: pieces small enough to estimate, give to one owner and track. Three rules make it work.

Three rules of a work breakdown structure

1The 100% rule
The pieces under any item add up to all of its scope, no more and no less
2Nouns, not verbs
Each item names something the project produces, such as Approved drawings
3Stop at the work package
Small enough to estimate, give to one owner and track as a unit

Build it top down with the people who will do the work, then test every branch against the 100% rule: does what sits under it cover all of it? Missing work never shows up as an error. It shows up as a project that finishes late for reasons the schedule never mentioned.

The showroom's WBS: five branches, 31 work packages (illustrative). The highlighted row held half of what the task list missed.
Branch (level 1)Work packages, for exampleMissing from the first task list
Fitted showroomApproved drawings, finished building works, display fixturesThe center's design approval; the fire safety inspection; the city's operating license
Stock on displayChosen range, supplier orders, imported sofas, stock in placeCustoms clearance for the sofas
Trained teamEight staff hired, training completed, uniformsNothing
Working systemsTill, card payments, stock system, networkThe card payment contract, which takes three weeks
Opening campaignShop sign, launch offers, opening eventThe center's approval of the sign

What goes inThe scope: what the project must deliver, and by when.

What comes outEvery work package, each with one owner.

Skip it if: the same project has been run several times and its WBS already exists.

Handed to stage 2: 31 work packages in five branches, each with an owner, including six the first task list had missed.

Decision point

If nobody can say what “done” means for a branch, stop and agree the scope first. A WBS organizes scope; it can't decide it. Here, the owner and the operations manager disagreed on whether the opening event was part of the project. It was, so it got a branch.

Stage 2 of 6

Which chain of work sets the finish date?

Tool: Critical Path Method · Time: 1 to 2 days

Next, put the work in order. The Critical Path Method lists each activity with what it must wait for, then finds the longest chain of dependent work from start to finish. The project can't finish before that chain does, so its length is the shortest the project can be.

Four terms the Critical Path Method works out

1Earliest start
From the forward pass: the soonest an activity can start
2Latest start
From the backward pass: the latest it can start without moving the finish
3Float
The gap between the two, also called slack
4Critical path
The chain of activities with zero float

Use one duration per activity for now, the most likely one. Then question the arrows, not just the durations: many dependencies are habits. The first network came to 21 weeks, one more than the company had, because display fixtures were to be made after the fire inspection. Made during the building works instead, and only fitted after it, they cut two weeks at no cost.

The main activities, in weeks (illustrative). Highlighted rows have zero float: they are the critical path.
ActivityWeeksWaits forCan start in weekFloat, weeks
A. Store design drawn3–10
B. The center's design approval3A40
C. Building works8B70
D. Display fixtures made6B73
E. Fire inspection and license1C150
F. Display fixtures fitted2D, E160
G. Range chosen, orders placed2–13
H. Sofas made and shipped10G33
I. Customs clearance2H133
J. Stock delivered and displayed2F, I180
K. Staff trained in the showroom1F181

What goes inThe work packages from stage 1, each with a most-likely duration and what it must wait for.

What comes outThe critical path, the project's shortest length, and the float on everything else.

Skip it if: the project is a few weeks of work in one obvious order.

Handed to stage 3: a 19-week critical path through the fit-out, one week inside the 20 available, and three weeks of float on the imported sofas.

Decision point

If the critical path is longer than the time you have, shorten it on the path and nowhere else. Speeding up an activity with float buys nothing. Recalculate after every change, because the path can move.

Stage 3 of 6

How likely are you to finish by the date?

Tool: PERT · Time: 1 to 2 days

A single duration is a guess that looks like a fact. PERT asks for three: optimistic (O), most likely (M) and pessimistic (P). From them it works out an expected duration, and a standard deviation that measures how uncertain the activity is.

The imported sofas, worked

Expected duration (PERT)(O + 4M + P) ÷ 6

  1. Three estimatesOptimistic 8 weeks, most likely 10, pessimistic 18, from the supplier and the shipping agent
  2. Expected(8 + 4 × 10 + 18) ÷ 6 = 11 weeks
  3. Standard deviation(18 − 8) ÷ 6 = 1.7 weeks, the widest range of any activity

Run it on the critical path and on any path with little float. Add the expected durations along the path. For the path's uncertainty, add the squares of the standard deviations and take the square root. Together they give the odds of finishing by the date.

The critical path's most-likely total was 19 weeks. Its expected total is 20.5, give or take 1.3: about a one-in-three chance of opening on time.

The critical path, re-estimated in weeks (illustrative). The highlighted row is the one the company paid to shorten.
ActivityOptimisticMost likelyPessimisticExpected
A. Store design drawn2343.0
B. The center's design approval2373.5
C. Building works78128.5
E. Fire inspection and license1131.3
F. Display fixtures fitted2232.2
J. Stock delivered and displayed1232.0

What goes inThe critical and near-critical activities from stage 2, with three estimates each from the people who will do the work.

What comes outAn expected finish, the odds of making the date, and the activities with the widest ranges.

Skip it if: the work has been done many times and its durations barely vary.

Handed to stage 4: an expected finish of 18.3 weeks after a second building crew was hired, about four chances in five of opening on time, and the sofas as the least certain activity.

Decision point

If the odds are lower than you can accept, buy time on the critical path now, while it is cheap. A second building crew cost PEN 38,000, less than two weeks of penalties, and cut building works to 5, 6 and 9 weeks. Counting the sofas' path, now almost as long, the odds rose to about four in five.

Stage 4 of 6

Is anyone booked on two things at once?

Tool: Resource Allocation Matrix · Time: half a day, then weekly

The network assumes everyone is free when it needs them. A resource allocation matrix checks: people down one side, the work claiming them across the top, and in each cell the share of that person's time it takes. Read across a row to see how loaded one person is.

Weeks 16 to 20, the busiest stretch, as a share of each person's time (illustrative). The outlined cell is the worst clash; any total over 100% is a promise nobody can keep.
PersonNew showroomFlagship's Christmas refitDay-to-day workTotal
Visual merchandiser70%60%20%150%
Operations manager60%10%40%110%
Buyer20%30%40%90%
HR officer30%0%50%80%
New store manager90%0%0%90%

Check the busiest weeks, not the average, and include work outside the project. Plan people at about 80%, not 100%; the rest absorbs illness, meetings and everything else. And name who decides between competing claims before the grid shows a clash.

What goes inThe schedule from stages 2 and 3, and everything else the same people are doing.

What comes outEach person's load in the busy weeks, with every clash settled.

Skip it if: every activity has its own people, doing nothing else.

Handed to stage 5: the visual merchandiser down from 150% to 90% and the operations manager to 80% in the busy weeks, with no date on the critical path changed.

Decision point

If someone is over 100% in a busy week, move work that has float, not work on the critical path. Setting up the displays had none. The flagship's refit could go four weeks earlier, and the owner, named in advance as the one who decides, moved it. The operations manager handed day-to-day work to a deputy.

Stage 5 of 6

What could knock the plan off course?

Tool: Risk Management Framework · Time: 1 to 2 days, then every two weeks

The schedule shows how long things take, not what might happen to them. Risk management lists the events that could knock the plan off course, scores each one, and gives it an owner and one of four treatments.

The four treatments for a risk

1Avoid
Don't do the thing that carries the risk
2Reduce
Lower the probability, the impact, or both
3Transfer
Move the cost elsewhere, by contract or insurance. The delay still happens
4Accept
Carry it on purpose, and watch it

Write each risk as a cause, an event and a consequence, so someone outside the room would recognize it. Keep spare time in one place. The plan's dates use most-likely durations and finish in week 17, so the three weeks before opening are one shared buffer, held by the operations manager; PERT expects about 1.3 of them to be used. Padding each activity hides it where it gets used up.

The showroom's risk register, probability (P) and impact (I) each scored 1 to 5 (illustrative). The highlighted row scored highest.
Risk: cause, event, consequenceScore: P × ITreatment and owner
A new customs broker; paperwork errors hold the sofas at the port; the showroom opens half empty12Reduce: the broker checks the papers before the ship sails. Owner: the buyer
Drawings use the old fire code; the inspection fails; weeks of rework on the critical path10Reduce: a fire engineer inspects first, in week 11. Owner: the operations manager
The builder moves the crew to another site; building works slip8Transfer: a late-finish penalty in the contract. Owner: the operations manager
New staff leave before opening; too few people for the opening weeks6Reduce: hire one more than needed. Owner: the HR officer
The printer is late; a week without posters2Accept. Owner: the store manager

What goes inThe schedule, the ranges from stage 3, and the WBS.

What comes outA risk register: each risk scored, owned and treated.

Skip it if: never. Even a short list with owners beats none.

Handed to stage 6: five risks with owners and treatments, a review every two weeks, and a three-week buffer held in one place.

Decision point

If a risk has no named owner and no review date, nobody is managing it. Review the register every two weeks and watch for early signs. For the sofas, the sign was the supplier's production photos, due in week 6.

Stage 6 of 6

How do you keep the plan on track, in plain view?

Tool: Gantt Chart · Time: a day to draw, then weekly until the finish

Nobody manages from a network diagram. A Gantt chart shows the schedule as bars against a calendar, which everyone from the owner to the builder can read. Arrows show what waits for what, diamonds mark milestones, shading inside a bar shows work done, and a today line shows where the project stands. The chart displays the critical path; it doesn't find one. That was stage 2.

The showroom's Gantt chart at the end of week 10 (illustrative). The sofas are two weeks late, so the stock activities are now the critical path, and two weeks of buffer remain.
Gantt chart of the showroom project at week 10. The sofas are two weeks late, so the stock activities G, H, I and J are now the critical path; two weeks of buffer remain before opening day in week 20.135791113151719WeekA. Store design drawnB. Design approvalC. Building worksD. Fixtures madeE. Fire inspectionF. Fixtures fittedG. Range and ordersH. Sofas made, shippedI. Customs clearanceJ. Stock displayedK. Staff trainedOpening dayToday: end of week 10
  • Bar: critical path at week 10
  • Bar: other work
  • Shading: work done
  • Slip reported in week 10
  • Shared buffer
  • Milestone
  • Arrow: the next bar waits for this one
  • Today

Record what actually happened every week, and let the dates move. Count finished work packages rather than percentages; work reported 90% done tends to stay there. In week 10 the supplier reported the sofas two weeks late. The stock activities became the critical path, and the buffer fell from three weeks to two.

What goes inThe schedule, the critical path and the buffer from stages 2 to 5.

What comes outA chart everyone can read, updated each week with actual progress.

Skip it if: never. It is how everyone else sees the plan.

Handed on: a schedule updated every Friday, a slip seen the week it happened, and a showroom open on the center's opening day.

Decision point

If the finish date hasn't moved in months while the work plainly has, the chart has stopped being a schedule. Look for dates someone fixed by hand. Here the slip showed the week it happened, and the supplier flew in 40 sofas for the displays, so the opening held.

Pace

Fast track or thorough

The fast track suits a project of a few weeks, run by one team. The thorough run suits a project of several months, or any date with a penalty attached. The order of the stages stays the same either way.

What changes between the two paces.
StageFast track (a day or two)Thorough (about two weeks)
1. Work Breakdown StructureA one-page list of deliverables, checked against the 100% ruleEvery work package, each with an owner
2. Critical Path MethodDependencies sketched on a whiteboardThe full network, with float for every activity
3. PERTThree estimates for the two or three longest activitiesEvery critical and near-critical activity
4. Resource Allocation MatrixAsk each person what else is on their plateA grid of the busy weeks, including work outside the project
5. Risk ManagementThe top five risks, each with an ownerA scored register, reviewed every two weeks
6. Gantt ChartBars on a shared sheet, updated weeklyScheduling software with arrows, milestones and a baseline

Failure modes

How project plans go wrong

Common failures and what prevents them.
What happensWhat it looks likeThe fix
Planning from a task listWork nobody listed turns up halfway throughBuild the WBS and test it with the 100% rule
One estimate treated as a promiseA date with no odds, missed by a little everywhereThree estimates for each activity on the critical path
Padding every activitySpare time used up wherever it sitsOne shared buffer, held by one person
Assuming people are freeTwo activities need the same person in the same weekCheck each person's load in the busy weeks
A chart that never movesThe finish date holds while the work slipsRecord actual progress weekly and let dates move

After the playbook

Measuring the finish

Compare the actual finish with the date, and each activity's actual duration with its three estimates. In the example, the showroom opened on the center's opening day with four days of buffer left.

The gaps show whose estimates run short and which risks came true. Keep the actual durations: they are the best starting estimates for the next project, here the company's third showroom.

Common questions

Project planning: quick answers

How do you plan a project so it finishes on time?

Break the scope into work packages, put them in order to find the critical path, and give the uncertain activities three estimates to get the odds of making the date. Check that nobody is booked twice, list the risks with owners, and keep one shared buffer. Then publish the schedule and update it weekly with actual progress.

What is the critical path in project management?

The longest chain of dependent activities from start to finish. The project can't finish sooner, so a delay on it delays the finish. Activities off the critical path have float: time they can slip without moving the finish date.

How do you calculate a PERT estimate?

Add the optimistic estimate, four times the most likely and the pessimistic, then divide by six. With 8, 10 and 18 weeks, the expected duration is 11 weeks. The standard deviation is the pessimistic minus the optimistic, divided by six: 1.7 weeks.

What is the difference between a WBS and a task list?

A WBS names what the project produces, broken down until each piece can be estimated and owned, and its pieces must add up to the whole scope. A task list names actions and has no test that it is complete, which is how work gets missed.

How much buffer should a project have?

Enough to cover the uncertainty on the critical path. PERT gives one way to size it: the gap between the most-likely finish and a finish you are confident in, three weeks on a 17-week plan in the example. Hold it as one shared buffer, not as padding on each activity.

What is the difference between a Gantt chart and the critical path method?

The Critical Path Method works out which activities control the finish date. A Gantt chart shows the schedule as bars on a calendar so people can read and track it. Most scheduling software does the first and shows you the second.

What if the deadline can't be met?

If the critical path is still longer than the time available after questioning its dependencies, there are three choices: more people or money on critical activities, less scope, or a later date. Raise it early, because each choice costs more the longer it waits.