Project Teams and Contracts by the Numbers
Chapter 8 of 10 · 11 min read
Staffing ramp-up, late additions and the cost of turnover
The DC core team's 194.6 FTE-months peaking at 16.2 in month 15, a leaver costed at 3 FTE-months (A$66k), turnover at 18% against 10%, and Brooks's law in a commissioning crunch.
Chapter 7 sized the core team as if its twenty people were all present from the first day to the last. They are not. The designers are busiest before the works are awarded, the builder’s team builds up as the slab goes in, and the racking, conveyor and software people arrive late and leave last. People also leave part way through, for other jobs, promotions or personal reasons, and every departure takes knowledge with it that the project has to rebuild.
This chapter plans the core team’s loading month by month for the distribution centre, separates the people the owner pays for from those inside the contractors’ prices, and puts a number on the cost of one leaver and on a year’s expected turnover. It then tests the old rule that adding people to a late job makes it later, using the conveyor commissioning as the example, and ends with a project manager who left in month 9 without a handover.
The core team’s loading plan
This is an illustration. All costs are in A$ excluding GST, with people’s time costed at a blended A$22,000 per FTE-month. The loading plan below covers the core management team of Chapter 1, not the trades and site labour inside the works. In month 1 only the owner’s people and the designer’s three are working, 4.7 FTE. Two of the builder’s people work from month 3 on tendering and pre-construction planning, and its team is at full strength from month 5, when the works are awarded. The racking supplier-installer, the conveyor and controls integrator and the WMS vendor join between months 6 and 8 at low intensity and build up for installation and commissioning. The team peaks at 16.2 FTE in month 15, when the builder is still completing the building, the racking installation is finishing and the conveyor installation is under way.
Over the 18 months the plan totals 194.6 FTE-months. By organisation: the builder 82, the owner 42.1, the designer 23, the conveyor and controls integrator 23, the WMS vendor 15 and the racking supplier-installer 9.5. At A$22k per FTE-month the whole core team represents about A$4,281k of people’s time across the six organisations.
Two features of the profile matter for planning. First, the peak comes late: in month 15 the project has its most people at the moment when the most organisations are on site at once, which is also when the interfaces of Chapter 1 are most active and the paths of Chapter 7 most crowded. Second, the late arrivals are the people with the least time to learn the project. The WMS vendor joins in month 8, at one FTE until month 14 and two from month 15, and must absorb decisions the rest of the team made over the previous seven months.
A loading plan like this starts from the work, not from the organisation chart. For each role, the hours of work in each month are converted to FTE at a realistic productive time per person, allowing for leave, training and the coordination load that Chapter 7 measured. The Staffing Need tool does that conversion from workload and productive time to a practical staffing requirement, and is a useful check on any month where the plan looks thin.
Bringing late arrivals up to speed
A late arrival’s ramp-up can be shortened by giving them the project’s memory in a form they can read. On the DC that means three documents. The interface register of Chapter 1 shows the 10 of its 38 interfaces that involve the WMS vendor (5 with the conveyor integrator, 3 with the owner and 2 with the builder for the IT network), with the named owner of each. The decision log shows what has already been decided on those interfaces and why. The programme shows when each interface must be closed. A half-day induction with the coordination group, walking through those three documents, costs a few hours of each lead’s time; it is the cheapest way to cut the 1.5 FTE-months of ramp-up loss that the next section assumes for a newcomer, and it applies just as much to a replacement as to a planned late arrival.
Who pays for whom
Most of the 194.6 FTE-months are not charged to the owner’s budget at all. The designer’s, builder’s, racking supplier’s, integrator’s and WMS vendor’s people are paid inside their fees and contract prices. Their cost reaches the owner only through those prices, and under the cost-based models of Chapters 4 to 6 through the reimbursable cost and the fee.
Of the owner’s own 42.1 FTE-months, only 25 are charged to the project: the project director full time for 18 months (18 FTE-months) and the contract administrator half time from month 5 (7 FTE-months). At A$22k that is A$550k, the owner’s project team line in the A$18.0m budget. The sponsor, at 0.2 FTE throughout, and the DC manager-designate, at 0.5 FTE to month 9 and full time from month 10, are existing staff. Their time is recorded in the plan, because they must be released from their day jobs to give it, but it is not charged. Recording uncharged time matters: if the DC manager-designate is pulled back to run the old warehouse in a busy month, the project loses a person the budget never showed.
The cost of one leaver
When a core team member leaves, the project pays three times over. There is the vacancy before the replacement starts, during which the work is either not done or done by someone else at the expense of their own. There is the ramp-up of the replacement, who is assumed here to be 25% productive in the first month, 50% in the second and 75% in the third, a loss of 0.75 + 0.5 + 0.25 = 1.5 FTE-months. And there is mentoring, the time colleagues spend bringing the newcomer up to speed, assumed at 0.25 FTE for two months.
- Vacancy
- 1 FTE-month between the leaver going and the replacement starting
- Ramp-up loss
- 25%, 50% and 75% productive in months 1 to 3: 1.5 FTE-months
- Mentoring
- 0.25 FTE of a colleague for 2 months: 0.5 FTE-month
- 1 + 1.5 + 0.5 = 3 FTE-months
- 3 × A$22k = A$66k per leaver
The A$66k per leaver does not include the cost of anything that goes wrong because knowledge left with the person. That cost is usually larger, and this chapter’s error shows one.
Expected turnover over 18 months
This is an illustration. At an annual turnover of 18%, a team of 20 can expect 20 × 0.18 × 1.5 = 5.4 leavers over the 18 months, costing 5.4 × A$66k = A$356k. At 10%, the expectation is 3.0 leavers and A$198k. The difference, A$158k, is the value to the project of keeping turnover down to 10%, and it is a direct cost before any knock-on delay. Neither rate is a market figure; a real project should use the turnover its own organisations have recorded on comparable work.
Turnover cannot be eliminated, but its cost per leaver can be cut. If the leaver and the replacement overlap for two weeks of paid handover, there is no vacancy; the replacement learns faster, so the ramp-up loss falls to 60% of its former size, 0.9 FTE-month; mentoring stays at 0.5; and the overlap itself costs 0.5 FTE-month. The total is 1.9 FTE-months, about A$42k (A$41.8k), against A$66k without the overlap.
Continuity has a value beyond the replacement arithmetic. A person who stays learns the project’s history, its people and its weak points, and the second half of a project depends on that learning. Experience, proven technology and keeping the team in Modular Projects by the Numbers prices the loss when a crew that has learned a repeated build is replaced, and Learning curves: unit and cumulative-average models shows how that learning is measured.
Key-personnel terms
The contracts are where continuity can be required rather than hoped for. A key-personnel clause names the people, usually each organisation’s lead and a few specialists, whose presence was part of the reason the organisation was selected. It requires the contractor to keep them on the project, and to obtain the owner’s approval of any replacement, who must be of comparable experience. Good clauses also require a handover overlap and allow the owner to recover the cost of an unapproved change.
- Do: Name each organisation's lead and critical specialists in the contractthe seven coordination group members plus commissioning and controls leads
- Do: Owner approval of any replacement, of comparable experienceapproval within 5 working days of nomination
- Do: Two-week paid handover overlap for every key person who leaves1.9 FTE-months instead of 3 per leaver
- Do: Named deputy for each lead, who attends the coordination group when the lead is absentkeeps the lead's paths open during a vacancy
- Do: Handover record: open decisions, interface positions, commitments madereviewed by the incoming person and the coordination group
- Avoid: Accepting a replacement on the day they arrive, without the outgoing person presentthe knowledge leaves before it is passed on
- Ask: Who holds the builder's open decisions if its project manager leaves tomorrowno deputy was named when the works were awarded
Adding people late: Brooks’s law
Brooks (1975) put the risk of late additions in one sentence: “Adding manpower to a late software project makes it later.” Brooks presented it as an outrageous simplification, and it is not a law of nature. It describes a mechanism: newcomers are not productive at once, the people who must train them stop doing their own work, and each newcomer adds communication paths. Whether adding people helps depends on how much work is left, how divisible it is, and how long the newcomers take to become useful.
This is an illustration. In month 17, the conveyor and controls integrator’s three experienced commissioning engineers have 960 engineer-hours of commissioning left on 4 conveyor zones, and every week of delay to the DC costs the owner A$80k. The model behind the numbers below is a teaching model, not research. Each engineer works 38 hours a week and is 85% on task; each spends 1 hour a week per communication path; newcomers are 25% productive in weeks 1 and 2 and 60% in weeks 3 and 4, then fully productive; each newcomer costs 6 hours a week of experienced engineers’ time in weeks 1 to 4; and because there are only 4 zones, engineers beyond 4 add only 30% of their output. An engineer costs A$5,320 a week (38 hours at A$140).
| Team | Weeks to finish | Weeks saved | Value of delay saved | Cost of added engineers |
|---|---|---|---|---|
| 3 experienced | 10.46 | | | | ||
| 3 + 1 new | 8.84 | 1.61 | A$129k | A$47k |
| 3 + 2 new | 8.98 | 1.48 | A$118k | A$96k |
| 3 + 3 new | 9.02 | 1.44 | A$115k | A$144k |
| 3 + 5 new | 9.01 | 1.44 | A$116k | A$240k |
| 4 experienced from the start | 8.07 | 2.39 | A$191k | A$43k |
With 3 engineers the work takes 10.46 weeks. Adding one engineer brings it to 8.84 weeks, saving 1.61 weeks worth A$129k of delay for A$47k of extra engineer time: a clear gain, because the fourth person has a zone of their own. Adding three brings it only to 9.02 weeks, later than adding one, saving 1.44 weeks (A$115k) for A$144k: a net loss. Adding five gives 9.01 weeks for about A$240k. Beyond one per zone, the newcomers’ training load and extra paths cancel their output. That is Brooks’s law in miniature: not that every addition makes a job later, but that additions past the point the work can absorb make it later than a smaller addition would.
The best line in the table is the last. Four experienced engineers planned from the start finish in 8.07 weeks, saving 2.39 weeks (A$191k) for about A$43k of extra engineer time, because nobody needs training at the critical moment. The lesson is to decide the commissioning team’s size when the programme is set, not when commissioning is late. Commissioning crunches are also what earlier testing prevents: Finding problems while they are cheap: the cost of change by stage in Prototypes and Simulation by the Numbers shows how the cost of a problem grows with the stage at which it is found.
The error: a project manager who left without a handover
This is an illustration. In month 9 the builder’s project manager resigned to take a job with another builder. The builder released the project manager at the end of the notice period and the replacement started 4 weeks later, with no overlap and no named deputy. The site manager covered day-to-day work, but the project manager’s open decisions sat unassigned.
One of them was the conveyor integrator’s access and staging plan, which set when and where the conveyor installers could work on the building slab while the builder finished the roof and services. Its approval normally took 2 weeks; it took 6. The conveyor installation start moved, and 2 weeks of the delay reached the critical path: 2 × A$80k = A$160k. Adding the replacement cost of A$66k, the departure cost the project A$226k.
It was found through the project’s decision log, which the coordination group reviews weekly. Normally about 2 items in the log are more than 10 working days old; in the third week after the resignation there were 11, eight of them assigned to the builder’s project manager. The contract administrator traced them, and the coordination group reassigned the open items to the builder’s site manager and contracts administrator until the replacement arrived.
The correction was contractual and practical. The builder agreed to a key-personnel clause covering its project manager, site manager and contracts administrator, with owner approval of any replacement and a required two-week overlap. Each organisation named a deputy for its lead, who attends the coordination group when the lead is absent. With the overlap and a deputy in place, the same resignation would have cost about A$42k in replacement and, with the open decisions held by the deputy, little or no delay.
Exercise: turnover at 14%
The owner’s procurement adviser suggests that 14% a year is a more realistic turnover rate for the core team than either 10% or 18%. Using the case’s assumptions, what is the expected number of leavers over 18 months, and their expected replacement cost without and with the two-week overlap policy?
Worked answer. Expected leavers are 20 × 0.14 × 1.5 = 4.2. Without overlap, each costs 3 FTE-months × A$22k = A$66k, so the expected cost is 4.2 × A$66k = A$277.2k. With the overlap policy, each costs 1.9 FTE-months × A$22k = A$41.8k, so the expected cost is 4.2 × A$41.8k = A$175.6k, a saving of about A$101.6k. The overlap is paid by whichever organisation employs the leaver, and an owner who wants it should write it into each contract and expect to see it in the prices. The larger saving, as the error showed, is the delay avoided when open decisions are handed over rather than left waiting.