Project Teams and Contracts by the Numbers
Chapter 9 of 10 · 11 min read
Shared goals, co-location and decision rights
One scorecard for six organisations, decision limits for 262 decisions, decision latency priced at A$320k against A$96k, and a joint site office that netted A$264k once limits were delegated.
A contract can share the money, and a team structure can shorten the paths, but neither makes decisions. On the distribution centre, several hundred decisions have to be made by somebody between award and handover: approvals of shop drawings, answers to requests for information, variations, access plans, test results. Each one waits until someone with authority makes it, and the waiting is where much of a project’s time goes. The speed of decisions depends on three things: whether the parties are aiming at the same result, whether the person in front of the problem is allowed to decide it, and how far apart the people who must agree are sitting.
This chapter maps Hackman’s conditions for effective teams to the case, sets out a shared scorecard for all six organisations, and divides the project’s decisions into three levels with dollar thresholds. It then prices the time decisions take, values a joint site office against its cost, and covers the safety coordination that a shared site requires. It ends with a joint office that changed nothing for six weeks because nobody had been given the authority to decide in it.
Hackman’s conditions in the case
Hackman (2002) argued that the conditions around a team matter more than its members’ personalities or a leader’s style. Chapter 7 used the first of Hackman’s five conditions, a real team with clear and stable membership, to size the coordination group. The other four apply just as directly to the DC.
- Real teamcore team of 20 with stable membership; coordination group of 7 that meets weekly
- Compelling directionone shared scorecard of five measures for all six organisations
- Enabling structuredecision rights at three levels with dollar thresholds; meeting design from Chapter 7
- Supportive contextcontract incentives aligned with the scorecard; information in one register and one decision log
- Expert coachingfacilitated launch workshop at award and a mid-point review in month 11
The conditions are not a sequence to be completed once. A launch workshop sets the direction, but the mid-point review is where the team checks whether its structure still fits the work: in month 11 the project is moving from building to installation, and the package meetings and decision limits that suited the slab may not suit the conveyors.
One scorecard for six organisations
This is an illustration. Each organisation arrives with its own measures of success: the builder’s margin, the designer’s hours against fee, the integrator’s commissioning window, the WMS vendor’s go-live checklist. None of these is wrong, but none is the owner’s result. A shared scorecard states the project’s result in five measures that every organisation reports against, and the coordination group reviews monthly.
The scorecard is only a compelling direction if it is connected to the money. Under the alliance of Chapter 6, four of the five measures are KRAs. Under the target-cost contract of Chapter 5, the share already ties the builder to the cost measure, and the schedule and readiness measures can be added as a small incentive. Under a lump sum, only the date and the defects are naturally the builder’s concern, and the readiness measure must be carried by the owner and the integrator.
The readiness measure deserves particular care because it is the one that pays for the project. A DC that is open but picking well below 1,150 order lines an hour still needs the third-party stores and the shuttle freight, which are A$65k of the A$80k a week cost of delay. The rate itself is a forecast of the benefit, and it should be checked as carefully as the cost estimate; Schedule, benefits and the cost of delay in Reference Class Forecasting by the Numbers shows how to test a benefit forecast against past outcomes and how to price delay, and Ramp-up curves: how fast the volume arrives in Demand and Benefit Forecasts by the Numbers covers how quickly a new operation’s output builds, which is what a week-4 target assumes.
Decision rights at three levels
This is an illustration. Over the works, the project expects about 262 decisions with a cost or programme consequence, plus some 360 requests for information that need an answer but rarely a commercial decision. The decision rights put each decision at the lowest level that can safely make it.
| Level | Limit | Programme | Expected decisions | Share |
|---|---|---|---|---|
| Owner's project director | up to A$50k | no change to go-live date or scope | 210 | 80.2% |
| Joint leadership group | up to A$250k | up to 2 weeks of float | 46 | 17.6% |
| Owner's board | above A$250k | any change to go-live date or scope | 6 | 2.3% |
The owner’s project director decides anything up to A$50k that does not change the go-live date or scope, and is expected to make about 210 decisions, 80.2% of the total. The joint leadership group, the coordination group’s leads meeting weekly with the sponsor, decides up to A$250k and may use up to 2 weeks of programme float: about 46 decisions, 17.6%. The owner’s board decides anything above A$250k and any change to the go-live date or scope: about 6 decisions, 2.3%.
Three features make the thresholds work. Each level’s limit is stated in dollars and in programme effect, because a cheap decision can still cost weeks. Decisions cannot be split to fit under a limit; a A$90k change presented as two A$45k changes is a joint leadership group decision. And each level must decide within a stated time, five working days for the project director and the next weekly meeting for the joint group, or the decision escalates automatically. Decision rights before meetings: who can decide, redirect and stop, and why a chart of names is not enough covers how to write rights that hold in practice, and Escalation rules that fire on the right things covers tolerances and authority that expires.
Pricing decision latency
This is an illustration. Of the 262 decisions, 8 are expected to sit on the critical path: approvals of the racking layout, conveyor access, the controls network and similar. The time each one waits for a decision is its latency. Not all of that wait delays completion, because programme float and parallel work absorb part of it; the case assumes, as a teaching assumption, that a quarter of each wait passes through to the completion date (a transmission of 0.25).
Under the original governance, with a monthly steering meeting making decisions above the site team’s authority, critical-path decisions waited an average of 10 working days. The completion delay is 8 × 10 × 0.25 = 20 working days, 4 weeks, and at A$80k a week, excluding GST, it costs A$320k. With a weekly joint forum and delegated limits, the average wait falls to 3 working days: 8 × 3 × 0.25 = 6 working days, 1.2 weeks, A$96k. The faster rhythm saves 14 working days, 2.8 weeks, worth A$224k.
The calculation is crude, and the transmission factor is the weakest number in it: on a programme with little float it could be higher, on one with plenty it could be lower. But it makes visible a cost that rarely appears in any budget, and it shows the order of magnitude. A monthly meeting that defers a critical-path decision by a month is not saving the leadership’s time; it is spending the owner’s money at A$80k a week.
Latency can only be managed if it is measured. The project’s decision log records, for every decision, the date it was raised, the level that must make it, the person responsible, the date it is due, whether it is on the critical path, and the date it was made. The coordination group reviews the log weekly and looks at one number first: how many items are more than 10 working days old. On the DC that number is normally about 2. Chapter 8 showed it jumping to 11 when a project manager left; the error below shows it climbing for a different reason.
Co-location
Allen (1977), studying engineers in research and development organisations, found that technical communication between people falls off steeply with the distance between their workplaces, even within one building. The finding has been used ever since to argue for putting people who must work together in the same place, and a construction site is a natural place to do it.
This is an illustration. From month 10, when the building is enclosed and the materials handling packages begin, the project sets up a joint site office for 12 people: the owner’s project director and DC manager-designate, the builder’s site team, and the leads of the designer, racking supplier, integrator and WMS vendor for the days they are needed. It runs for 9 months, to the end of month 18, at A$15k a month, A$135k in total.
- Office for 12 people, A$15k a month for 9 months: A$135k
- Furniture, network and meeting room, inside the monthly rate
- Leads away from their own offices on site days
- Decision delay avoided with a weekly forum and delegated limits: A$224k
- Travel saved: 4 people × 1 day a week × 39 weeks × 8 h × A$140 = A$174,720
- RFI turnaround from 8 working days to 3 on 360 RFIs
The office saves travel: four people who would otherwise travel between the site and their own offices one day a week save 4 × 39 weeks × 8 hours × A$140 = A$174,720. It also supports the weekly forum and the delegated limits that bring the decision delay down, worth A$224k. The net is A$224k + A$174,720 − A$135k = about A$264k. Requests for information, answered across a desk instead of by email to three offices, are expected to fall from an average of 8 working days to 3; across 360 RFIs that is 1,800 RFI-days of waiting that no longer holds up the people who asked.
The A$224k belongs to the package of weekly forum, delegated limits and joint office, not to the office alone. An office in which nobody can decide anything saves travel and little else, as this chapter’s error shows.
Safety coordination on a shared site
Sharing a site shares safety duties, and the law that sets them depends on where the site is. Safe Work Australia lists Victoria, where the DC is built, as the one jurisdiction that has not implemented the model WHS laws. Victoria has its own Occupational Health and Safety Act 2004, which is structured differently and sets its own thresholds, and WorkSafe Victoria is the regulator whose guidance applies to this site. The duties each business on the DC holds, including those of whoever manages the construction work, must be taken from the Victorian Act and regulations and WorkSafe Victoria’s guidance, with advice where needed.
The model laws used in the other states and territories show, as an example only, the kind of duty that arises when businesses share a site. Under the model WHS Act, where more than one business has a duty in relation to the same matter, each must, so far as is reasonably practicable, consult, cooperate and coordinate activities with the others (model WHS Act s 46), and Safe Work Australia’s model Code of Practice on consultation, cooperation and coordination gives guidance on doing so. Under the model WHS Regulations a construction project is one whose construction work costs A$250,000 or more, and it has a principal contractor with extra duties. None of these model provisions applies on the DC as written.
Whatever the law in force, the practical work is the same and the joint office helps. On the DC, racking installers, conveyor fitters, electricians and the builder’s trades share one slab, with mobile elevating work platforms, forklifts and live conveyor sections during commissioning. The coordination group reviews the combined work plan weekly, the builder’s site rules, as the business managing the construction work, apply to everyone, and the serious-injury measure sits on the scorecard for all six organisations, not only the builder.
The error: co-located but not delegated
This is an illustration. The joint site office opened at the start of month 10 as planned. The decision levels above had been agreed in principle, but the owner had not yet formally delegated the A$50k limit to its project director, so approvals under A$50k still went back to the chief operating officer at head office, as they had before.
For 6 weeks nothing improved. The average turnaround on decisions and RFIs stayed at 8 working days, the same as before the office opened. A check of the decision log at week 6 found 23 items under A$50k waiting on owner approval, each of which the project director, sitting in the office, could have decided on the day.
The correction took a week. The board formally delegated the A$50k limit, with the go-live and scope conditions, to the owner’s project director on site, and the coordination group published the decision levels to every organisation. Within a fortnight the average turnaround fell to 3 working days. Over the six weeks, the office had cost about A$20.8k (six weeks at A$15k a month), and the 23 waiting items had each waited about 5 working days longer than they needed to, 115 item-days of avoidable waiting. The project was fortunate that none of the 23 was on the critical path.
The point is the one Bresnen and Marshall (2000) made about partnering: collaborative arrangements do not change behaviour by themselves. A shared office, like a charter or an alliance agreement, works only when the authority and incentives move with it. Collaboration with contractors is a practice, not a contract type makes the same point in words; the 23 items are what it looks like in a decision log.
Exercise: a fortnightly forum
The sponsor suggests that a weekly joint forum is too much of the leadership’s time, and proposes a fortnightly forum instead, with the same delegated limits. The average wait for a critical-path decision is then expected to be 5 working days. Using the case’s 8 critical-path decisions and transmission of 0.25, what completion delay does the fortnightly forum cause, what does it cost, and how much does it forgo against the weekly forum?
Worked answer. The completion delay is 8 × 5 × 0.25 = 10 working days, 2 weeks, costing 2 × A$80k = A$160k. The weekly forum’s delay is 6 working days, 1.2 weeks, A$96k, so the fortnightly forum forgoes A$64k of avoided delay. That is still far better than the monthly meeting’s A$320k. Whether A$64k is worth the leadership’s time depends on what that time costs. If the forum is the 8 members of the joint leadership group for 1.5 hours, a year of fortnightly forums holds 26 fewer meetings and saves 26 × 8 × 1.5 = 312 person-hours, about A$43.7k at A$140 an hour. That is less than the A$64k of delay it forgoes, before counting the non-critical decisions that also wait longer, so the weekly forum is the better use of the leaders’ time.