12+
Leading When the Plan Breaks

Бесплатный фрагмент - Leading When the Plan Breaks

Decisions and Dependable Commitments When Construction Completion Slips

Объем: 155 бумажных стр.

Формат: epub, fb2, pdfRead, mobi

Подробнее

A Note from the Author

The most expensive promise on a construction project may be the one that still sounds reasonable after its supporting conditions have disappeared. An equipment order remains valid. Work continues. The monthly report explains the variance. Meanwhile the operator plans a move, a contractor releases a crew and the last opportunity to buy a different route passes.

This book concerns the decisions that turn a slipping completion plan into an instruction people can use. Its main setting is the final stage of a construction or engineering project: equipment delivery, installation, testing and handover. I have written it for project and construction directors, engineering leads, commercial managers and the people receiving the asset.

I have led the delivery of major construction projects, including six ground space-infrastructure facilities at the Vostochny Cosmodrome built under state contracts. The concern I bring to this subject is practical. A decision has to become something people can build, test and use. A programme must allow for the resources and permissions each activity requires when it is due to start. The consequences of a missing resource or permission eventually become visible on site.

The central argument is that a commitment becomes dependable when its recipient has adequate grounds to plan around it. That requires an intelligible result, a feasible sequence, available capacity, effective authority and a receiving organisation ready to use what is delivered. Material changes have to reach the people relying on the promise. Confidence cannot substitute for a missing condition.

Chapter 2 develops the commitment test and introduces RESTORE, the seven-step sequence I use when a condition fails: Read the reality, Expose the constraint, Surface the stakes, Trace decision authority, Open the options, Reset the commitment and Examine execution. The worked cases include refusals, unresolved instructions and competing estimates. The book concentrates on those decisions rather than on every cause of project distress.

Northbank is the main case of the book. It includes a forty-activity completion model and completed working records. Westhaven is a separate countercase in which a well-presented issue remains unresolved and the delivery position deteriorates. Both are teaching cases, built so that every figure can be checked.

Published reviews of Crossrail, London 2012, the Edinburgh Tram and the Ottawa Light Rail Transit system provide a second kind of evidence. They are named and sourced, and the discussion keeps what their authors found separate from my application of it.

The book addresses management judgement. Technical acceptance, statutory permissions and contractual rights remain with the people and bodies who hold them, and immediate protective action never waits for a management exercise.

I judge an intervention by the choice it improves and the change a receiver can inspect. Sometimes the responsible result is a later date or a refusal. The book follows those outcomes as carefully as the occasions when time is recovered.

What this book is built on

Preparation combined a comparative reading of selected public material from the United States, the United Kingdom, Canada, Australia and New Zealand with the construction of a worked completion case. The source collection includes government audits, engineering guidance, procurement descriptions, an infrastructure inquiry and research on team learning. The source notes identify the material and the extent of use.

The analytical question was consistent across them: what made a delivery statement credible, what changed its basis, who could act and what happened after the decision? The comparison focuses on the decisions and delivery conditions described in each source. A programme report, an audit finding and an empirical study carry different kinds of evidence and support different claims.

For the worked schedule, GAO’s discussion of critical-path validity and float provides a reference for checking the logic behind a date. NASA’s decision-analysis guidance informs the examination of criteria, alternatives and uncertainty. Northbank’s calculations, capacity examples and decision records were built for this book and can be checked from their stated inputs.

The published project readings include Crossrail at two review dates, the London 2012 post-Games review the Edinburgh Tram Inquiry and the Ottawa Light Rail Transit inquiry. These accounts complicate a simple message about good planning: a reset can still be inadequately founded; commercial terms can frustrate the intended allocation of risk; and a fixed date requires examination of the actual arrangements surrounding it. The project comparisons keep to their historical scope; old forecasts are not presented as present facts.

Other materials broaden the comparison. Canadian procurement descriptions show how early collaboration changes the point at which a firm price is set. Australian sources connect a project’s resource promises to market capacity and contingency. New Zealand audit material shows what happens when investment decisions are made quickly on thin information. The team-learning study supplies a narrower empirical reference for reporting conditions.

My contribution is the commitment test and its application throughout the delivery process: a warning becomes a forecast; a forecast informs a decision; a decision changes an obligation; the recipient then has to use the result. The test draws on established scheduling, assurance and decision analysis and connects them at the point where people rely on a promise.

The completed forms expose the information needed for those exchanges. The separate countercase illustrates the procedure when cooperation fails. Exercises ask the reader to choose under incomplete information, and the commentaries discuss conditions that could reverse the choice. The resulting book is intended to be usable at a decision meeting and sufficiently explicit to be challenged afterwards.

How to use the book

If a decision is due today, use Before the Next Meeting. Start with the recipient’s next action: placing an order, reserving a crew, cancelling storage or accepting a result. Select the smallest record that makes that action defensible. The tools can be incorporated into existing project records.

For a continuous reading, follow Northbank through Chapters 1–20. Day 40 is the original full-handover milestone T; the delayed route finishes on day 60. The original replacement hypothesis gives day 45 but is never authorised before its offer expires. Limited warehouse use is a separate result. The case record and Appendix A retain the calculations.

Chapter 4 shows how to reconcile conflicting estimates. Chapters 5, 9 and 11 examine repeated refusals and approval that does not become an instruction. These passages use clearly marked variations that leave the main Northbank figures unchanged.

Write your recommendation before reading each exercise commentary. Name the decision, the authority and the finding most likely to change your answer. Appendix A supports calculation; the source notes distinguish published findings from my interpretation.

Before the Next Meeting

Choose the row that describes the immediate problem. Use the first tool listed; add the second only if it answers a further question. The numbers refer to the twelve tools after Chapter 20.

Здесь есть иллюстрация

Зарегистрируйтесь или войдите, чтобы увидеть ее и другие изображения

For routine, reversible coordination within existing powers, a short entry may be enough: fact, choice, authority, owner and next check. Chapter 17 gives a five-line example. Use the fuller record when the decision changes expenditure, technical configuration, an important date or another party’s reliance. Applicable approvals remain part of that work.

Leave the meeting able to state who will do what next and how the result will be checked. A completed form is useful only if it helps that happen.

The Northbank project

Northbank is a £120 million engineering and logistics complex. Its first complete operational handover was planned after an eighteen-month programme. The case models forty selected completion activities; it does not pretend to be the full project programme. When the story begins, eight weeks remain before that date. We call the original handover milestone T. All dates are expressed relative to the start of the case.

Maya is the project director. Daniel represents the client and sponsors requests that exceed Maya’s delegated authority. Priya is the engineering lead. Owen manages the principal contractor. Leah plans the remaining work. Ben leads commissioning. Ruth manages commercial controls. Elena represents the receiving operating team. Their responsibilities overlap, but their powers do not.

The operating building depends on control cabinets. The original sequence requires delivery at the end of week 4, installation during weeks 5 and 6 and integrated testing during weeks 7 and 8. A reliable warning puts delivery at the end of week 8. With the remaining sequence unchanged, full handover moves to the end of week 12, four weeks after T.

An alternative supplier offers delivery at the end of week 5 for an additional £180,000, subject to technical compatibility and commercial confirmation. That could move full handover to the end of week 9, one week after T. The original commercial offer remains open until Thursday of the first week. It is not a proven recovery plan.

An independently usable warehouse may offer limited value at T while the operating building remains closed. Its separation, permitted use and acceptance route must be established before it becomes a delivery commitment.

Part One. Seeing What Is Actually Happening

1 The Meeting Where Nobody Lies

At nine on Monday morning, Northbank was still on schedule.

The equipment had been ordered. The installation team was available. The contractor expected to finish the work in time. Each statement had a place in the report, and each speaker could explain why it was true.

Then Ben asked whether the operating building would be ready for integrated testing at the start of week 7.

Nobody answered immediately.

The equipment was ordered, but its production position had changed. The team was available in principle, but its work area had not been released. The forecast assumed that an outstanding operating requirement would be settled without disturbing the release of the cabinet drawings. The necessary client decision was absent from the decision list.

The report had preserved the date by leaving its dependencies outside the conversation.

The promise hidden in a status report

There is usually an unspoken promise inside a project dashboard. A delivery entry in a report can tell the operator when to book staff. A completion percentage encourages the client to end a temporary arrangement. A green milestone can lead another contractor to release a crew for work elsewhere. The report starts to matter when someone relies on it.

My starting question is therefore about the reader: what are they preparing to do on the strength of these words?

At Northbank, Daniel understood the report to mean that the operating team could take over at T, the original end-of-week-8 milestone. The contributors understood it to mean that their own packages were progressing. They had supplied descriptions of activity. He had received an assurance about a working facility.

That gap is the subject of this book. I call a commitment dependable when its recipient can reasonably plan around it, with the important conditions and uncertainties understood. Dependability does not require certainty about the future. It requires adequate support for the decisions the recipient will make, and a workable response if that support changes.

A forecast estimates an outcome. A commitment adds an accepted responsibility to provide a defined result. Confusing them transfers uncertainty to someone who may never have agreed to carry it.

Begin with the act of reliance

Imagine the operator must cancel external storage by Friday to avoid another month’s rent. A forecast of warehouse availability now bears directly on a payment decision. The question is no longer whether the building looks nearly finished. It is whether the operator has sufficient grounds to terminate that particular arrangement.

The construction team may have useful information without being entitled to make that decision. The operator may accept a financial uncertainty while still requiring all conditions for lawful, safe occupation. A clear account preserves those distinctions.

This approach changes the first meeting. Instead of asking each discipline whether it is on track, ask what capability the recipient needs next and what must be true for it to exist. Work backwards far enough to expose the conditions that can still defeat it. Then assign the missing checks.

Maya began with Ben’s testing question. Installed cabinets were necessary, but so were released configuration information, available supporting systems, the right people and the required permissions. An order number, by itself, confirmed none of those conditions.

Leah wrote down the conditions. Owen identified the delivery assumption. Priya identified the unresolved design input. The team had found a useful starting point within twenty minutes. Verification would take longer.

A report can be accurate and still mislead

The word «complete’ is especially troublesome. A fabricator may mean manufacture is finished. An installer may mean physical installation is finished. An operator may hear that the system can be used. Each meaning can be legitimate within its own scope; the trouble begins when a report crosses between them without saying so.

Keep the underlying measurements. If 90 per cent of a measured quantity has been installed, report it accurately. Add the outstanding condition that governs the next activity. Swapping a useful quantity for a dramatic red status just trades one simplification for another.

Nor should every uncertain item become a crisis. A late activity with available float may leave the receiving date unchanged. The task is to trace the effect, including cases where the effect is small. A method that discovers only bad news is another form of selective reporting.

What Maya could change that morning

Maya wanted to ask why nobody had told her. She would eventually need an answer. First she needed to stop the organisation making further plans around a date it could no longer support.

«Which decisions will be made before we have checked these conditions?» she asked.

That question produced a shorter list than the general risk register. Ruth needed to preserve a supplier decision window. Ben needed a credible basis for booking the test team. Elena needed to know whether warehouse availability was sufficiently established to alter storage arrangements.

Maya assigned owners and requested the first verification results that afternoon. The team retained the original milestone as its reference. It stopped presenting that milestone as an assured completion date.

This was an incomplete correction. Nobody had recovered a day. The report had become more useful because it no longer invited an unsupported next step.

The limit of a better question

A question can expose a gap without giving the manager power to close it. The client may withhold an operating requirement. A supplier may refuse further detail. A sponsor may dislike the consequence and keep repeating the old date. Those situations require decisions about resources, authority or commercial action. They cannot all be solved by greater candour.

The chapters that follow examine that harder part of the work. The standard is a promise another person can use, a decision made by someone entitled to make it, and enough traceable reasoning to revise either when circumstances change.

On site: the boiler house that was 100 per cent complete

On a remote, state-funded site, the monthly report showed the boiler house at 100 per cent installation. The client booked its acceptance commission for the following Tuesday and flew three members in from the regional capital. On Monday the commissioning engineer asked for the pressure-test records. The welding was finished. The tests had not been witnessed, because the inspector who had to witness them was on another site four hundred kilometres away until the end of the month.

Nobody had lied. The installer meant «installed’. The client heard «ready for acceptance’. The commission’s wasted journey was the first cost of that gap; a month’s delay in heating the neighbouring buildings was the second.

The fix was one line in the report template: next to every percentage, the next act it permits and the evidence that act requires. «Installation 100 per cent; pressure test not witnessed; acceptance not possible before the inspector’s visit on the 28th.»

Decision practice The report that triggers a cancellation

A warehouse package is reported as 92 per cent complete. The operator must decide by 15:00 tomorrow whether to cancel external storage. Cancellation would save rent but remove guaranteed capacity; replacement storage is not assured. The remaining works include the only delivery entrance and a fire-system interface whose review is due in three days. The project director is asked to confirm the opening date now.

Draft the reply and decide which action can be supported today. Identify a piece of information that could reverse your recommendation. You may propose a provisional plan, but state what the operator must not infer from it.

2 What Makes a Commitment Dependable

Before asking whether Northbank could keep its date, Maya had to establish what the date meant. Full operational handover and limited warehouse use were different results. Combining them would make almost any progress report easier to defend and the operator’s planning harder to trust.

The commitment test begins with a sentence: who will provide what usable result, to whom, by when, and on which material conditions?

If that sentence cannot be written clearly, the team is not ready to assess how much confidence the promise warrants. It is still defining the obligation.

Follow the promise through its necessary conditions

A significant commitment rests on several kinds of support. I examine them separately because strength in one cannot compensate for a missing essential condition elsewhere.

Здесь есть иллюстрация

Зарегистрируйтесь или войдите, чтобы увидеть ее и другие изображения

This is a set of questions, not a scoring model. Five satisfactory answers cannot compensate for one failed mandatory condition. Nor do the questions demand that every uncertainty be eliminated. They identify who needs to understand and carry the remaining uncertainty.

The replacement supplier illustrates the difference. A price and delivery slot may be sufficiently firm to justify funding a technical assessment. They are insufficient to promise operational handover if compatibility is still unresolved. The same evidence can support one act and fail to support another.

Distinguish uncertainty from a broken condition

There are three materially different findings.

An unknown condition requires investigation or an explicit decision about exposure. An unmet condition requires a change to the route or the claim. A previously adequate condition that has ceased to hold requires prompt review of the commitment and of the actions people have taken in reliance on it.

For example, uncertainty about a productivity rate can sometimes be carried through a range of completion dates and a funded contingency. A confirmed inability to meet an essential functional requirement excludes the configuration until an acceptable change is established. A supplier’s changed factory slot alters a dated production premise and requires recalculation.

Recording all three as «risk’ discards the distinction that should govern the response.

Four possible decisions

When the support is incomplete, the next step is a choice about the commitment itself.

Retain it when the changed information is immaterial to the promised result or the remaining margin absorbs its effect. Record why. The presence of a warning alone does not justify rewriting every agreement.

Qualify it when a specific uncertainty remains and the recipient can knowingly plan with that condition. Identify who will resolve it, when, and what follows if it fails. «Subject to everything going well’ is no usable qualification.

Reset it when the scope, timing, resources or accepted exposure must change. Obtain the powers and instructions needed for the new arrangement. A revised forecast can be issued before that process is complete, provided its status is clear.

Suspend reliance on it when a mandatory condition has failed or support is too weak for the next act that depends on it. This can mean withholding an opening announcement while construction continues. It does not automatically mean stopping the entire project.

These are management descriptions. A contract’s obligations, notice requirements and approval arrangements still determine the formal effect of any proposed change.

Match the burden of proof to the next act

The evidence needed to fund an investigation differs from that needed to release a warehouse for use. The comparison below follows the same proposition through successive decisions, so that the change in burden is visible.

Decide what the evidence permits

Take one proposition: the warehouse can operate independently of the unfinished process building. The following examples concern that same proposition at successive decisions.

Здесь есть иллюстрация

Зарегистрируйтесь или войдите, чтобы увидеть ее и другие изображения

An absent mandatory finding remains a failed gate. The investment authority may accept a forecast cost range within its delegation; it cannot accept an unverified essential service on behalf of the person responsible for that technical finding. Priya coordinates the specialist evidence. Elena confirms operational suitability and the operator’s readiness. Daniel exercises the delegated investment power. Any statutory decision stays with the body under the local arrangements.

Write residual uncertainty in terms of a consequence and a response. For example: «The remaining logistics estimate is £12,000, with a further £3,000 exposure within the approved contingency; Ruth will confirm the order before commitment.» That can be a tolerable investment uncertainty. «The essential fire-system interface has not been resolved’ cannot be turned into an occupation condition by getting a commercial signature.

Evidence is sufficient for an act when it answers that act’s necessary conditions, has been assessed by the competent people and leaves only uncertainty the authorised decision-maker may carry. If two reviewers disagree, identify the disputed condition and seek the required finding. Averaging their confidence would settle neither the condition nor the power to decide.

Keep the baseline forecast and approved changes distinct

The original baseline belongs in the record. So do authorised changes, the current forecast and any provisional options. Each answers a different question.

Northbank’s full-handover baseline remains day 40. Its delayed model produces day 60. A replacement hypothesis produces day 45, but never becomes an approved route under the original offer. Limited warehouse use may begin on day 40 if its separate conditions are met.

None of these facts needs to erase another. Reporting only day 40 conceals the forecast. Replacing the original reference with day 60 conceals the variance. Presenting the warehouse as «the project opening’ conceals the reduced result.

Preserve the history and state which decisions the current information supports. When a recipient knows what they can use and what would change it, the report has done more than describe progress.

RESTORE: the response when a condition fails

The commitment test tells you what support a promise needs. Sooner or later that support is questioned. A warning arrives, an estimate moves, a decision is withheld. At that point I work through seven steps. Their initials spell RESTORE.

Read the reality. Separate what has been observed from what is forecast, assumed or unknown, with a date and a source for each claim. Chapters 1 and 3; Tools 1 and 2.

Expose the constraint. Trace the missing condition through the remaining work to the result, and check the people behind the bars. Chapter 4; Tool 3.

Surface the stakes. Establish the pressures and obligations shaping each party’s response, and test promises against evidence rather than confidence. Chapters 5–8; Tools 4 and 5.

Trace decision authority. Name who can make the specific change, who advises, who funds, who implements and the last useful time to act. Chapters 9 and 11; Tool 6.

Open the options. Compare the admissible routes, including waiting, a narrower result and stopping, with the assumptions that could reverse the choice. Chapters 10 and 12; Tools 7 and 8.

Reset the commitment. Agree the result again with its recipient, scope, timing, owner, resources and review trigger, keeping the original baseline visible. Chapters 13–15; Tools 9 and 10.

Examine execution. Inspect the first observable act and the recipient’s ability to use the result, then carry the lesson forward. Chapters 16–20; Tools 11 and 12.

The sequence is not a ritual to complete in order every time. Enter at the step the situation requires and return to an earlier one when a governing condition changes. A routine, reversible matter may need only a five-line entry; a change to expenditure, configuration or another party’s reliance needs the fuller record. The rest of the book follows Northbank through each step.

Decision practice One condition changes

A client has accepted a provisional warehouse move, subject to independent electrical services being demonstrated. That demonstration succeeds. The operator now proposes storing temperature-sensitive goods, whereas the assessment covered ambient storage only. The move date and the supplier’s physical work have not changed.

Choose whether to retain, qualify, reset or suspend reliance on the commitment. State which part of the existing assessment remains useful and who must resolve the changed use. Would your answer differ if the goods were unchanged and only their quantity increased?

3 Why Bad News Arrives Late

The supplier warning had existed before Monday’s meeting.

Owen had received a cautious message late the previous week. It referred to pressure on the production line and asked for confirmation of several details. He had spoken to the supplier, heard that recovery was still being explored and decided not to raise an alarm until the position was clearer.

He did not describe this as hiding information. He described it as avoiding unnecessary escalation.

For Maya, the distinction offered little comfort. The alternative supplier’s commercial offer would expire on Thursday. Waiting for certainty was consuming the time available to investigate a different route.

A warning has a personal cost

A person reporting trouble may lose credibility, inherit a problem outside their powers or face questions about prevention. They may also doubt an interpretation based on incomplete information.

A manager who asks for early warnings but punishes tentative information creates a contradiction. People can comply with the request for speed or the demand for certainty, but not both. Many resolve the contradiction by delaying the warning until they can defend it.

Research by Amy Edmondson in 51 manufacturing work teams found an association between team psychological safety and learning behaviour. The study supports attention to the interpersonal conditions in which people raise questions and discuss errors. It does not prove that one reassuring statement will create reliable reporting in any project. [S2]

Your task is to make the reporting rule workable. State what level of uncertainty justifies notification, what initial evidence is needed and what happens next. Then behave consistently when the next uncomfortable warning arrives.

Separate notification from accusation

An early-warning record should not read like a completed investigation. It can say: «The supplier indicates a possible production delay. The effect on delivery is unconfirmed. If the later date is verified, integrated testing may move. Owner to obtain written production position by 15:00.»

That statement does not blame the supplier or announce an established four-week delay. It gives the organisation a reason to protect the decision window.

Maya adopted that notification rule with Owen. She also changed her first response: establish the signal and assign verification before examining responsibility. Otherwise the request for speed would still reward people who waited until they could defend every detail.

Ask about the route the information took

«Why didn’t you tell me?» often produces a defence of the person’s intentions. A more useful investigation begins with the route.

When was the signal first observed? What did it say? Who saw it next? What action was expected? What made the person wait? At which point did the information stop moving, lose specificity or become disconnected from the milestone?

These questions can reveal a missing reporting threshold, a commercial instruction, a repeated demand to «come with solutions» or an overloaded manager. They can also reveal deliberate concealment. The diagnosis should follow the evidence.

The instruction to bring solutions deserves care. It can encourage ownership when the person has relevant authority and resources. It can suppress warnings when the problem is outside their power. A supplier’s production failure may require client funding or a technical substitution. An engineer should not have to solve those matters before reporting the signal.

Make response predictable

When a report says «awaiting clarification’, I look for the question, its recipient and the date when the answer could still change the decision. Without those details, the phrase can conceal an ownerless problem. Tell the originator what happened to the warning; otherwise early reporting becomes unpaid work with no visible consequence.

When reassurance is insufficient

Sometimes the person’s fear is reasonable. Their immediate manager may retaliate, the organisation may reward unrealistic forecasts or a contractual process may discourage openness. A team-level conversation cannot solve every structural problem.

Use established escalation, assurance and reporting channels appropriate to the issue. Keep records accurate. If you lead the organisation, inspect whether incentives, payment rules and performance reviews contradict your reporting policy.

At Northbank, Owen’s delay had both personal and commercial dimensions. He wanted to retain control of the problem and avoid appearing unable to manage his supplier. Maya could reduce the cost of early notification. She could not erase his contractual responsibilities through a friendly conversation.

The aim was to report uncertainty early enough to influence a decision.

On site: the wagons that stood for eleven days

Cable drums for a remote site travelled by rail and then by road. A supply foreman learned that the wagons were standing at a transshipment station because of a mismatch in the shipping documents. He expected it to clear in a day or two and said nothing. Raising it, he felt, would look like losing control of his own delivery.

The wagons stood for eleven days. By the time the drums reached site, the temperature had fallen below the limit at which that cable could be pulled cold. The heated tent and the extra shift cost more than the original freight.

The review did not blame the foreman. It found that the site’s reporting rule asked for «problems with solutions’, so a problem he could not yet solve had nowhere to go. The new rule was simple: any delivery that stops moving for more than 48 hours is reported the same day, with what is known and who is checking.

Decision practice The uncertain email

A junior engineer receives a cautious supplier email about a possible component-test failure. Verification will take two days. A scarce replacement slot must be reserved tomorrow, at a non-refundable cost within the manager’s normal spending limit. The message is credible but does not establish that the specified component will fail.

Decide what should be notified now and whether reserving the slot is justified. Identify the evidence needed for that limited expenditure and the separate evidence needed for a replacement purchase.

4 Find the Constraint Before Adding Pressure

Daniel asked for a second shift before Leah had finished explaining the supplier’s revised dates. Owen said he could price it. Ben asked which activity the shift would advance.

That question changed the discussion. It was possible to spend money that afternoon. It was harder to show what the money would buy.

Leah put the remaining sequence on the screen. The old report had shown packages. Her new view showed the route from an unanswered operating requirement to an accepted working system. Some activities could proceed together. Others required a particular result from somebody else. The difference mattered more than how busy each team appeared.

Make the logic open to examination

The commitment test asks whether the sequence can produce the result people plan to use. Here is the calculation behind that question.

Northbank’s detailed model contains forty activities. Appendix A gives their durations, predecessors and calculated dates. It covers a selected completion sequence; the rest of the £120 million project sits outside it. The case assumes the excluded work is sufficiently advanced and does not govern handover; discovering otherwise would require that work to be added. There is no hard date constraint forcing the model to finish on time.

Day 0 is the beginning of the first working week. The calendar contains five eight-hour working days per week, without holidays. A zero-duration event records a decision or acceptance boundary. Its zero duration means that preceding work has provided for the review; it does not mean that a real reviewer needs no time. All modelled links are finish-to-start. These choices make the calculation inspectable. A live programme may require other calendars, relationships and approval allowances.

The baseline sequence is specific. The operating requirement and interfaces are resolved by day 5; the configuration is frozen by day 7; the drawing package is released by day 10. A verified material kit is available by day 7. Factory queue and assembly then occupy six days, factory acceptance checks two and transport two. Cabinets reach site on day 20. Positioning, termination and pre-energisation checks take ten further days. The operational test sequence takes another ten. Full handover is supported at day 40, which the case calls T.

These durations are case estimates. Their arithmetic can be checked; their realism on an actual facility would need production records, supplier commitments, quantities, methods and the engineering assessment. I have made the assumptions visible because a plausible-looking schedule can conceal weak estimates as easily as a persuasive speech can.

The supplier’s new position changes factory queue and assembly from six days to twenty-six. Cabinet arrival moves from day 20 to day 40. The remaining logic then produces full handover at day 60. The finish date has not been imposed as a constraint. The result follows from the changed input and the dependencies.

Trace the route that governs the finish

In this model, each activity’s earliest start is the latest early finish among its predecessors. Add the duration to obtain its earliest finish. Work backwards from the calculated project finish to establish how late each activity could start without delaying that finish. The difference between late and early start is total float in this simple unconstrained model.

This is elementary critical-path arithmetic, but what it shows has consequences. An attractive intervention must reach a path that governs the result. Otherwise it may make a package look better without advancing handover.

Consider the cable branch. Area release finishes on day 8. Containment takes five days, cable pulling four and circuit testing three: the branch finishes on day 20. Termination begins on day 22 after cabinet positioning. The cable branch therefore has two days of total float in the baseline model. It has twenty-two days of total float in the delayed model, because cabinet arrival has moved. The additional float does not indicate improved cable-team performance. It is room created by deterioration elsewhere.

Conversely, a baseline cable overrun of three days would use the two-day reserve and delay handover by one day, assuming the rest of the model holds. The governing route would change. The critical path must therefore be recalculated when durations or dependencies change.

GAO’s schedule guidance is a useful reference here: its discussions of critical-path validity and reasonable float examine how missing logic, constraints and resource limits can distort the apparent route. The Northbank calculations are original worked examples rather than a reproduction of GAO’s case studies. [S12]

Check the people behind the bars

Leah’s first network calculation assumes that the stated resources are available. That assumption requires a separate examination. Dependency logic does not prevent the same six electricians being promised to two sites.

For A18, termination and identification, the case allows 240 productive person-hours. With six electricians delivering eight productive hours per day, the arithmetic gives five days. If only four are available, the same quantity takes 7.5 days before any further effects are considered. The delayed finish moves from day 60 to day 62.5. The assumption about productive hours excludes disruption; it simplifies the case and sets no overtime rule or productivity rate.

The resource record therefore names the crew, period, supervision, access and required inspection support. A separate two-person test team is reserved for the sequential checks. The training cohort does not supply the witness team at the same time. Appendix A states these boundaries so that the model does not borrow the same people twice.

This is where remote-site plans often become less convincing. A workforce number says little until it is connected to skills, travel, accommodation, induction, materials and the work area. A recruitment plan does not confirm that a suitably qualified termination crew will be available next Tuesday.

Infrastructure Australia’s market-capacity work provides the wider context: competing infrastructure demand and shortages make resource assumptions a portfolio question as well as a site question. Its forecasts are not a substitute for confirming Northbank’s particular team. [S3]

Build a forecast from estimates that disagree

The following estimating variation starts with unfinished termination work. It is separate from Northbank’s fixed 240-person-hour A18 account. The contractor says four days, the planner says six and the supervisor says nine. Averaging them would conceal the different work and resources each has assumed.

Leah asks them to reconstruct the remaining quantity before debating the date. The current approved register contains 120 connections. Traceable records support completion of 24, leaving 96. A markup proposes another 12 connections, but their design status is unresolved. The four-day proposal assumed all 120 still remained and eight fully productive workers; the six-day estimate assumed 96 with six workers. The supervisor expected only four workers and allowed time for the additional scope. None of those is yet a common basis.

Здесь есть иллюстрация

Зарегистрируйтесь или войдите, чтобы увидеть ее и другие изображения

The base allowance is 96 × 1.75 +30 = 198 productive person-hours. The additional-scope scenario is 108 × 1.75 +30 = 219. Both include the same setup once.

With six people available throughout, capacity would be 36 productive person-hours a day. The two durations would be 5.5 and approximately 6.08 working days. That explains the planner’s six-day neighbourhood, but it ignores the actual reservation.

The confirmed four-person crew supplies 24 productive person-hours a day for the first three days: 72 in total. At six people thereafter, the remaining 126 base hours take 3.5 days, making 6.5 days overall. With the additional scope, the remaining 147 hours take approximately 4.08 days, making 7.08 days overall. This staged-capacity calculation assumes the work can be allocated to those people without another bottleneck; if inspection or access limits output, that condition must also be modelled.

Leah issues a conditional task forecast of 6.5–7.08 working days, separating the scope scenarios. She does not call it a probability range or a full-handover forecast. The lower end requires the approved base scope, the confirmed crew and the assumed productive output. The upper end adds the twelve connections without further redesign or delay. Late design approval could put completion beyond it.

For today’s decision, the resource owner confirms the staged reservation, the design lead resolves the twelve disputed connections, and the supervisor checks output against the same quantity register after the first representative shift. Existing approved work may continue only under its established releases and method. The unresolved addition is not released by appearing in a forecast. Leah then links the revised duration to downstream installation, review and testing windows before publishing its handover effect.

The comparison has reconciled three claims without declaring one person the unreliable estimator. The useful question was what each estimate contained. I distrust a precise duration when the team cannot show which remaining quantity it covers.

Decide what additional effort would change

Maya asked Owen to return with a proposal against named activities. It had to show the revised method, crew and access arrangement, the receiving team’s response and the resulting finish. «More people» was an input to that explanation.

The team first protected what could still be influenced. Priya investigated the replacement’s interfaces. Ruth obtained the actual commercial exclusions. Leah checked the later crew reservation. Daniel was asked to resolve the outstanding operating requirement through the client’s authority. These actions addressed different links; none could stand in for the others.

A proposal to reduce termination from five days to four would move the delayed model from day 60 to day 59 if its assumptions held. It could be worth buying. It could not recover the twenty days lost in factory availability. The calculation gave Maya a measured answer to the pressure for a dramatic recovery.

I would give greater weight to a recovery proposal whose effect can be traced through the remaining work. The former can be challenged and improved. The latter usually becomes somebody’s promise before anyone has found the missing conditions.

The working question is simple: after this intervention, what can start earlier, and what still prevents the intended result? Keep asking until the answer reaches the milestone or exposes the boundary of the proposal.

Decision practice Which branch should receive the crew

Use the Northbank baseline in Appendix A. Cable circuit testing A12 is now forecast to take six days rather than three. A supplier proposal could shorten A06 by one day, at a price the client considers modest. Another proposal restores A12 to three days at a higher price. Only one intervention can be funded.

Calculate the finish under each proposal, with the changed cable duration in place. State what you would recommend if time were the only objective, then name the missing information required for an investment decision.

Part Two. People and Uncomfortable Truth

5 The Promise That Asked for Another Week

On Tuesday, Owen brought a revised promise from the original supplier. There might be a way to recover some time. A production manager was reviewing priorities. A fuller answer would follow next week.

Daniel heard a reason to wait. Maya heard an untested possibility competing with a commercial offer that expired on Thursday.

Owen had not invented the supplier’s statement. He had also not obtained a reserved production slot, a verified recovery sequence or a date the supplier was prepared to support with evidence.

The statement invited another week of patience before anyone had established what that week could achieve.

Distinguish intention from forecast

An intention expresses a desired action. A forecast estimates an outcome under assumptions. A commitment assigns responsibility for delivering a defined result under agreed conditions. People use the same future tense for all three.

«We will recover» might mean «we want to recover», «our current model shows recovery if several things happen» or «we have secured the means and accept responsibility for this date.» Those are materially different statements.

Ask the speaker to describe the basis without forcing a legal classification in a casual conversation. What has changed since the previous forecast? Which resource or sequence now supports the improvement? What early evidence will demonstrate that the recovery is occurring?

What would have changed things at Northbank was a confirmed manufacturing arrangement. A more senior person’s confidence did not. A senior person might be able to secure that arrangement. Their seniority did not establish that they had done so.

Investigate incentives without mind-reading

A repeated promise can reflect optimism, weak planning, incomplete information, cash pressure, a dependency or deliberate delay. Selecting the most suspicious explanation prematurely can make your response less effective.

If the supplier’s actual problem is an unresolved configuration, a threat will not resolve it. If the supplier is protecting a production slot for another customer, technical assistance alone may not change its priority. If it lacks funds to purchase components, a new schedule may simply conceal a financing problem.

Ask questions that separate explanations. «What is the next production event that has not yet occurred?» «What stops it happening?» «Who can change that condition?» «What information can we inspect?» The answers may reveal a problem the other party has been reluctant to state directly.

This is an inquiry into performance. Avoid diagnosing character where an observable explanation will suffice. If evidence establishes misrepresentation, preserve it and follow the commercial or organisational process.

Inspect the conditions created by the contract

The party asked to promise may not control every relevant risk. The client may retain design decisions, access or approvals while expecting a contractor to provide an unconditional delivery date.

UK Cabinet Office guidance on risk allocation emphasises placing risks with parties reasonably able to manage them and warns of the consequences of inappropriate allocation. This guidance has a defined procurement context; it is not a substitute for interpreting an individual construction agreement. It does offer a useful commercial question: what behaviour does this allocation make likely? [S4]

The answer may include a risk premium, narrow compliance, disputes or reluctant early disclosure. None of those outcomes is inevitable. They should be considered when a promise seems detached from the conditions needed to honour it.

For an existing project, start by establishing the actual agreement and the authorised route for change. A manager should not casually waive obligations, promise payment or alter rights in an attempt to improve a relationship. Better information and properly authorised commercial action belong together.

What Edinburgh shows

The Edinburgh Tram Inquiry identified the infrastructure contract as the principal cause of the project’s delays: its terms led to disputes over interpretation and to work stopping until they were resolved. It recommended that where a business case assumes design, approvals and utility diversions will be complete before the construction contract is awarded, negotiations should wait until they are, or a new business case should be prepared. [S16]

The lesson I take is specific: a commercial label does not establish the state of the dependencies or determine what the full contract does when an assumption changes. A delivery director needs the commercial and legal assessment of that mechanism before relying on the label.

What Priya needed before recommending replacement

The replacement’s envelope drawing looked close enough to encourage optimism. Its interface schedule also listed the required 48 digital inputs, 24 digital outputs, eight analogue channels and two communication links. Matching the counts did not establish that the equipment would behave as the receiving design required.

Priya identified two unresolved items. The existing design and the offered cabinet did not yet show the same division of responsibility for supplying loop power. The proposed communications information also failed to establish the required restart behaviour after a loss of connection. Neither point proved the replacement impossible. Both required a resolved design and supporting evidence before the route could support a handover promise.

Her review separated matters that could be checked from documents from matters requiring demonstration or further design. A vendor’s general statement of equivalence would not close the second group.

Здесь есть иллюстрация

Зарегистрируйтесь или войдите, чтобы увидеть ее и другие изображения

By Wednesday’s review, the record still had the loop-supply and restart items open. Ruth could not price the potential site changes accurately, and Ben could not yet confirm whether the quoted factory work supplied all the evidence his site sequence required. The £180,000 offer therefore remained a commercial offer attached to a technically unverified option.

This is the precise uncertainty behind Daniel’s refusal. Whether or not replacement would have worked, the record shows why the apparent three-week recovery and £90,000 advantage had not become an adequately supported recommendation before the offer expired.

Make another week earn its place

Sometimes waiting is the right decision. An investigation may resolve uncertainty cheaply. A supplier’s recovery may become verifiable within hours. The test is whether waiting has a defined information benefit that is worth the options it consumes.

Maya asked Owen what would be known by Wednesday afternoon that was not known on Tuesday. He requested a written production sequence, confirmation of component availability and a named escalation contact. Those were useful outputs. «We will know more next week» was not sufficiently specific when the alternative offer expired this week.

She set a parallel path. The team would continue investigating the original supplier’s recovery while testing the replacement option. This consumed some professional effort, but it protected a choice. It also made clear that parallel investigation was not authority to incur the full replacement cost.

A conversation that tests the promise

Try the following sequence, adapting it to your role.

«What result are you now forecasting, and what does complete mean here?»

«What has changed since the last date?»

«Which necessary conditions are already secured, and which depend on somebody else?»

«What will we be able to verify before the next decision point?»

«What should we do if that evidence does not appear?»

The final question matters. Without it, the group may have an elaborate discussion and return to the same open-ended extension. A fallback can be a further review, a different supplier, a scope decision or an escalation. It must be an action the people in the room can authorise.

When a second reply still contains no evidence

Consider a variation on the first Northbank week that tests the response to a supplier who continues to withhold detail. The main chronology is unchanged.

At Tuesday’s review Owen supplies a recovery slide showing delivery restored to the original date. It has no component-release dates or factory allocation behind it. Asked for those records, he says the supplier will not disclose its production plan to a customer.

Maya narrows the request. She needs confirmation that the critical components are available, that assembly capacity is allocated and that the proposed checks and transport connect to the receiving date. Ruth offers a redacted extract or a call with an authorised production manager. They do not demand unrelated customer information.

By Wednesday morning, the second reply is another slide. Owen says, «They have confirmed it. Asking again will damage the relationship.»

«What did they confirm that I can now use to book installation?» Maya asks. «A named slot and its prerequisites would change the forecast. The repeated date does not.»

She sets the next useful return before the replacement decision window closes. Priya continues the separate compatibility review; Ruth checks what limited reservation, if any, the supplier would really offer and who could authorise it. No reservation is assumed from an unanswered request.

The third reply refuses both the redacted evidence and the production call. Maya stops commissioning versions of the same unanswered enquiry. The recovery claim is recorded as unsupported; the latest evidenced late-delivery position remains the planning basis, with its remaining assumptions stated. The replacement still requires its own technical and commercial decision. Owen’s supply-chain management obligation is not transferred to Maya because she asked the questions.

Her briefing is short: «Original-date recovery remains unverified after two specific evidence requests and an offered confidential route. We cannot release the receiving resources on that promise. The live decisions are the qualified replacement investigation and planning around the supported later route.»

This response does not determine dishonesty, waive a contract or prove the supplier incapable of recovering. Ruth handles any required commercial notice through the actual contract route. The immediate management consequence is narrower: the unsupported claim cannot govern another team’s irreversible action. Further investigation needs new evidence or a new question, not another copy of the request.

I would stop asking whether the supplier is confident at this point. I would ask which commitment I am still being invited to make on its behalf.

The promise and the person

A fourth missed date invites a judgement about the person who gives it. Sometimes that judgement is supported. Sometimes the date conceals a requirement the client has still not supplied. The manager needs a way to distinguish the two without allowing either to become an excuse for endless deferral.

The same review should expose both the changed production basis and the unresolved dependency. The owner of the external condition receives an action just as explicit as the supplier’s. A commitment that depends on a client instruction is reported with that condition visible; it is not quietly presented as unconditional and disputed later.

This approach does not absolve the supplier. It creates a fairer and more useful performance record. If the client answers on time and the supplier still cannot support the date, the next discussion begins from clearer evidence. If the client does not answer, the forecast should show the effect before the date is missed.

Decision practice A date with an unanswered client input

A supplier has missed three dates. It now offers a fourth, conditional on receiving a client configuration decision by noon tomorrow. The client decision is late, but the supplier’s own material availability is also unconfirmed. The sponsor asks you to «hold the supplier to its promise’.

Write the questions and actions you would assign to each party. Explain how you would report the date if the client responds on time but the material position remains unknown.

6 When Disagreement Looks Like Disloyalty

Priya challenged the replacement option before the team had finished discussing its price.

«We are assuming the controls will work with what is already installed,» she said. «I haven’t established that.»

Daniel responded that the project needed solutions. Priya said the project needed an answer to the compatibility question before it purchased one.

The exchange was becoming personal. Owen later described Priya as technically excellent and permanently opposed to anything urgent. Ben said she was the only person asking the necessary question.

Both descriptions were too broad to support a decision.

Extract the claim from the performance

A person can identify a real problem in an unhelpful way. A person can also express an unsupported objection calmly and fluently. Tone and content require separate attention.

Ask what precisely could fail, through what mechanism, under what conditions and with what evidence. Priya’s objection concerned compatibility between the alternative cabinets and the installed system. That was a testable claim. Her manner might need discussion, but it did not determine whether the claim was true.

Whether the speaker is respected or irritating, tone cannot establish technical validity.

A useful response would have been: «We need to test that. What information is missing, who can provide it and when could you give a conclusion with its limits stated?» The phrase «qualified conclusion» allows for a result such as compatible subject to specified modifications. It does not demand a simple yes from a complex assessment.

Convert opposition into an examinable question

People often say «that won’t work» because they have compressed a long history into a short sentence. Ask them to unpack it.

When they say they have tried before, establish what the previous attempt involved and whether the same conditions remain. The earlier failure may identify a relevant dependency. It may also be a poor analogy.

Ask what observation would change their view. An expert who can state that condition helps the team investigate efficiently. If no possible evidence would change the position, examine whether the objection is really about a different concern: loss of control, unacceptable liability or an organisational conflict.

Do not require the objector to produce a complete alternative before their warning is heard. A structural engineer may be right to identify a technical concern without being the person who can secure another supplier. Asking for constructive participation is reasonable. Making the warning conditional on solving the whole project is not.

Set a boundary around conduct

Useful dissent does not authorise humiliation. If someone insults a colleague, interrupt the behaviour and return to the issue.

«The technical concern needs an answer. Personal comments will not help us establish it. Describe the failure mechanism and the evidence we need.»

Apply the same standard upwards. A sponsor who dismisses a specialist as negative can suppress information as effectively as an abrasive specialist can intimidate a junior colleague. Your process loses credibility if only lower-status people are expected to behave professionally.

Address repeated conduct privately with specific feedback and a clear expectation: let others explain the evidence, avoid personal remarks and state the objection precisely.

Decide who judges the technical question

A meeting chair is not automatically the competent technical authority. If the group cannot resolve a significant issue, assign the review to qualified people. Clarify who can accept the resulting recommendation and what approvals remain.

At Northbank, Priya would lead the compatibility assessment with the supplier and relevant specialists. Ben would identify implications for integrated testing. Maya would integrate their conclusions into the decision note. Daniel would decide on the expenditure within his authority. These were connected but distinct responsibilities.

Raising a concern did not give anyone a veto over unrelated matters. Equally, a commercial majority could not vote technical compatibility into existence.

Give the finding a visible consequence

If an objection is substantiated and affects the proposed course of action, revise the plan accordingly. If it is not substantiated, explain the basis for proceeding and any monitoring required. If uncertainty remains, state its significance and who accepts the decision within their authority.

An organisation that invites objections but never shows how they affect decisions teaches people that the invitation is ceremonial. That can be worse than a clear, limited process because it consumes their effort and expectations.

By Wednesday, Priya had established that the replacement option still had unresolved configuration questions. The initial quotation did not establish compatibility at every required interface. This did not prove that replacement was impossible. It meant the advertised delivery date alone could not support the recovery promise.

Daniel disliked the uncertainty. Maya needed it preserved in the decision paper. A later, more expensive discovery would be less convenient still.

Decision practice A valid warning delivered badly

During a recovery meeting, a senior engineer calls the planner incompetent and identifies a possible access conflict. The engineer has not reviewed the latest access drawing. The drawing’s author says it resolves the issue, but the operating team has not reviewed the revised arrangement. The purchase decision is due in an hour.

Describe your first response and the shortest credible review route. Identify a finding that would permit the decision and one that would require it to return, regardless of the participants’ manner.

7 The Expert Everyone Depends On

Ben could answer a readiness question in minutes. Without him, the same question could occupy a meeting and still remain open. Northbank had gradually organised part of its commissioning work around that difference.

Maya appreciated the rescues. She also kept assigning the next ambiguous interface to Ben. His indispensability was partly a consequence of her choices.

The important distinction is between expertise the project really needs and dependence it has created around the expert. Removing the latter should improve access to the former.

Find what only the expert can do

Take a recent interrupted handover and reconstruct the missing act. Did it require professional judgement? An interpretation of an agreed criterion? Access to configuration records? A decision that nobody else was authorised to make?

These are different dependencies. A competent deputy may be able to apply a documented ordinary-case rule. An administrator can arrange access to a controlled record. Neither can acquire engineering judgement by receiving a longer checklist.

Ben and Maya reviewed the last few questions together. Some were significant exceptions. Others recurred because the supplier’s configuration records were poorly indexed or because the receiving team did not know whose finding it needed. The project had been paying for scarce judgement to perform avoidable searches.

The claim that nobody else is competent

Sometimes that claim is true. Require a concrete account of the tasks and competencies involved, then address the capacity problem. Naming a deputy does not provide cover unless that person can perform the required tasks.

Sometimes the claim protects an expert from review. The response is still a concrete examination. Ask for the reasoning behind one important finding, the information used and the conditions that would change it. Arrange competent independent review where appropriate. The aim is to find out what the decision rests on. Nobody is trying to prove the expert replaceable.

Бесплатный фрагмент закончился.

Купите книгу, чтобы продолжить чтение.