There is a window early in every initiative when changing your mind is nearly free and nobody yet knows enough to want to. It is also the window most organisations staff most thinly.
Draw four curves against time for any capital initiative. Level of effort starts low, rises through delivery, falls at handover. Potential for adding value starts high and declines steadily. Influence of stakeholders does the same. Potential cost of changes to plan does the reverse — near zero at the outset, climbing through the life.
That diagram appears in teaching material as an illustration of lifecycle characteristics. Read as a resourcing argument it says something uncomfortable: the period when your ability to shape the outcome is greatest is the period when you have committed the least effort to shaping it. The curves cross somewhere early, and after the crossing every organisation is buying decisions at a premium.
Most executives know this in the abstract. Very few have looked at where their own effort actually sits relative to that crossing, and fewer still have distinguished the two entirely different things that get called a change.
The Strategic Context
Flores and Chase, writing on front-end project controls in Cost Engineering in April 2005, make the case from construction cost control, and their framing is more useful than the familiar version.
They observe that projects rarely run exactly to plan, that changes are inevitable, and that "the later the change enters the project process, the more costly". They then offer a proportional claim that must be reported with the hedge they attached to it: "a case can be made that approximately 80 percent of the project costs are established in the first 20 percent of the project life". That is a proposition the authors advanced, not a finding they demonstrated, and it should be used to frame a question rather than to close one.
They attribute the underlying principle to Vilfredo Pareto and describe him as having coined it. The observation about wealth distribution is his; the naming of the principle is generally credited to a later author [FACT CHECK REQUIRED]. The correction is small and worth making, because it is a reminder that a widely repeated framing can carry an error for decades without anyone checking.
Their most valuable contribution is not the proportion at all. It is a distinction.
What Leaders Commonly Misread
The first misreading is that all change is the same problem. Flores and Chase separate two families that most organisations report in one column.
Scope changes are, in their words, "discretionary, intended, and controllable" — additions or deletions to the original work, decided by the client because circumstances or intentions moved. They argue these should not be counted as overruns at all. They are purchases.
Overruns come from a different family entirely: clarifications, errors and omissions, and varying site conditions. These are not decisions anyone made. They are the cost of having defined the work inadequately, arriving later as a bill.
Combining the two in a single variance figure destroys the only information that matters. An initiative that is fifteen per cent over because the enterprise bought more is in a different condition from one that is fifteen per cent over because the definition was poor — and the second is the one that says something about your organisation.
The second misreading is that the front end is where you produce documents. It is where you produce decisions that will be expensive to revisit, and the documents are only the record. Flores and Chase describe the mechanism precisely: low-quality bid documents invite requests for information; those become change orders; and change orders arrive at a premium to original contract pricing. They put a figure on it — that "changes after construction contract award can cost as much as, if not more than, 50 percent of a similar scope of work included in the initial bid documents" [FACT CHECK REQUIRED] — which is asserted rather than demonstrated, and is directionally consistent with everything else in the paper.
The third misreading is that this is an argument for planning longer. It is not, and the distinction matters. The case for planning under uncertainty — that not knowing is a reason to start planning earlier rather than later — is a separate argument made in [Related article: Uncertainty Is the Case for Planning, Not the Excuse Against It]. This article says something narrower: that influence and knowledge are inversely distributed across time, and that the economics of a change depend on when it arrives. Both can be true, and an organisation can respond to the second by defining better rather than by planning longer.
Reframing the Issue
The reframing is to treat front-end work as the purchase of optionality, and to price it as such.
Every hour spent clarifying a requirement early buys the right to change position cheaply. Every hour deferred converts that right into a variation negotiated under time pressure with a counterparty who now holds information you do not. The front end is not a preparatory phase. It is where the organisation buys or forgoes its own future flexibility.
Two consequences follow.
The right question is not "have we planned enough?" but "what have we already committed?" A decision becomes expensive not when it is written down but when others have built on it. Mapping which early decisions are load-bearing — which ones twenty subsequent choices depend on — identifies where definition effort actually pays.
Definition quality is a measurable property, not a feeling. The rate at which a delivery team asks clarifying questions is an available signal, and it is generated continuously from the moment work starts. Most organisations collect it, in ticketing systems and RFI logs, and read it as administrative traffic rather than as a verdict on their own front end.
The one thing this reframing does not settle is who should carry the exposure created by a poorly defined scope. That is a question about commercial structure, treated in [Related article: Risk You Transfer Is Risk You Still Own]. Definition quality determines how much late change a contract will generate; allocation determines who pays for it.
Strategic Analysis
The front end is where scope is decided, and scope decides the rest
The teaching notes behind this material put the point more bluntly than most consultancy would: "If the project scope is flawed then no matter how effective the management processes are the project will not achieve its objectives."
That is worth taking literally. Governance quality, delivery capability and reporting discipline are all downstream of a definition. Excellent execution of a poorly specified thing produces a well-managed disappointment — on time, on budget, and not what was needed. The organisation will then examine its delivery function, which was not the problem.
A pharmaceutical manufacturer scaling a new production line illustrates the shape. The decisions with the longest reach are made in the first months: the process configuration, the containment strategy, the degree of automation, whether the line is designed for one product or several. Each is made when the commercial forecast is least reliable and the regulatory position least settled. By the time both are clear, the steel is ordered.
Nothing in that sequence is avoidable. What is available is a deliberate choice about which of those decisions to hold open, at what carrying cost, and which to fix early in exchange for the ability to proceed. That is a portfolio-of-options judgement, and it is almost never made explicitly.
What starting with incomplete definition actually costs
A vendor-sponsored column in Offshore Magazine in October 2005 — written by the president and cofounder of the software company that initiated the underlying study, and therefore treated here as illustration rather than evidence — asserts that designs are "often 60 to 70% complete upon start of construction". The claim is the author's own, unattributed and unevidenced, and it is offered here only because the shape it describes is recognisable to anyone who has run capital work: construction beginning against a definition that is substantially incomplete, with the remainder resolved through change.
The same column notes that project data was reported to be leaving organisations as the workforce retired — again, an assertion in a sponsored piece rather than a finding.
Both point at the same structural condition. Front-end definition is under-resourced because its cost is visible and its benefit is counterfactual. A month of additional definition appears in the schedule as a month of delay; the three months of variation it would have prevented never appear anywhere, because they did not happen.
That asymmetry is why front-end investment loses every argument it is not deliberately protected from. The remedy is to require the counterfactual to be estimated and recorded, so the comparison is at least made.
Where the gate sits in this
An obvious response is to hold a decision point before commitment and test whether the definition is adequate. That is right, and what such a decision point is for — whether it is a genuine control or a status review with a budget attached — is examined in [Related article: What a Stage Gate Is Actually For]. The argument here is upstream of that one: a gate can only test the quality of definition that exists, and no amount of governance manufactures definition that was never funded.
Decision Framework
Five steps, applicable to any initiative before major commitment.
1. Separate the two change families in your reporting. Report discretionary scope change and definition-driven change as distinct lines. This is a reporting change, not an analytical one, and it usually takes a quarter to implement and immediately alters what the board discusses.
2. Identify the load-bearing early decisions. For the initiative in front of you, list the decisions that twenty later choices will depend on. Five to eight is typical. These are where definition effort pays, and where the option to defer is worth buying.
A caution applies to the schedule those decisions produce. Load-bearing decisions are commitments whose reversibility declines; the orderings around them are assertions that may never have been true — see [Related article: Which of Your Dependencies Are Real?].
3. State what each early decision costs to reverse, by month. Not precisely — by order of magnitude, at three points in the life. The exercise takes an hour and makes the crossing point visible for that specific initiative rather than in the abstract.
4. Measure clarification traffic as a definition metric. Count requests for information, by originating discipline, against elapsed time. Rising traffic in a discipline is a verdict on the front-end work in that discipline, available months before the variations arrive.
5. Require the counterfactual before declining front-end investment. When additional definition work is refused on schedule grounds, record the estimated late-change cost that decision accepts. The estimate will be poor. Making it at all changes the decision more often than anyone expects.
A supporting convention: treat the assumptions the definition rests on as a separate register. Assumptions are beliefs to be tested; the decisions in step two are commitments whose reversibility declines. Confusing the two is common, and the discipline of separating them is set out in [Related article: What Must Be True: The Assumptions Register as a Strategy Instrument].
From Strategy to Execution
Immediate. Take the initiative with the largest committed capital and run step one on its last twelve months of variations. Most organisations find that much of what was reported as overrun was discretionary purchase, and that the remainder concentrates in two or three disciplines. Both findings are actionable within a week.
Medium term. Change the funding rule for front-end work. Definition effort competes badly against delivery effort because its benefit is invisible, so it needs protection: a fixed proportion of an initiative's budget ring-fenced for definition, released before commitment and not available for reallocation. The proportion matters less than the ring-fence.
Long term. Build the organisational memory that front-end quality depends on. The decisions that go wrong early are usually decisions someone in the enterprise has seen go wrong before. Whether that experience is available at the moment of definition is a matter of how the organisation retains and routes what it has learned — and where it is not retained, every initiative pays the tuition again.
Signals to Monitor
- Clarification traffic by discipline. The single most useful leading indicator in this article, and one most organisations already generate without reading.
- Variations arriving in the first third of delivery. Late change is expensive; early change concentrated immediately after commitment indicates a definition that was signed off before it was finished.
- Overrun reported as a single figure. Wherever discretionary and definition-driven change are combined, the organisation cannot tell whether it has a purchasing pattern or a capability problem.
- Front-end effort as a share of total effort. Track it across initiatives. A consistently low proportion, coupled with high late-change cost, is the pattern this article describes.
- Definition work cut when schedules tighten. Watch what is removed first under pressure. Front-end effort is the usual casualty and its removal is the least visible decision in the programme.
- Post-completion reviews attributing failure to execution. Where reviews consistently find delivery at fault on initiatives whose definitions were rushed, the review instrument is looking downstream of the cause.
Questions for the Leadership Team
- On our largest initiative, what share of reported overrun was change we chose to buy, and what share was definition we failed to complete?
- Which five early decisions is that initiative's outcome most dependent on, and how much would each cost to reverse now?
- What proportion of that initiative's budget was spent before commitment, and is that proportion a decision or an accident?
- What is our clarification traffic telling us, and does anyone read it?
- When we last shortened a schedule, what definition work was removed — and did we estimate what that would cost later?
- Do we know the difference, across our portfolio, between initiatives that are expensive because we bought more and initiatives that are expensive because we specified badly?
Closing Perspective
The economics of an initiative are largely settled in a period that most organisations treat as preliminary. Influence is highest, cost of change is lowest, and effort is minimal — a combination that would be recognised as an arbitrage in any other part of the enterprise.
It persists because the argument is structurally unfair. Front-end investment costs visibly and pays counterfactually, and no business case has ever been strengthened by a line describing a variation that did not occur. So the month is cut, the definition ships incomplete, and the difference is paid later at a premium by people who were not party to the decision.
The remedy is not more planning, and it is not better governance downstream. It is to fund definition deliberately, to separate what you bought from what you failed to specify, and to accept that the cheapest decisions your enterprise will ever make are the ones nobody is yet paying attention to.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.