From Estimate to Commitment

A range becomes a point somewhere between the estimator's spreadsheet and the board paper. No one makes that decision. It happens by the deletion of a column.

An agribusiness co-operative is upgrading its grain terminal. The scenario is hypothetical; the sequence is not. At concept stage the number is forty-two million, and the estimator who produced it would tell you, if asked, that it could plausibly land anywhere between thirty and fifty-five. Nobody asks. By the time the paper reaches the board it reads "$42m", and four pages later it reads "the approved budget". Two years on, at $49 million, the conversation is about delivery.

Nothing improper occurred. The estimate was competent, the paper accurate, the board reasonable. What went missing was the range — and with it, the only honest signal in the document about how much anyone actually knew. The deletion is invisible: there is no meeting at which someone proposes converting a range into a commitment. It happens through formatting.

The Strategic Context

Course material on project cost management is more candid about this than most corporate practice. It reports, as received wisdom in the field, that there is no such thing as an accurate estimate — and then sets out what accuracy can reasonably be expected at each stage of definition.

[FACT CHECK REQUIRED — these figures are printed unattributed in the teaching source; reading them as symmetrical bands is an inference from the surrounding text rather than something the source states]

Two things about that table matter more than the numbers. The first is that accuracy is a function of definition, not of effort or seniority. No estimating skill closes a concept-stage range, because the information required does not exist yet. The source says as much: until a detailed design is complete, it is difficult if not impossible to commit to an accurate project cost.

The second is that this is not merely didactic. Winch, Usmani and Edkins, studying a major research-campus construction project in Construction Management and Economics in 1998, record the same structure occurring in practice: an initial estimate at plus or minus 35 per cent accompanying the client's control documents, narrowing to plus or minus 15 per cent as the design problem was resolved. The values differ from the teaching figures; the shape does not. A range that is wide early and narrows as definition improves is how estimating actually behaves.

And the same course material states, in terms, what should follow: expectations must be correctly established at concept stage, and project managers, clients and sponsors must all understand the estimating methods used and the variability associated with them. That instruction is almost universally ignored, and its absence is what this article is about.

What Leaders Commonly Misread

That a single number is the more decisive form of information. It feels decisive. It is less informative, and the decisiveness is borrowed — the reader supplies a confidence the estimator never had. A board told "$42m" and a board told "$42m, order-of-magnitude, plus or minus a third, method: analogous from two comparable terminals" are being asked to make different decisions with different reversibility. Only the second can sensibly ask whether to spend money on definition before committing to construction.

That the range is an admission of weakness. In most organisations it is presented, and received, as though it were. The perverse consequence is that the most rigorous estimator — the one who has thought hardest about what is unknown — produces the least confident-looking paper and loses the argument to a colleague with a rounder number and less analysis.

That "estimate" and "price" are the same act. They are two decisions with different owners, and the source material draws the line explicitly, though it scopes the distinction to contract work: cost estimating develops an assessment of the likely result — what it will cost the organisation to provide the thing. Pricing is a business decision — what the organisation will charge. Extending that distinction to internal approvals is this article's move rather than the source's, and it is worth making, because internally the same two acts occur and only one of them is ever minuted. Someone assesses likely cost. Someone else — often nobody in particular — converts that assessment into a figure the organisation will be held to.

Reframing the Issue

Treat the conversion as what it is: a commercial decision, made by a person, on a date.

An estimate is a technical output. It has a method, a basis, and a variability, and those three properties travel with it or it is not an estimate — it is a number. A commitment is something else entirely: a statement that the organisation will be judged against a figure. Converting the first into the second involves choosing where within the range to sit, deciding what contingency is held and by whom, and accepting a specific probability of exceeding it.

Every one of those is a judgement. In a well-run organisation each is made deliberately and recorded. In most organisations they are made by the act of copying a number into a different document.

Three things this article deliberately does not argue. That uncertainty has an upside as well as a downside — that favourable deviation is information a risk register should be capable of holding — is the argument of [Related article: Risk Is Not the Chance That Things Go Wrong], and it is not re-made here. The narrower observation is that both halves of the band are deleted together, and the organisation loses the good news with the bad. And whether the investment should be made at all, against what alternatives and against the option of doing nothing, belongs to [Related article: What Is the Risk of Not Doing It?]. Nor is it about whether to plan at all under uncertainty, which [Related article: Uncertainty Is the Case for Planning, Not the Excuse Against It] settles. This article assumes the decision to proceed, and asks only how the number accompanying it is qualified.

Strategic Analysis

Method is information, and it is the first thing discarded

The teaching material lists the techniques by which a cost estimate is produced: analogous estimating from similar projects, parametric estimating from quantity multiplied by a historical rate, reliable quotations, vendor bid analysis, determination of resource cost rates. These are not interchangeable. An analogous estimate from two loosely comparable projects and a tendered price from a competitive process are different species of claim, and a governance paper that presents both as "$42m" has destroyed the distinction that mattered.

The consequence is a specific and common governance error: an organisation compares two initiatives whose numbers were produced by methods of wildly different reliability, and treats the comparison as meaningful. One came from a supplier's firm quotation, the other from a planner's recollection of a similar job five years ago. Ranked side by side, they look like peers.

The fix is cheap and unpopular in equal measure: require the method on the face of the number, every time, in every paper that reaches a decision-making forum. It takes eight words, and it is unpopular because it shows how many of the enterprise's numbers rest on analogy.

Contingency is where the conversion becomes traceable

The one place most organisations do reason explicitly about the range is contingency, and the way they do it is revealing. Teaching material on determining a budget names reserve analysis — the setting of contingency sums — alongside funding-limit reconciliation, tying releases to milestones.

Contingency is, in effect, a purchased position within the band. Where it is held tells you who owns the uncertainty. Held inside the delivery team's budget, it will be consumed and the commitment will be met, which is comfortable and teaches the organisation nothing about its own estimating. Held centrally, released against defined events, it stays visible and the enterprise learns how wide its bands actually are. Not held at all, the commitment is the point estimate and the first surprise becomes an overrun.

None of those is wrong in every case. What is wrong is not knowing which one you have chosen, which is the ordinary condition of most enterprises.

A related discipline follows. If contingency is a purchased position in the range, then releasing it should require the same evidence as the original approval — what was learned, what narrowed, what is now known that was not. Contingency drawn down without that evidence is not risk being realised; it is the band being spent quietly.

Where the conversion actually happens

Take a waste-to-energy plant, hypothetically, moving from concept to commitment. There are four candidate moments at which the number hardens, and an organisation should be able to say which one it uses.

The estimator's handover, where a range passes to a sponsor. The business case, where a single figure first enters a formal document. The approval decision, where a governing body records a budget. And the tender award, where the market supplies a firm price — the one point at which, as the teaching material notes, an accurate or fixed price can actually be determined.

In most organisations the hardening happens at the second of these, in a document-preparation step nobody regards as a decision, and is merely ratified at the third. That sequencing is the problem in miniature: the commitment is manufactured by a drafting act and then blessed by a governance act that believes it is approving something already settled.

The corrective is not a new committee. It is to name which moment is the conversion, and to require that whoever performs it does so on the record. [Related article: The Front End Owns the Outcome] argues that the front end determines the cost; this article argues something narrower and adjacent — that the front end also determines how honestly the cost is described, and that honesty is discarded separately and later.

Decision Framework

Two rules, and a test that exposes whether they are being followed.

Rule one — every number carries three things. No figure enters a governance paper without its method, its basis and its range — a $200,000 line and a $200 million programme alike. Where a range genuinely cannot be given, that is itself the disclosure, and the paper says so.

Rule two — the conversion is minuted. Turning a range into a commitment is a decision with a named owner, a date, a stated position within the range, and a stated contingency arrangement. It may occur at any of the four moments above. It may not occur silently.

The reconstruction test. Take three completed initiatives and try to establish, from the record, who converted the estimate into a commitment and on what basis. If the trail runs cold — and it usually does — the enterprise has been making one of its more consequential financial decisions anonymously. That is more persuasive to a board than any argument, because it is a fact about their own organisation rather than an opinion about estimating.

Note what this does not ask for: more precision. Concept-stage precision is unavailable at any price, and demanding it produces false confidence rather than better numbers. It asks only that the imprecision which genuinely exists survives the journey to the decision-maker.

From Strategy to Execution

Immediately. Add two columns to the standard investment paper: estimate method, and range. No process change, no system, no training. The first paper produced under the new template will start a useful argument.

Over the next two or three quarters. Establish where contingency is held and who releases it, and require evidence at release. Then begin recording the conversion — the single line that says who committed the organisation to a figure, when, and where in the range they chose to sit.

Longer term. Build the feedback loop that makes the bands real. Compare original ranges to outturns across a portfolio of completed initiatives, and publish the comparison internally. Most organisations have never done this, and the result is invariably instructive: it shows which estimating methods are reliable in their hands, which sponsors are systematically optimistic, and how wide the concept-stage band really is in their business rather than in a textbook. It replaces an argument about honesty with a table.

Signals to Monitor

• Papers in which every number is a point estimate. The bands are being removed upstream of governance as a matter of routine.
• Contingency drawn down early and without narrative. The range is being spent rather than managed.
• Variance reported against a concept-stage figure as though it were a budget. The organisation is holding people to a number that was never a commitment.
• The most cautious estimator consistently losing internal arguments. Rigour is being punished — a reliable predictor of future optimism.

Questions for the Leadership Team

1. What was the range around the number on our biggest approved initiative when it was first produced, and who removed it?
2. Who, by name, converted each estimate into a commitment — and on what date?
3. Where is contingency held on each, who releases it, and what evidence is required?
4. When we compare two investment options, do we know whether their numbers were produced by comparable methods?
5. What do our last twenty completed initiatives tell us about the real width of our concept-stage estimates?
6. Would a rigorous estimator in this organisation be rewarded for showing a wide range, or quietly penalised?

Closing Perspective

The instinct behind the point estimate is understandable. Executives are paid to decide, decision requires commitment, and a range can look like an evasion of the responsibility to commit.

But the range is not a hedge. It is a measurement of how much the organisation currently knows, and deleting it does not increase knowledge — it only conceals the deficit and transfers the consequences to whoever is holding the work when the deficit surfaces. That transfer is usually described afterwards as a delivery failure, which is both unfair and expensive, because it sends the organisation looking for the problem in the wrong place.

An enterprise that carries method, basis and range on every number it governs by will make slightly more uncomfortable decisions and considerably fewer bad ones. It will also, over time, learn something it currently cannot: how good its own estimates actually are. That is the same discipline this collection applies to a priority label in [Related article: What Does "Critical" Actually Commit You To?] and to the definition of success in [Related article: Who Is Entitled to Say It Worked?] — in each case, a word or a figure that reads as a commitment, and specifies far less than everyone believes.

About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.

Copy link Print