Estimating Costs: Accuracy Is Not the Same as Precision
A capital figure arrives at a board meeting quoted to the nearest pound. It has been built up properly, it adds correctly, and at concept stage it carries an uncertainty of perhaps forty per cent in either direction. The arithmetic is exact and the number is a guess in a suit, and the exactness is what causes the damage, because people plan against figures that look definite.
Section 2.4.2 of the PMBOK® Guide Eighth Edition covers developing cost estimates, and the distinction worth holding is between two qualities that sound similar and behave differently.
Two different qualities
Precision is how finely a number is expressed. To the pound, to two decimal places, plus or minus two per cent. It is entirely under the estimator's control and it costs nothing to produce, which is exactly why it is such an unreliable signal.
Accuracy is how close the number is to what will actually happen. It comes from the quality of the definition underneath the estimate, the relevance of the data it was built from, and the honesty of the assumptions. It is largely outside the estimator's control early on, because the work has not been defined well enough for anybody to be accurate about it.
Why single numbers keep winning
They are easier to act on. A range cannot be put in a budget line, and somebody has to write something in the cell. This is a real problem and the honest response is to record the point figure used for planning alongside the range it came from, rather than pretending the range never existed.
Ranges get read as their low end. Give an audience a figure between four and six million and a proportion of the room will plan for four, particularly the people who wanted the project approved. Stating which figure should be used for planning, and why, prevents most of that.
Asking for a range sounds like evasion. In organisations where confidence is rewarded, an estimator who offers a range can look less competent than one who offers a number, whatever the evidence says. This is cultural and it is best addressed by making ranges normal for everybody rather than by individual courage.
A concept figure that was believed
A pharmaceutical manufacturer was planning a new packaging suite. At concept, with a layout sketch and a process description, the project engineer produced an estimate of a little over four million pounds, presented as a single figure to the nearest ten thousand because that was the format of the capital template.
The figure went into the annual capital plan and was communicated to three other projects whose own funding depended on what remained. By the time the design was developed, the estimate had risen by about twenty-eight per cent, which was entirely normal for the movement between concept and detailed design and would have surprised nobody if a range had ever been attached.
It surprised everybody. Two of the other projects had committed against the remaining capital, one of them contractually, and the recovery took a supplementary paper to the capital committee and a deferral that cost a quarter of a year on an unrelated piece of work.
The correction was a change to the template, which had never had a field for uncertainty. Capital estimates now carry a range with the class of estimate stated, the basis attached, and the events that will narrow it: layout freeze, supplier quotations, and design completion. The point figure still goes in the plan, because a plan needs a number, and it now travels with the band around it.
The validation office where the estimate was written has a micrometer on the bench, and it made the same point in miniature every day: a hacksawn bar measured to a hundredth of a millimetre, an exact reading of a rough edge.
How to express an estimate
Range, basis, and what narrows it. Three short lines. The range in round figures, what the estimate was built from, and the specific events after which a tighter figure will be available. This turns an estimate into something a decision-maker can use and prevents it being quoted back as a commitment.
Round to the honesty of the input. An estimate with forty per cent uncertainty should be expressed in round millions, and one built from firm quotations can carry more digits. Rounding is a communication act: it tells the reader how much to trust the number without requiring them to read the basis.
For a PMP® candidate, what helps is seeing that precision and accuracy are different properties, so a scenario about an early estimate treated as a commitment is describing a presentation problem before it is a costing one. A response that refines the arithmetic adds precision to something whose definition has not changed. Estimating situations where the pressure is for a single confident figure come up early and often in a structured PMP exam preparation course.
Take the headline number on your own project and ask what range it sat in when it was first produced, and whether anybody outside the project has ever seen that range. If the answer is no, the organisation is planning against a figure whose uncertainty is invisible to it, and adding one line to the next report begins to fix that.
Interested in going further?
An early estimate becomes a commitment the moment somebody writes it in a plan, and how it is expressed decides how much damage that does. Omega's PMP® Exam Preparation works through estimating and how estimates are communicated.
Cost estimating and its interpretation are covered in the PMBOK® Guide Eighth Edition.
Ad · Amazon affiliate link.
References
A123: The PMBOK 8 Finance Performance Domain: What It Really Covers
A124: Cost Is Not the Same as Value
A125: Budget vs Cost Baseline vs Funding
A126: Contingency Reserve vs Management Reserve
A127: CPI and SPI Explained Without Formula Memorisation
PMP and PMBOK are registered marks of the Project Management Institute, Inc.
