Estimating questions look like arithmetic questions, so candidates prepare for them by memorising the PERT formula and stopping there. Then the exam hands them a scenario with no numbers in it at all: a sponsor wants a budget figure by Friday, the charter was signed last week, and the work breakdown structure does not exist yet. Which estimating technique should the project manager use?
That is the question the exam actually asks, and the formula does not help with it. What helps is knowing what each technique needs before it can produce anything. Every estimating question embeds a description of the information available, and the information available decides the technique. Once you read for that, the whole category becomes mechanical.
Start from what the technique requires
There are four techniques worth knowing cold for duration and cost estimating. They differ less in how they calculate than in what they demand as input.
| Technique | Needs | Relative accuracy | Effort | Typical timing |
|---|---|---|---|---|
| Analogous | A genuinely similar past project | Lowest | Low | Early, little detail available |
| Parametric | Historical data plus a scalable unit rate | Middle | Low to moderate | Early to mid planning |
| Bottom-up | A decomposed WBS and resource detail | Highest | Highest | Late planning |
| Three-point | Three estimates per activity from people who know the work | High, and expressed as a range | Moderate to high | Where uncertainty is the problem |
Read that table once more as a decision tree rather than a reference list. If the scenario says detail does not exist, bottom-up is off the table no matter how attractive accuracy sounds. If it says a reliable unit rate exists, parametric is almost certainly the answer. The constraint in the stem is doing the work.
Analogous estimating
Analogous estimating takes the actual cost or duration of a similar past project and adjusts it for the current one. The For Dummies PMP exam material describes it as combining expert judgment with historical information, adjusted for cost drivers such as size, weight, and complexity, and notes that it is typically used in the earlier phases of a project when detailed information is limited.
The qualifier that matters is "similar." The same source is explicit that the technique works only when projects are alike in substance rather than merely in appearance — two office fit-outs of the same square footage are not comparable if one is in a listed building. Exam scenarios exploit this by offering a past project that superficially matches and inviting you to scale from it.
Analogous estimating is the fastest of the four and the least accurate. PM Study Circle, comparing the techniques for PMP candidates, puts its accuracy at roughly −25% to +75%. Treat that figure as the order of magnitude to expect rather than a number to memorise: the point is that analogous estimates are wide, and that is acceptable when the alternative is no estimate at all.
Parametric estimating
Parametric estimating multiplies a quantity by a rate derived from historical data. The Dummies material gives the cleanest illustration: a contractor's rate of $120 per square foot across 4,000 square feet produces $480,000. Project Management Academy, a PMI Authorized Training Partner, works the same logic the other direction — a past 1,000-square-foot project that cost $200,000 implies $300,000 for 1,500 square feet, assuming material costs hold.
Two conditions have to be satisfied, and exam questions turn on both. The parameter must be quantifiable, and the relationship must be scalable. Project Management Academy notes that the simple version assumes a linear relationship between the parameter and the cost or duration, and that missing or non-scalable data undermines the result. Where the relationship is not linear — where the twentieth floor costs more than the second — the simple rate fails and the estimate needs a statistical model or regression analysis instead.
On accuracy, Project Management Academy places parametric estimates considerably above analogous ones but below bottom-up and three-point. That middle position is the reason parametric estimating shows up so often as the correct answer: it is the technique that buys meaningful accuracy without requiring detail the project does not yet have.
Bottom-up estimating
Bottom-up estimating works from the decomposed work. The Dummies material lays out the sequence: determine the cost of each resource for each deliverable, sum the deliverables, add contingency reserve, and aggregate to a total.
It is the most accurate technique available, and PM Study Circle puts the range at roughly −5% to +10%. It is also the most expensive, in both effort and elapsed time, and it cannot be started until deliverables are defined in significant detail. The Dummies material flags two failure modes worth remembering: interaction costs between work packages get overlooked, and contingency applied at every individual task inflates the total.
The exam cares about the precondition. "Bottom-up" is a plausible-sounding distractor in any scenario where scope has not been decomposed, and it is wrong there for a structural reason, not a judgment one. There is nothing to add up yet.
Three-point estimating
Three-point estimating collects an optimistic, most likely, and pessimistic value for an activity and combines them. The weighted form, variously called PERT or the beta distribution, is:
Expected value = (O + 4M + P) / 6
The most likely value carries four times the weight because it reflects the probable outcome. PM Study Circle's commute example makes the arithmetic concrete: with O = 30, M = 60, and P = 120 minutes, the expected duration is (30 + 240 + 120) / 6 = 65 minutes. The simple average of the same three numbers is 70 — the triangular distribution, which the Dummies material describes as the less accurate of the two forms.
The reason to reach for three-point estimating is uncertainty rather than precision. PM Study Circle positions it where activities are unique, where new technology or complex dependencies are involved, and where subject-matter experts can supply realistic values — and notes it is less necessary for repetitive work with ample historical data, where analogous or parametric estimating will do. Its output is a range, which is what justifies a contingency reserve.
Reading the clue in the stem
Most estimating questions can be resolved from a single phrase:
- "Limited information," "high-level," "the charter was just approved" — analogous
- "A reliable rate per unit," "cost per square metre," "historical productivity data" — parametric
- "The WBS is complete," "the team has detailed the work packages," "the most accurate estimate possible" — bottom-up
- "Significant uncertainty," "the team disagrees on duration," "a range is needed for the reserve" — three-point
When two techniques both look available, the constraint in the question decides. A demand for an estimate by tomorrow rules out bottom-up. A demand for the most defensible number, with time to produce it, rules out analogous.
Relative estimating in adaptive work
In adaptive and hybrid delivery the team usually sizes work relative to other work rather than in absolute hours. Techniques such as planning poker, t-shirt sizing, affinity mapping, and bucket systems all compare items against each other and converge through discussion, typically using a Fibonacci-style scale. The underlying principle, as PM Study Circle puts it, is to focus on relative sizing rather than absolute numbers.
For exam purposes the important move is recognising which world the question is in. If the scenario describes a team, a backlog, and iterative delivery, an answer that sends the project manager off to build a parametric model alone is the wrong shape — the team estimates its own work. If the scenario describes a sponsor needing a budget figure for a predictive project, the relative techniques are not the answer.
How to practise this
Take twenty estimating questions and, before looking at the options, write down one line: what information does this scenario say is available? Then pick the technique that information permits. You will find the correct answer falls out of that line far more often than it falls out of the arithmetic — and on the handful of questions that do want a calculation, it is almost always the weighted three-point formula.