Why Your Estimates Keep Missing — And the Habits That Fix Them
Ask a room of project managers why their last estimate was wrong and you will rarely hear "we did the math badly." You will hear that scope crept, a supplier slipped, a key person was off sick, or the number was set before anyone understood the work. Estimating is not a calculation problem so much as a discipline problem. The teams that estimate well are not blessed with better intuition — they have simply stopped making the same handful of mistakes.
In the spring of 2021, that lesson is sharper than usual. Hybrid teams are harder to read across a screen, materials and equipment lead times are still unpredictable after a year of disruption, and the easy assumption that "things will be back to normal by then" has burned a lot of plans. A good estimate now has to carry uncertainty openly rather than hide it inside a single confident number.
The mistakes that wreck estimates
Estimating before you understand the work. A number produced to satisfy a deadline question — "roughly how much, roughly when?" — has a way of hardening into a commitment. Separate the rough order-of-magnitude from the figure you are willing to be held to, and say which one you are giving.
Treating one number as the truth. Reality is a range. "Eight weeks" pretends to a precision you do not have; "seven to eleven weeks, most likely nine" tells the reader where the risk lives and invites a real conversation about it.
Anchoring on the first figure said aloud. Once a sponsor floats "this should be about fifty thousand," every later estimate drifts toward it. Gather independent inputs before anyone names a number, not after.
Forgetting the work that is not the work. Testing, rework, reviews, coordination, handover, and waiting on other people routinely add up to more than the build itself. Estimates that count only the obvious task are optimistic by design.
Padding in secret instead of pricing risk in the open. A quietly inflated number erodes trust and gets cut by a sponsor who assumes you padded. A visible contingency tied to named risks survives scrutiny.
Techniques that beat guessing
None of these is exotic, and none requires a tool you do not already have. They work because they replace a single person's optimism with structure and evidence.
Analogous estimating — start from a comparable past project, then adjust for what is genuinely different. Honest history beats a blank page every time.
Three-point estimating — capture optimistic, most likely, and pessimistic figures and combine them, so the number carries its own uncertainty instead of pretending it away.
Bottom-up estimating — decompose the work to a level you can actually reason about, estimate the pieces, and roll them up; the act of breaking it down surfaces the tasks you would otherwise forget.
Wideband estimating — have several knowledgeable people estimate independently, then discuss the gaps; the disagreement is where the hidden assumptions are buried.
Make the estimate a living thing
An estimate is a forecast, not a promise carved in stone. Record the assumptions you made and the date you made them. When an assumption breaks — the supplier confirms a longer lead time, the scope grows by a feature — re-estimate openly and tell the people who are counting on the number. A team that revises early and explains why keeps its credibility; a team that clings to a stale figure until it fails publicly loses it. Track your estimates against actuals on a few projects and you will quickly learn where your own optimism reliably lives, which is the single most useful calibration you can do.
If your organization needs estimates that hold up to scrutiny and forecasts that stay honest as conditions change, XNM's program & project delivery advisory can help you build the discipline into how your teams plan.