The symptom
Reliability and accuracy are not the same thing, and a team that always hits its numbers is not automatically a team whose numbers reflect what the work genuinely requires. This is a subtle symptom because it hides inside a result that looks, by every conventional standard, like success. A team that never misses a deadline is usually praised for exactly that. The question this symptom raises is not whether the team is reliable — it clearly is — but what that reliability is actually made of, and what it might be costing that no dashboard currently shows.
The pattern to watch for is a widening gap between effort estimated and effort actually required for comparable work, tracked over time, in a team whose delivery record looks stable or improving. A stable delivery record combined with a quietly inflating estimate is a specific and checkable symptom — and one that a delivery-focused view of performance, which mostly asks “did they hit the date,” is not built to surface on its own.
Why it happens
The mechanism here is best understood as asymmetric consequence. An organisation’s response to a missed deadline and its response to an early or comfortable delivery are rarely symmetrical, and that asymmetry teaches people something, whether or not it is intended to. Research on process-improvement and delivery pressure — including Nelson Repenning and John Sterman’s work at MIT Sloan on why improvement efforts default to “working harder” under pressure — points to a related dynamic: people and teams respond rationally to the incentives actually in front of them, which are frequently about avoiding visible failure rather than optimising for the most accurate possible forecast.
Applied to estimation specifically, the pattern is straightforward once named. A team that overruns a deadline typically faces scrutiny, explanation, sometimes escalation. A team that delivers comfortably early rarely faces any equivalent attention — there is no meeting convened to ask why the estimate was too generous. Given that asymmetry, the rational long-run response is to build in enough margin that overruns become rare, even if that margin has nothing to do with the actual complexity of the work. Reliability, in this shape, is not free. It has been purchased with padding, and — this is the part that tends to go unexamined — nobody explicitly decided to make that trade. It accumulated, estimate by estimate, as a reasonable individual response to an environment that punished one kind of error far more visibly than the other.
The competing, more comfortable explanation is that the work has simply become more complex over time, which would fully justify larger estimates without requiring any behavioural explanation at all. This is often partially true — some complexity growth is real and legitimate. The distinguishing question is whether the estimate growth is proportionate to genuinely comparable work, or whether it has outpaced any plausible increase in complexity, which is a checkable, specific question rather than a matter of impression.
What it looks like in practice
Consider an organisation where a project team has consistently delivered on schedule for two years running — a track record that has made the team a reference point for reliability elsewhere in the business. A routine review, prompted by nothing more than curiosity about how estimation practice had evolved, finds that the team’s own estimates for comparable work have grown by roughly forty per cent over that period. The first explanation offered, and the most flattering one, is that the work genuinely has become more complex.
Some of it has. The review also finds a second factor: the team received significant, visible criticism after two schedule overruns early in the two-year period, and received no comment at all — positive or otherwise — on the occasions when it delivered early. The organisation’s response to being wrong was loud; its response to being cautious was silent. Over time, the estimates adjusted to reflect that asymmetry, not merely the complexity of the work. Reliability was, in effect, purchased with margin, and the purchase was never a deliberate decision by anyone — it was the accumulated result of many individually sensible responses to an environment that reacted strongly in one direction and not at all in the other.
The organisational cost of this pattern is easy to miss precisely because it produces a result that looks like success. Padded estimates consume capacity that could be allocated elsewhere; they distort planning and prioritisation decisions made on the assumption that the estimates reflect the work rather than the team’s risk calculus; and they are self-reinforcing, because a track record of hitting padded estimates provides no visible reason for anyone to question whether the padding is there in the first place.
A further cost worth naming is what padded estimation does to comparisons between teams. A team that estimates accurately, without margin, will appear less reliable on paper than a team that pads generously, simply because accurate estimates carry a genuine chance of an occasional overrun that padded ones are specifically designed to avoid. An organisation comparing team reliability without accounting for this difference risks rewarding the padding behaviour directly, reinforcing exactly the asymmetric incentive that produced it in the first place, while an equally capable but more accurately forecasting team is quietly penalised for a kind of honesty the organisation’s own reward pattern does not actually recognise as such.
The same pattern appears across many kinds of organisation; see Other Sectors for how it is read elsewhere.
Questions worth considering
- What does this organisation’s reaction to being wrong actually teach people to do differently next time — and is that lesson the one leadership intended to teach?
- If a team’s estimates have grown steadily, is the growth proportionate to a genuine, checkable increase in the complexity of comparable work?
- Is early or comfortable delivery ever examined with the same attention as a missed deadline — or does only one direction of surprise get noticed?
More on this topic: Performance.
Your team is facing similar challenges?
Subscribe to The Riddle Ledger
Perspectives are sent as they are published, with no promotional email, and you can unsubscribe at any time.