For years I was an estimation sceptic. I saw it as a relic of the waterfall days, along with Gantt charts and process-heavy project managers. Agile had saved us from story points and man-hour breakdowns. Why play planning poker when the future is unpredictable anyway?
Estimation comes with baggage. The three-hour T-shirt size debates. The manager who turns story points into a stick. The agile transformation that becomes velocity theatre. It’s tempting to swing the other way and say estimation is broken, so let’s just build. I’ve been there.
But businesses don’t run on faith. Budgets exist, customers have deadlines, and the rest of the company can’t wait for engineering to find enlightenment. Without any signal at all, trust erodes. Estimation earns its place as a way of staying aligned.
What changed my mind was watching a promising project implode. We’d refused to estimate, saying it would be ready when it was ready. Six months in, the stakeholders were in the dark, the dependencies had drifted, and the trust was gone. The failure modes are easy to name: using estimates for control, treating an uncertain number as a commitment, and letting metrics replace conversations. That’s when estimates turn into weapons rather than bridges.
Done well, estimation is quiet and useful. It’s an early warning rather than a contract. It gives the business something to work with, lets customers plan, and stops engineering looking like a black box. You’re not aiming for accurate numbers. You’re aiming to reduce surprise.
Here’s what good looks like. A team I worked with broke a complex integration into rough T-shirt sizes and found three critical dependencies they’d missed. They started talking to security early. That’s the value, and it’s de-risking rather than fortune-telling. Estimates surface what might explode and what needs more thought. Once you admit the uncertainty up front, you can prototype and slice the work smaller before the risks turn into blockers.
Estimates do get weaponised. I’ve seen managers turn story points into productivity metrics. That’s bad leadership rather than bad estimation, and the answer is to use the tool properly instead of abandoning it. The best teams treat estimation as a dialogue. They’re honest about what they don’t know and quick to change the number when they learn something.
Estimation isn’t about control. It’s about partnership.
Published on . Co-authored with AI.