Graham McNicoll

I spoke to an engineering lead last week whose team still tracks progress in story points. In a world where AI is doing most of the execution, what does a story point even measure anymore? Honestly, I am not sure they ever measured much to begin with. Story points are a proxy metric. They stand in for something you actually care about (or you should care about). Delivering things of value, ensuring they are released in a timely and predictable way, and ultimately making progress against the roadmap. Those things are harder to see and measure directly, so we invented a number to approximate them- but missed completely and ended up only measuring perceived "effort". Effort does not make a good product; it doesn't guarantee a product will be valuable, nor does it mean your team is working efficiently or effectively. Yet so many product teams still use story points as their primary metric, often unaware that it is a made-up proxy. These kinds of proxy metrics have a known failure mode too. It even has its own law: Goodhart's Law. When a proxy metric becomes the target, it stops being a useful measure of anything. The cleanest version of this story is Hanoi under French occupation. The administration had a rat problem, so they offered a bounty for every rat tail turned in. After a while, people started noticing tailless rats running around the city. Locals were cutting the tails off rats and releasing them back into the breeding stock. Some were farming rats outright. The bounty got paid. The rat population got worse. Story points follow the same pattern. The moment a team is measured on them, they inflate estimates, pad the unit, and protect the velocity number. A PM pushes for more points next sprint. The team just adjusts what a point is worth. That is a team learning to play the game they are being scored on. It is not a team getting better at their job. The real question for an engineering org is different. What did you learn this sprint, and what changed because of it? A one-day build that disproves a bad hypothesis is worth more than a six-week feature that ships to no effect. That is what a learning rate looks like. If you are still tracking your team on story points in 2026, I am curious what value you get from them.