When I was on the Gold Coast for AREC, I shipped a pull request and released it from my phone while sitting on the G. I was, of course, using an AI coding agent to help me get there.

Fast forward to a little over a week ago, someone asked me the obvious question: "Great, but did it make you any money?"

No. It didn't.

His point was about the cost of AI relative to its benefits. My point is that the experience still mattered.

A tension exists with speed: Shipping something quickly doesn't prove the thing was worth shipping. Demo velocity can easily turn into theatre: a stream of things going out the door that look like progress but don't move the needle on anything that matters. We've all seen teams optimise for that feeling of motion and mistake it for value.

None of this makes speed irrelevant. The habit of shipping quickly is still worth building, even when individual things you ship don't move a number. If a team has built the muscle to move fast consistently, then as genuinely valuable work shows up in the roadmap, it moves fast too.

Speed as a general capability compounds. Speed as a one-off flex doesn't.

The Paradox of Measurement

A harder problem underneath all of this is measurement. Engineering and product teams lean on operational metrics because the real business outcomes - like revenue, retention or adoption - can take months to show up. So we track cycle time, throughput, escape rates, and how much critical maintenance gets cleared, because those numbers are available this week. Bookings, revenue, and retention are available this quarter, or this year.

That's a reasonable trade-off, but it comes with a catch. High velocity on the operational side doesn't prove the team is creating business value. It only proves the team is executing the roadmap it's been handed, quickly. If that roadmap is focused on the wrong things, all you've built is a very fast way of counting motion instead of progress.

The way I think about it now: manage flow frequently, using the leading indicators, because you need a live read on whether work is moving. Validate value over a longer period, using the lagging indicators, because that's the only honest test of whether the work mattered.

Leading indicators guide us. Lagging indicators judge us.

Don't let the first one convince you the second one is already settled.