Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Why the last 10% matters

As AI compresses the first 90% of any project into hours, leaders are discovering that the final, slowest stretch is where real judgment—and the real work—still lives.

[Photo: Getty Images]

Profile photo of Kurt Donnell

CEO · Freestar

Every team using AI has felt this by now. You get to 90% done in record time. A draft, or a prototype, or an analysis that used to take weeks now comes together in a few days or even hours. Then everything hits a snail’s pace, and inevitably the team is wondering: How are we not done yet?

The honest answer is summarized beautifully by productivity writer James Clear: “When you are creating something, you will usually find once things appear 90% done, you are actually about halfway there.”

The last 10% of a project has always been the hard and slow part, and AI is magnifying it.

The new math

The old math: it will take 50 hours to get a project to 90%, then another 50 hours of refinement to go from good to great. The second 50 hours never bothered anyone because the first stretch was just as long.

The new math: AI delivers the first 90% in 10 hours. The refinement still takes 50 hours. Same work, same value, but now the finish is five times longer than everything that came before it. It feels like stalling, but it isn’t.

The 50-hour finish has long been the case, even before AI. I’ve experienced this with almost every system migration I have been part of over the last 20 years, including a recent CRM accounting system migration. It will feel like we’re nearly done, but nailing the final details drags on, and it can take months for things to be fully operational.

What has changed is the temptation to reach the finish line in a similarly compressed timeframe. When the first 90% comes so easily, spending extra time on the rest now feels painful.

Almost done is the most dangerous moment

I fall into this trap myself. Not long ago, my team spent days on a major RFP, and when I went to bed one night, I was certain we were a few finishing touches away from being done.

The next morning, as I read it with fresh eyes, I immediately realized we were not nearly ready to ship it. We spent another dozen-plus hours on it over the next two days. In the moment, I was annoyed. In hindsight, that stretch was the difference between a submission that was good and one that was a contender to win.

That’s the reframe leaders need to make, for themselves and for their teams. The last 10% is not the boring part that comes after the real work. It is where the real work truly kicks into gear.

This is when leaders should get in the dirt

There’s a paradox here for executives, and I’ll own it. I spend a lot of time preaching delegation: Focus on what you should do, not everything you can do. AI has made delegating the first 90% easier than ever—and you should take advantage of that.

But the last 10% is when your involvement as a leader is most valuable. This is where you catch the edge cases, pressure-test the assumptions, and find the details that make something memorable instead of forgettable. AI can’t shortcut the work that takes experience and judgment.

The trick is timing. You should spend time with the work at the 90% mark. Don’t wait until it’s fully baked. By then, your input means undoing things that were built in that final stretch instead of helping to shape them. Come in at 90%, and you’re the fresh eyes, showing up with questions the team may not have yet asked because they are so close to the product. What happens if this breaks? What are the legal implications if something goes wrong? What if the system is offline for six hours, and what is the fail-safe? How does this touch billing, data reporting, and our pricing model? My favorite: What are three reasons customers would say no to this, and how do we build the answers in before anyone asks?

These types of questions come from the experience of being in the game longer. When you have decades of experience dealing with challenges, you see them coming sooner. You can see what a team that has been heads-down for weeks can’t, and you can widen the view before everyone locks into one way of thinking. As a leader, you don’t necessarily need to weigh in on the technicalities; you need to bring the perspective.

That perspective cuts both ways. Sometimes it highlights what’s missing. Sometimes it’s the restraint to slow down. During our recent quarterly planning meeting, three people on my team each built a prototype to solve the same problem, literally as the meeting was taking place. Were the prototypes close to what we needed? Yes. Were any of them close to the final product? No.

My reflex was exactly what you would expect: set up a meeting on Tuesday and deliver something by Friday. It feels like that should be possible now. But I had to restrain myself from setting that expectation because it would not be done well if we rush it. The whole world has accelerated, and the leadership discipline that matters now is knowing when not to pass the pressure to accelerate onto your team. It is okay, and in fact important, to take longer to get it right and to run it by people who don’t have an immediate stake in the outcome.

This is the dirt. And the dirt is where the magic happens.

Build your culture for the slow finish

The slow finish isn’t just a mindset at Freestar. We are deliberately realistic about timelines because the final stretch always takes longer than expected. We build review gates into projects, so that fresh eyes are a requirement, not a last-minute scramble. And we budget refinement time into planning, which means the last 10% is a line item.

If you lead a team, the specifics matter less than the signal. Pull business owners and stakeholders into projects before they’re done because the cheapest time to catch a problem is while there is still time to fix it. Reward people for doing thorough work, not just for shipping fast. And when a project feels like it is dragging at the end, your team is watching whether you      practice what you preach about spending the extra time to get it right.

I used to think the biggest risk in business was moving too slow. I’ve learned the opposite risk is just as real: being done too soon. AI has handed every team the ability to have a fast start.

What you do with the slow finish is what will separate you. The last 10% is where your product is forgotten or where you win.

Kurt Donnell is CEO of Freestar.