You're Measuring AI Wrong.

A company deploys AI to summarize meetings. It saves their team about thirty minutes each per week. Real savings—maybe $50,000 a year across the organization.

They're proud of it. The vendor makes a case study out of it.

Meanwhile, their revenue base is $30 million. A 1% improvement in how they price their core product would generate $300,000. Six times the savings. Every year.

Very few companies are working on that.

Most AI deployments start with what is easy to measure—time saved, emails drafted, documents searched—rather than what is important to measure.

The question most companies ask is: “Where can AI save us time?”

The question that actually matters is: “Where does a small improvement generate an outsized return?”

In most businesses, that answer lives at the top of the value chain—pricing, market timing, capital allocation, contract terms. Places where a 1% improvement doesn't save you an hour. It makes you $300,000.

Almost nobody is pointing AI there. Because it's harder to measure. Because it's easier to demo a meeting summary than a pricing model. Because activity feels like progress.

Peter Drucker said it decades ago: “There is nothing so useless as doing efficiently that which should not be done at all.” AI just makes it possible to do the wrong thing faster.

Before your next AI investment, do the math on one question:

If AI made your most important decision 1% better, what would that be worth?

If the answer dwarfs what you're currently saving on meeting summaries and email drafts, you're measuring the wrong thing.

The companies that win with AI won't be the ones that adopted it fastest. They'll be the ones that aimed it at the right thing.