Most Businesses Are Paying for AI They Don't Actually Use

The tool isn't the problem. It never was.

I talk to business owners every week who are paying for 3, 4, sometimes 6 AI tools. When I ask what results they’re getting, the answer is almost always some version of “well, we use ChatGPT for emails sometimes.”

That’s not adoption. That’s a subscription.

And apparently it’s happening everywhere. New data from Inc. and Pertama Partners puts the AI project failure rate at over 80%. MIT found that 95% of generative AI pilots get abandoned before they ever become part of the actual business. Not because the tech broke. Because nobody knew how to make it stick.

I’m not surprised. I’ve watched it happen in real time for three years.

Here’s what it actually looks like

Someone hears about AI, gets excited, signs up for a handful of tools. Maybe watches some YouTube videos. Builds a prompt or two. Gets a decent output once and thinks “okay cool, this works.”

Then a week goes by. The tool sits there. Nobody built it into a repeatable process. Nobody measured whether it saved time or made money. It just... exists in the stack, billing monthly.

Multiply that by a few tools and suddenly you’re spending $300 to $500 a month on AI that produces essentially nothing. I’ve seen this exact pattern dozens of times. It’s not a motivation problem. It’s a “nobody showed me what to do after the free trial” problem.

The stat that explains everything

75% of executives admit their company’s AI strategy is “more for show” than actual internal guidance. Three quarters. And these are people with teams, budgets, and consultants. If they’re winging it, imagine what it looks like for a 4-person company with no dedicated ops person.

The wild part: only 29% of companies report meaningful ROI from generative AI. That means the majority of businesses paying for these tools are getting... vibes. A sense that they’re “doing the AI thing.” But no actual business outcome they can point to.

Why small businesses have an unfair advantage here

Big companies fail at this because of politics, committees, and employees who use the tools at surface level out of anxiety about their jobs. You don’t have that baggage.

A small team can go from “this task is eating 6 hours a week” to “AI handles 80% of it” in a single afternoon, if someone who’s done it before sets it up. No change management. No committee approval. No six-month pilot program. Just: identify the bottleneck, configure the solution, train the person, measure the result.

The operational advantage of being small is massive right now. But only if you skip the “wander around trying tools” phase and go straight to “someone who knows what works tells me exactly what to do.”

The market is splitting and it’s obvious

There are businesses using AI as a line item that produces results every week. And there are businesses using AI as a vague subscription they feel good about having. The gap between those two groups is getting wider, not smaller, because the first group compounds their advantage every month.

The difference isn’t intelligence. It isn’t budget. It’s whether someone in the operation has enough reps to know which tool goes where, and more importantly, which tools to ignore entirely.

That’s my entire job. Not teaching AI theory. Not running workshops. Getting it producing results in your specific business, this week. If your AI tools are collecting dust, hit reply and tell me what’s not working.

See you tomorrow,

Andrew