The Good-Student Paradox: Companies That Win with AI Were Already Winning

There's a classic trap in reading statistics, and AI is a perfect victim of it. When we observe that companies using AI are more productive, we naturally conclude that AI made them productive. The 2026 Statistics Canada study shows that a good part of the relationship runs the other way. The raw 16.8% advantage falls to 10.2% once you account for the fact that these companies were already, before AI, more productive than the rest.

In other words, part of what we attribute to AI is really a portrait of the companies that adopt it. They're often better-managed organizations, more curious, more disciplined in their operations. AI didn't turn them into good companies; they were good companies that added AI. The good student stays a good student, with or without a new tool.

You could draw a discouraging conclusion from this: if success depends on being good already, why start? That would be the wrong lesson. The right one is that the qualities that make an AI adoption succeed aren't technical and can't be bought. They're management habits: knowing what you're talking about, measuring, finishing what you start. Any SME can cultivate those habits, regardless of size or technology budget.

For an owner, it shifts the question. It's no longer "which AI tool to buy," but "does my company have the habits that make any tool, whatever it is, produce something." That's a less flattering and far more useful question. Answering it honestly, before spending a dollar on technology, is probably the best return available.

At Paquin & Co., we help Quebec SMEs install those management habits, that's what makes an AI project succeed, whatever the technology. paquinco.com