AI Alone Doesn't Create Productivity: The Number That Deflates from 16.8% to 5.1%

There's a number that makes for great slides: companies using artificial intelligence are about 17% more productive than those that don't. It's true, and it's misleading. A 2026 Statistics Canada study did the exercise almost everyone skips: looking at what's left of that edge once you strip out everything that has nothing to do with AI.

The result deserves to be shown as is. The raw advantage is 16.8%. But part of that gap comes from the fact that the companies adopting AI were already more productive before adopting it. Account for that, and the advantage falls to 10.2%. Then the researchers remove the effect of the investments that almost always come with AI: data analytics, R&D, cloud, training. What remains attributable to AI alone: 5.1%.

This isn't an argument against AI. It's an argument against a certain way of buying it. The data's message is plain: the researchers' own conclusion is that "AI adoption alone is probably insufficient." In other words, plugging an AI tool into a company that hasn't organized its data, rethought its ways of working and trained its people yields a small gain. The bulk of the result lives elsewhere, in everything you build around it.

For a small-business owner, that's the best possible news, provided you read it right. It means productivity isn't reserved for those who buy the most expensive technology. It's available to those who do the less glamorous work: clarifying a piece of data, simplifying a process, showing a team how to use it. The tool is the easy part. The result is in the rest.

At Paquin & Co., we help Quebec SMEs build that "rest" (the data, the processes, the people) so AI actually pays off, not just 5%. paquinco.com