Data-informed, before I had a word for it

I didn't start in product. I started in digital marketing, advising clients on campaigns and websites, and the move into product happened almost by accident, through conversion.

In 2018 I joined MatHem as Conversion Manager. The job was to make the customer journey work better, and the main tool was experimentation: form a hypothesis, test it, look at what customers actually did, decide. I ran a lot of tests in those years, and most of them taught me more about our own assumptions than about our customers.

One test I still think about was a change someone very senior was certain would work. It made complete sense in the meeting, and nobody really wanted to argue. We tested it anyway, and customers did the opposite of what we all expected. What stayed with me wasn't that the idea was wrong. It was how much easier the next conversation became, because we were talking about what customers had done, not whose opinion carried the most weight.

That's where being data-informed started for me, long before I had a word for it. The tests that changed something were the ones where we had agreed beforehand what we would do with each possible result. The rest mostly ended up as slides.

I carried that through e-commerce at Ahlsell, lead generation at Universum and subscription growth at Nextory. Very different products, same question underneath: which change is actually worth making, for the customer and for the business?

AI changes the economics of that work. Trying something is now much cheaper, both to build and to analyse. That makes the old discipline matter more, not less, because someone still has to decide what you're trying to learn and what you'll do with the answer. Teams that already had that habit are getting a lot out of AI. In teams that didn't, I mostly see the number of experiments and dashboards go up.

That thread runs through everything I do now, whether I'm helping an organisation work out where AI can create business value or coaching a team on how they make product decisions.