Alex Steer

272 words | ~1 min

Data is good at saying no.

And those of us who work in analytics tend to think of ourselves as the checkers and balancers of ideas. Given a hypothesis, we go and find all its rough edges, all the reasons why it might not work.

That critical faculty is important and powerful. But after a while people get tired of hearing no, and they assume that it will be the answer to any question they ask us, so they stop asking. We risk becoming the footnotes to a poem - maybe important, but distracting, so unread.

This is bad for us, and even worse for the work we're trying to make better. Nobody likes a prophet of doom - so we have to earn the right to be heard when it's important.

Saying yes with data is more difficult. It means thinking about the spirit of every question, not just the letter. If someone asks 'will X work?', and the answer is no, the next question should be, what would?

At our best, we should be using data to prompt ideas, not just test them. That, bluntly, means working harder - exploring the data we have, not just to answer a specific question, but to generate new ones, which identify problems, which spark ideas. There's no quick way to do this, any more than there's a quick way for an art director to know what visual cues will make a brand feel 'contemporary' and 'authentic'. That kind of fluency requires spending a lot of time with our craft. It means being more than the people who know how the database works.