Most data failures start before the analysis

When was the last time an analysis you built actually changed someone's mind?

Week one of teaching Data-Informed Organisations at Hyper Island, and the thing I spent two sessions drilling into a new cohort has nothing to do with data. Most data-informed failures are not data failures. Almost nobody has bad numbers. What's missing shows up in how the request gets made, before any analysis starts.

First tell: ask what someone wants, and you get a request. “A dashboard.” “Better visibility on our numbers.” A request names a thing to build. It never names what should be different afterwards. You can deliver it perfectly and change nothing, because nobody committed to an outcome, only a deliverable.

Second tell, the same failure from the other side: ask what changes depending on how the analysis comes back, and you get a shrug. If nobody can say what they'd do differently either way, the analysis was never going to drive a decision. It was going to decorate one that had already been made.

Both are the same failure in miniature. Data-driven treats the analysis as the answer. Data-informed treats it as one input to a decision a human still has to own.

We worked through one example. A consultant gets an email asking for a delivery-times dashboard. She spends thirty minutes asking three questions before opening her laptop, and what she finds underneath is a budget-holder who never wrote the email, an unresolved conflict between two people who both act like the decision is theirs, and a request that would have taken two weeks to build and answered nothing. Watching a room of new analysts find that same gap in a stranger's inbox, and then go looking for it in their own, was worth more than either framework alone. Neither session mentioned a single chart.

Try this on whatever you're working on right now: if the answer came back the opposite of what you expect, what would you actually do differently? If nothing comes to mind, that's the real finding, and more useful than whatever the analysis says.