Data earned its reputation. Opinion with nothing behind it is how bad decisions get made in confident voices, and the first thing we do on any job is go and get the facts. Nothing in this note argues for less of that.
But somewhere along the way "data-driven" slid from meaning informed by the evidence to meaning the numbers decide. Those are different things, and the second one fails in predictable ways.
Data is a record of the past, gathered by somebody, for a purpose. It answers the questions somebody thought to ask, about the things somebody chose to measure, inside lines somebody drew. Three failures follow from that:
The way we work: data is the instrument panel, judgement is the pilot. A pilot does not ignore the altimeter because she feels lucky. But nobody asks the altimeter where to fly. The two questions belong in the same breath. What does the evidence say? And what can this evidence not see?
In practice, we quantify everything that can honestly be quantified, say how much confidence each number deserves, and then say out loud which parts of the recommendation rest on judgement and experience instead. A client can weigh that. What a client cannot weigh is a number with the judgement smuggled inside it.
Use data hard where it can answer the question. Say so plainly where it cannot. Most of the job is knowing which is which.
We produce market assessments, vendor due diligence and feasibility studies for investors, companies and governments.