According to Forrester, between 60% and 73% of the data a company holds is never used for any analysis at all. The figure says it all: most organizations aren’t short of data, they’re buried under more of it than they can read. We collect, we store, we stack up dashboards, and when the time comes to decide, intuition takes over as if nothing had ever been measured. The real subject of data analysis, then, isn’t having more of it, but knowing which piece actually serves a decision.
The Problem Is Almost Never a Lack of Data
This mountain of unused data has a name: « dark data ». It piles up because measuring everything has become trivial—every click, every sale, every interaction leaves a trace. But collecting has never enlightened anyone. One more dashboard doesn’t produce one more decision; more often than not, it produces one more meeting, where people comment on curves without doing anything with them. The capacity to measure has exploded while the capacity to decide on that basis has stayed flat, and the second is the one that actually sorts companies out.
The usual reflex only makes the imbalance worse. Faced with a hard decision, people call for « more data », an extra dashboard, one more study—an elegant way of putting off the moment of choosing. Yet past a certain threshold, additional data no longer raises the quality of the decision; it raises the noise and the delay. Knowing when to stop measuring and start deciding is a skill as rare as it is valuable.
Data Only Answers the Question You Put to It
The most widespread methodological mistake is to start from the available data to see « what it has to say ». The right move is the opposite: start from the decision to be made, then look for the data that informs it. Before opening a single table, the right question is « which decision do I have to settle, and what, in the numbers, would tip it one way or the other ». An indicator that changes no decision, whatever its value, ends up as little more than decoration on the dashboard.
Take that reversal into practice. A company weighing a market expansion doesn’t begin by contemplating its traffic statistics; it first defines the demand threshold that would make the move viable, then goes to see whether the data confirms it. Likewise, knowing where the real growth potential sits means having chosen which question to dig into—retention? margin per customer?—before going to rummage through the numbers. Data never tells you what to do; it decides between options you’ve framed beforehand.
This discipline carries a decisive advantage for an SME: it spares you from building a data factory before getting anything out of it. You don’t need a team of data scientists or a data warehouse to decide whether a diversification holds up—you just need to have framed the right question and to go find the few numbers that answer it. Three relevant data points beat a data lake with no question. Starting with the rare decisions that genuinely matter is also the most realistic way to install a real data culture, without drowning your budget or your teams in it.
The Trap: Data That Confirms What You Wanted to Believe
Once the decision is framed, a second trap waits, more insidious: using data not to decide, but to justify a decision already made. You hunt for the figure that validates the boss’s intuition, isolate the curve that suits, and conveniently forget the one that doesn’t. That’s the opposite of an informed approach—data you never gave a chance to contradict you proves nothing. The honest test of any analysis comes down to a single question: what, in these numbers, would make me drop what I was about to do?
This is also what separates a useful measure from a flattering one. The number of views, followers or visitors climbs steadily and feels good, but steers no action: these are vanity metrics. Data has value when it is actionable—when crossing a threshold triggers a specific move. A handful of indicators tied to real decisions does more for a team than a wall of screens it watches without ever acting on them. The real test of « data-driven » maturity is the number of decisions a dashboard has actually changed, not the number of dashboards on the wall.
Conclusion
Data analysis doesn’t replace judgment so much as it forces judgment to justify itself. Its value comes neither from the volume collected nor from the number of tables, but from a simple sequence that’s too rarely respected: start from the decision, choose the data that sheds light on it, and accept that it might contradict you. A company that follows that order gets more out of a handful of well-chosen indicators than its rival drowning in « dark data ». The question to ask in front of any dashboard isn’t « what do these numbers say », but: which decision are these numbers supposed to help me settle?
If you’re piling up data without being sure it serves your decisions, get in touch—clarifying the decision before digging into the numbers is often what turns a dashboard into a tool.
