When I joined the New York Times to run data and insights for the digital analytics team, my boss told me something on my first week.
Don't trust the numbers unless you trust the person.
I had come from a different world. My first real job out of school was as an analyst in the M&A division of a publicly traded services and real estate company. My team's job was to build hundred-page decks — analysis after analysis — that the business would use to make billion-dollar decisions about what to acquire, divest, or merge. We swam in oceans of Excel sheets and elaborate SQL queries, tracking dozens of moving pieces at once. And my bosses, who had been doing this for twenty years, could open a fresh deck and find a single wrong number in a tiny column buried on page sixty just by scanning it with their eyes.
It looked like magic. It wasn't. They had been trained to read numbers the way a doctor reads an x-ray. They knew which numbers should land in which ranges, which ratios should hold against which others, what the story behind a row of figures should sound like. When something didn't fit the story, they noticed before they consciously knew why.
That training was the first half of my education. The Times was the second half.
At the Times, the analysts and editors and operators in our data group spent the first ten minutes of every important meeting not talking about the analysis. They were vetting the numbers themselves. Where did this come from? What's the denominator? When did you last validate it against the source of truth? What changed between this version and the one we saw last week? It looked, at first, like nitpicking. It wasn't. It was a culture protecting itself.
Because in that culture, when you put a number on a slide and used it to make a decision, you owned that number. Your reputation rode on it. If the number turned out to be wrong, it wasn't a small thing. It wasn't the spreadsheet that lost trust. You lost trust. And in a building full of people whose careers were built on being trusted with the truth, that mattered enormously.
What I came to understand is that the integrity of the numbers and the integrity of the people were the same thing. You can't have one without the other. A high-trust data culture isn't built by buying better software or hiring better analysts. It's built by making it expensive — socially, reputationally — to be casual with the truth. And it's built by making it safe to ask the dumb question, to admit you don't know, to say I'm not sure the calculation is right, can we walk through it together. Peter Senge writes about this kind of organizational habit in The Fifth Discipline. The frame he gives it is useful. But the practice itself is older and simpler than any framework: it's a culture of people who would rather be slow and right than fast and impressive.
I took that culture with me when I joined Recharge to build out the data function. I built the team from scratch, which meant I had the rare privilege of designing the culture before any habits had calcified. The rule we hired against, again and again, was this one: don't be afraid of numbers. If you're afraid, simplify them until you're not. Break them apart. Build them back up. Track the lineage. The moment you start nodding along with a number you don't actually understand, you've taken the first step toward bad decisions. The analysts and engineers we hired weren't the flashiest. They were the ones who would slow down to be sure.
I'm now doing this work as a fractional CDO/CPO for founder-led ecommerce brands, and the first thing I look at when I come into a new company is not the dashboards. It's the meetings. How does the team talk about numbers in front of each other? Who's allowed to challenge a chart? When a metric disagrees with another metric, what happens? Is anyone embarrassed to ask where a number came from?
You can tell more about whether a business will make good decisions from ten minutes inside one of those meetings than from a quarter of looking at its data stack.
If you're a founder building a brand past $1M and you're starting to feel the wobble — finance and growth telling different stories, dashboards multiplying, decisions getting harder — the temptation is to think you need better tools. You probably don't. You need a culture where it's safe and expected to say let's slow down and check the number. You need the kind of team where the integrity of the data and the integrity of the people are inseparable.
That's the work. And it starts long before the dashboards.