Somewhere in a university I know, a set of test scores lived in a table built for athletics. Not a scandal. Years ago somebody needed a place to put those scores, the sports table had a column going spare, and it worked. It kept working. Then people who never met that decision started pulling those numbers, and the table quietly told them a story that wasn’t true: that these had something to do with athletes. The scores were fine. The meaning had drifted, because meaning doesn’t live in the number. It lives in the story around it.
That’s the part nobody warns you about when you go into data. A number is not a fact sitting still. It’s the residue of some operation, some form, some Tuesday when a person in a specific office needed to record a specific thing for a reason that made sense to them and to no one downstream. Strip that away and the number is generic. Keep it and the number means something. The data was never the thing. The work was the thing. The data is what the work leaves behind.
We get this backwards constantly, and I get it backwards professionally. My title is Chief Data Officer. The whole field is arranged as if the data sits at the center of the institution and everything else revolves around it. Feed the warehouse. Serve the dashboard. Protect the data. It’s a tidy picture, and it’s wrong in exactly the way the old picture was wrong, the one where the sun went around the Earth. You can chart the heavens that way. The math even mostly works, for a while. You’ve just put the wrong thing in the middle, and every conclusion you draw from there inherits the mistake.
The operations are the center. The data goes around them.
Strip a data set down to what it can’t lose and you don’t find data. You find a story. Who created this, for what, on what day, under what pressure, not thinking for one second about you or the four departments that would one day lean on it. That origin story is the primitive. The number is its shadow. And a shadow read without the object that cast it will fool you every time.
I want to be fair to the data-first crowd, because I’ve carried the card. Centering the data is how you get a single source of truth, and a single source of truth is worth fighting for. I once spent three years getting a university to agree on what the words new student mean, and that fight was won on this exact ground: one number, shared, governed, believed. Governance is real work and it holds institutions together. So this isn’t numbers-bad. It’s a question of what orbits what.
Watch what happens at the two ends of the room. On one end, the analyst. Fast with the query, fluent in the join, reading a number with no idea what work produced it, ready to conclude that a school full of test-takers is a school full of athletes because that’s where the column sat. On the other end, the operator. The administrator, the person who runs the process that makes the data. They know the story in their bones and can’t see the structure. They don’t know that the innocuous field they cleaned up on Tuesday just broke a federal report three semesters downstream, because no one ever showed them the whole. Each end is half blind. The rare person reads both, and can walk from one landscape to the other and carry the meaning across without spilling it. That translation is most of a job nobody wrote into the job description.
It’s also why optimizing one department can quietly wreck three others. A department hits its number, tidies a process, renames a field, all of it locally sensible, none of it aware that the same data feeds a formula, a report, a decision somewhere it will never see and never get the bill for. The win is local. The loss is global, and it shows up late, in a place with no way to trace it home. Without someone holding the long view of the whole life cycle, the data takes on a life of its own and starts making decisions nobody signed off on.
Here’s the seat I’m speaking from, since it’s the reason I can see any of this at all. The data of an entire institution flows past the CDO. Recruitment to alumni, plus everything the student never sees: finances, facilities, the machinery under the floor. Most roles get a segment. The CDO watches the whole institution move, in the one medium that ignores the org chart. And what that vantage teaches you, faster than you’d like, is that you are not watching data. You’re watching the institution’s operations, recorded. The data is the institution talking about itself. It is secondary to what it describes, the way a transcript is secondary to the conversation.
Which is also why so much of the job turns out to be language. When two departments use the same word for two different measures, they don’t have a data problem. They have a translation problem that the data merely exposes. Part of governing is holding one language for the institution, and yes, policing it, so a word means one thing across every office that says it. Not because tidiness is a virtue. Because a number you can’t trace to a shared definition is a rumor with decimal places.
So the useful question, the one I reach for in real meetings: before you trust a number, can you tell its creation and evolution story? Who made it, for what, and what were they not thinking about when they did? If you can’t answer that, you don’t have a fact. You have a shape, and you’re about to make a decision by reading tea leaves that happen to have a decimal point.
The data is worth defending. Build the warehouse, guard the definitions, keep the dashboards honest. But none of it is the point. Put the data at the center and you get a picture that’s elegant, confident, and aimed at the wrong thing, the same way the old star charts were: careful math around a mistake. The operations are the center. The data goes around them. The work is the sun, and the data is only what we can see of it from here.