TL;DR: Data-driven means the data makes the call. Data-informed means the data informs the call, but humans make it. For most government and business decisions, you want to be data-informed. Keep humans in the loop while leveraging good data. But here’s the catch: neither approach works if your data is a dumpster fire. Get your data governance house in order first.
“You Should Be More Data-Driven”
I was watching a YouTube video the other day on, you guessed it, data governance. For those of you that have been around long enough, you know I listen to YouTube in the background while I work. It’s how my brain processes things. Anyway, this particular video was about how to be more “data-driven,” and the presenter said something that stopped me mid-task. He drew a line between being data-driven and being data-informed. Two different things. And it clicked immediately, because I realized I’d been using the wrong term for what I actually believe in.
I’ve been telling people we need to be more data-driven. What I actually meant, what I’ve been building toward, is data-informed. The distinction sounds academic until you see it play out in real decisions. Then it matters a lot.
The Actual Difference (And Why I Care)
Data-Driven: The data makes the decision. You’re executing what the numbers say. Your credit card company flagging a weird transaction at 3am? That’s data-driven. No human reviewed it. Algorithm said “suspicious,” card got locked. Done.
Data-Informed: Data provides input, but a human makes the call. Context matters. Experience matters. That thing your gut knows but can’t quite articulate? It gets a seat at the table too.
Andrew Chen, General Partner at Andreessen Horowitz, nails it:
“The difference between data-informed versus data-driven is that you weigh the data as one piece of a messy problem you’re solving with thousands of constantly changing variables. While data is concrete, it is often systematically biased.”
Read that last part again. Data is often systematically biased. Not everything is an optimization problem. Some decisions need human judgment, local knowledge, and the ability to weigh factors that don’t fit in a spreadsheet.
The Pothole Parable
Let me paint you a picture. Fictional, but anyone who’s worked in local government will feel this in their bones.
It’s budget season. (When isn’t it?) You’ve got a pavement management system that scores every road segment. Cracks, potholes, surface degradation, the works. Purely data-driven approach says: repave the ten worst-scoring streets. Easy. Meeting adjourned. Lunch at 11:30.
Except.
Street #3 on your list is getting completely torn up next spring for a water main replacement. You’d be paving something you’re about to jackhammer into oblivion. That’s not efficiency. That’s lighting money on fire.
Street #7 is in a neighborhood where Mrs. Patterson has been calling the city manager every single week for two years about the road in front of her house. Her street scores “moderate.” But Mrs. Patterson votes, Mrs. Patterson talks to her neighbors, and Mrs. Patterson has a point. Her street genuinely sucks to drive on even if the algorithm disagrees.
Street #9 runs past Lincoln Elementary. The potholes aren’t even the real problem. It’s drainage. Every time it rains, there’s a lake in front of the crosswalk. Kids are literally walking through puddles to get to school. Repaving won’t fix that. It’ll just give you a smoother surface under six inches of water.
A data-informed approach uses those condition scores as a starting point, then layers in capital project schedules so you don’t pave what you’re about to dig up. Community input, yes, even from Mrs. Patterson. Institutional knowledge from your DPW crew who actually drive these roads every day. Practical constraints the data doesn’t capture.
The data didn’t get worse. Your decision got better.
Why This Matters for Places Like Rochester
I’m CIO for a city of about 33,000 people. Around 350 employees. We’re not Google running A/B tests on millions of users. We’re making decisions that affect actual neighbors. Real families. The guy who owns the pizza shop downtown. The tax preparer who lives on Autumn Street.
When we allocate resources, whether it’s road repairs, community events, or outreach programs, we need data. Absolutely. But we also need the institutional knowledge from the staff member who’s been here 28 years and remembers why we don’t do things a certain way. We need community context that doesn’t show up in any dashboard. Political realities, because democracy is messy. Ethical considerations that go beyond pure efficiency.
Harvard Business Review found that even well-intentioned data-driven approaches fail when leaders “either take the evidence presented as gospel or dismiss it altogether.” The answer isn’t less data. It’s better integration of data with human judgment.
Here’s the Part Nobody Wants to Talk About
None of this matters if your data is garbage.
I know. Unsexy topic. But hear me out.
Most organizations have data everywhere. Permit systems. Finance software. HR databases. GIS mapping. Citizen request portals. It’s all there. The problem is it doesn’t talk to each other. You can’t get a unified view of anything. You’re making decisions with fragmented, incomplete, sometimes outright contradictory information and then wondering why your “data-driven” initiative flopped.
The Supermetrics research team found that 56% of marketers don’t have enough time to properly analyze the data they collect. That’s not a data problem. That’s a data governance problem.
Even small organizations need to think about this stuff. Is our data actually accurate? When was it last updated? Who’s responsible for it? Can we connect information across departments without someone manually exporting CSVs? Can the people who need data actually get to it, or is it locked in someone’s desktop folder? And are we protecting the sensitive stuff?
This is the foundation work. It’s not flashy. Nobody’s winning awards for “implemented proper data governance.” But it’s what makes everything else possible. It’s what we’re doing right now with our data governance RFP, and it’s why we didn’t rush into AI tools before we understood the state of our own data. The governance comes first. The tools come after.
So What’s the Actual Takeaway?
Data-driven means the data makes the decision. Good for automation, fraud detection, algorithms that need to move fast. Data-informed means the data informs the decision and humans decide. Good for policy, strategy, anything involving actual people. And neither one works if your underlying data is a mess.
For most of what we do, especially in government, data-informed is the right call. We’re not trying to remove humans from the loop. We’re trying to make humans smarter.
But first, you’ve got to get your data house in order. That’s the work. That’s the job.
And if anyone tells you they’re “fully data-driven” and everything’s going great, they’re either lying or they’re running a very simple operation. No shade. Just reality.
*Andy Watkins is the CIO of Rochester, NH. AV-8B Avionics Marine. Recovering sysadmin.
Sources:
1. Chen, Andrew. “Know the difference between data-informed and versus data-driven.” andrewchen.com. Link
2. Luca, Michael and Edmondson, Amy C. “Where Data-Driven Decision-Making Can Go Wrong.” Harvard Business Review, September-October 2024. Link
3. Supermetrics. “Data-Driven vs. Data-Informed: A Comprehensive Guide.” supermetrics.com, May 2025. Link