I believe in data. I have a degree in Marketing Analytics. I've built scorecards, run test-and-learn portfolios, and designed frameworks that turned messy operational questions into measurable ones. I don't think decisions should be made on gut feel alone.
But the most important things I've learned about data-driven decision making came from the moments when the data wasn't enough. When the spreadsheet showed me what was happening but couldn't tell me why. When the numbers were technically correct but the decision they pointed toward was wrong because something human was missing from the picture.
I've spent my career in operations and strategy across manufacturing and retail. In every role, data was central and the decisions that actually worked were the ones where someone paired the data with context, experience and the voice of a person who was closer to the problem than any dashboard could get.
That's what I want to talk about. Not whether to be data-driven. Of course you should be data-driven. But how to be data-driven without losing the human in the numbers.
Data Without Context Is a Status Update
Early in one project, I took over responsibility for tracking guest satisfaction across all guest-facing programs at a large retail company. The data was there in the form of NPS scores, guest feedback themes and trend lines going back months. I could tell you exactly what the satisfaction number was for any program in any period.
And it was almost useless.
Because satisfaction data on its own is a symptom report. Guests can tell you what felt good and what felt wrong. They can't tell you why. Only the teams running each program know what operational decisions, staffing changes, or process shifts might be creating what guests are experiencing. A guest tells you the experience felt rushed. She can't tell you that a scheduling change compressed appointment windows that month.
So I built a bridge between the guest data and the people closest to each program, by creating a regular rhythm of context-sharing so that when feedback moved, we understood why. The report going to decision-makers stopped being just numbers and started being the complete health scorecard which was actionable.
The difference was immediate. Before, a satisfaction drop was a status update. "The number is down." After, it became a strategy conversation. "The number is down because this specific change affected this part of the experience, and here's what the team is doing about it."
That's the human layer. The data tells you what. The person closest to the operation tells you why. Pairing those two is where actionable insight actually lives.
The Best Analytical Move I Ever Made Wasn't Analytical
In another project, I was trying to understand why certain guests weren't coming back after a first visit. We had the quantitative data in the form of guest feedback, retention numbers, competitive research. All of it was informative but not truly complete.
So I did something that felt different from my usual analytical work. I booked appointments at competitor locations - as a customer (not a strategist doing competitive research). I experienced the consultation, sat through the service, paid at the checkout counter. Then I did the same at our own locations.
The gap I noticed as a customer was not the gap I had expected from the data. The data suggested the core service quality was fine. But the experience around the service was where the real gap lived - The greeting, consultation, communication, the feeling of being cared for, not just served. A guest can rate the service highly and still not come back, because satisfaction with the service and satisfaction with the experience are two different things.
Being the customer changed what I saw when I went back to the data. The numbers hadn't changed. My understanding of what they meant had.
I think this is one of the most undervalued practices in data-driven work. Going out and feeling what the customer feels. Not as a mystery shop with a clipboard. As a genuine attempt to stand where the customer stands and experience what the customer experiences. You notice things you didn't think you would but then thats where the insight lives.
The data said "retention is low." Being the customer told me where the experience broke. No dashboard in the world would have surfaced that.
Measure Three Things, Not One
In a test-and-learn initiative, my team ran a portfolio of structured experiments across service offerings. We needed a way to evaluate each test and decide whether to expand it, pivot it, or stop it.
The instinct in most organizations is to build a financial scorecard. Did it make money? What's the margin? Is it profitable? Those are important questions. But a financial-only scorecard misses real signals.
So we built a scorecard with three lenses. Financial impact. Guest impact. And associate impact.
Financial measured the obvious things. Guest impact measured whether the experience was working. And associate impact measured something most scorecards ignore entirely. Whether the role was sustainable for the person doing it.
That third lens changed decisions.
A test might look financially promising but operationally unsustainable for the people running it. A different test might have modest financial returns but strong guest signals and a role that associates could actually do well. The three lenses together painted a picture that any single lens would have distorted.
When we reviewed the full portfolio, the scorecard gave clear answers. Some tests moved forward. Others didn't. The three lenses together made it possible to have honest conversations about what was working and what wasn't. And the associate lens was a big part of why the right calls got made.
The human in the numbers, in this case, was literally the human doing the work. If your scorecard doesn't measure whether the operation is sustainable for the people inside it, you're optimizing for something that won't last.
Data Tells You What's Possible. It Also Tells You What Isn't.
In one of my most instructive projects, we used a hypothesis-driven approach for why a set of locations were underperforming. We tested it against multiple data sources and then built a diagnostic tool that helped translate what we were seeing at the portfolio level into something actionable at the individual location level.
What we found was that the constraint was structural. The talent pool was small and competitive. Every player in the industry was hiring from the same limited group. No amount of operational creativity on our side was going to expand the supply of qualified people quickly enough to close the gap. It meant the operational levers we'd designed would work where conditions allowed, but a meaningful share of the gap couldn't be closed by operations alone.
The finding was that there was a structural ceiling which meant redirecting efforts to where it could actually move the needle. Data-driven decision making means listening to the data even when it says "this problem isn't solvable with the tools you have." That's one of the hardest forms of being data-driven. Because it requires you to stop, not do more.
The Translator Makes the Numbers Useful
In one project, an enterprise scorecard got reviewed by leadership every period. It had metrics from every major department. The data assembly was extensive which included gathering actuals and targets across teams, reconciling them and tracking movement.
Every period, the work involved partnering with the teams behind the numbers to understand why a metric improved or declined. What changed operationally or what was being done about it or what the trend meant. Then translating all of that into a story leadership could act on. Not just "this number went up" but "this number went up because of this, and here's what it means for where we focus next."
That translation layer was extremely manual. It had a quick turnaround. And it resisted automation. Not because the technology wasn't there to automate the data assembly. But because the value lived in the human judgment that connected the dots between raw metric movement and operational reality. A dashboard can show you the number moved. Only a person who talked to the teams can tell you what the movement means.
I think this is true across more of operations than most people realize. The cells aren't the deliverable. The story around the cells is the deliverable. And that story requires a person who understands both the data and the context well enough to connect them into something a decision-maker can use.
That's the translator role. And in my experience, it's one of the most undervalued roles in any organization.
Data-Driven Doesn't Mean Data-Only
If I had to name the through-line across everything I've learned, it would be this.
Data is essential. It brings rigor. It brings honesty. It keeps you from making decisions based on hope or politics or whoever talks loudest in the room. I would never go back to a world without it.
But data is incomplete. It tells you what happened. It doesn't always tell you why. It measures what you designed it to measure but often misses what you didn't think to ask. And it can't hold the texture, the context, or the human experience that explains what the numbers actually mean.
The best decisions I've been part of happened when someone paired the data with something else such as going out and being the customer, talking to the people closest to the operation, measuring the associate's experience alongside the guest's - basically translating raw metrics into a narrative that connected the dots.
Data-driven decision making that loses the human in the numbers isn't data-driven. It's data-limited. The spreadsheet is the start. The human is the rest.