Data Is a Mirror, Not a Map
How first-time CEOs can uncover the story behind the metrics.
Did you know? I’m writing a book, and just launched its pre-sale! The First-Time CEO: 50 Conversations on Leading While Becoming will include 50 podcast conversations with modern leaders, hosted by me, reflections on building an AI startup as a first-time CEO, and some of my most-read newsletters. Pre-order your copy and follow my progress as I write it in public!
P.S. Wanna get some evergreen publicity? Choose a Founding Membership Pack instead of just the book!
Dashboards are seductive.
They promise clarity, but what they often deliver is a kind of quiet frustration. Pageviews climb. That’s good, isn’t it? Yet conversions barely budge. The numbers alone offer no explanation. They are, at best, a mirror of what happened, not a map of why it happened.
In the early days of my career, I treated dashboards like crystal balls. I would stare at them, sometimes for hours, hoping the lines and bars would reveal some hidden logic of human behavior. But over time, I realized that numbers rarely speak for themselves. They reflect, they hint, they suggest—but they do not explain.
So I learned to approach metrics differently. Instead of counting, I question. If something shifts, I ask:
Who exactly changed their behavior?
What was their last interaction before the shift?
When did this pattern begin, and what else was happening then?
It turns out, the work of understanding isn’t in the dashboard—it’s in connecting the dots the data leaves behind. Metrics, in this sense, are not maps. They do not tell you where to go; they reflect where you’ve been, where attention lingers, where behavior veers.
I follow a simple rhythm:
Signal. Identify the metric that feels “off.” Not every dip or rise warrants investigation, but some stir a quiet unease—a spike in churn, a drop in engagement. Those are your signals.
Segment. Break it down—by cohort, by channel, by time. Often, overall trends conceal nuance: perhaps only one cohort reacted differently, or a particular channel behaved unusually. Segmentation reveals the details hidden in the aggregate.
Trace. Follow the user path backward, from outcome to trigger. Who saw what, clicked what, ignored what? It is a detective’s work, tracing footsteps through an invisible landscape.
Hypothesize & Test. Define exactly what you expect to see if your idea is correct. Test your assumptions. Let the data speak back to you.
And then, the hardest part: leave space for surprise. For what you didn’t predict. That’s where insight lives—not in the tidy lines of a graph, but in the quiet, stubborn questions it provokes. It is in those anomalies, those stubborn outliers, that you begin to understand the human patterns behind the numbers.
I remember one particular instance: we had launched a new onboarding flow, and engagement looked promising. Conversions, however, remained stubbornly flat. Segmenting by cohort revealed a small subset abandoning midway. Tracing their path backward uncovered a tiny copywriting change on a single page. Hypothesis confirmed: that line of text, innocuous to most, was a trigger for confusion in just that cohort. We reversed it, and conversion rose. A small anomaly had held the story; the dashboard alone would never have told us.
Numbers are mirrors, not maps.
They reflect patterns, but the understanding—the insight—comes from curiosity, patience, and careful attention. Measure with questions, not just counters. Connect the dots. Trace the paths, test your assumptions. And always leave room for what you did not predict.
And perhaps the most important lesson of all is this: data is never neutral. It is shaped by the questions we ask, the moments we notice, and the patterns we are willing to follow. It is a reflection of our curiosity, our patience, our willingness to linger with uncertainty. To read numbers as stories, rather than answers, is to acknowledge the complexity of the people behind them—their unpredictability, their stubbornness, their capacity to surprise. And it is in that space, between reflection and interpretation, that insight becomes almost poetic: fragile, illuminating, and human.
The Data Interpretation Playbook for First-Time CEOs
To turn reflection into action, I’ve developed The Data Interpretation Playbook for First-Time CEOs. It’s a practical, no-BS guide for moving from raw metrics to insight:
start with the question
identify the signal
segment ruthlessly
trace the user path
hypothesize and test
leave room for surprises
document everything
Each step pairs with real-world examples—from spotting churn in a SaaS startup to tracing drop-offs in onboarding—and recommended platforms like Mixpanel, FullStory, Optimizely, and Google Analytics.
Think of it as a full feedback loop: metric → question → cohort → path → hypothesis → test → insight → learning. Numbers are mirrors, not instructions—and this playbook shows you exactly how to read them.
#GoldenFindings
From Reports to Insights: Why Dashboards Alone Don’t Drive Better Decisions – Explores how to turn data into actionable insight.
Data Storytelling for Managers: Turn Dashboards Into Decisions – A narrative approach to using dashboards effectively.
Why Data Interpretation Is More Important Than Data Collection and Analysis – Contrasts having numbers vs. making sense of them.
13 Data Analysis Questions to Improve Your Business Reporting Process – A list of probing questions to elevate reports from noise to narrative.
#CEOCheck
When a key metric shifts, what story does the data actually tell — and what assumptions are you making if you don’t dig deeper?
Let’s Connect!
For more authentic content on leading companies for the first time, follow me on LinkedIn and Instagram, and The First-Time CEO podcast on Spotify, YouTube, and Apple Podcasts. Season 2 coming soon! Here you can find more information about me and my career.
Brilliant framing on the mirror vs map distinction. That onboarding flow example is clutch becasue it shows how aggregated data masks the actual problem - without segmenting by cohort you'd never catch that one copywriting change tanking conversions for a specific group. I've been in situations where teams stare at dashboards waiting for answers when the real work is asking better questions about the anomalies.