Every dataset tells a story. The hard part isn’t finding the data — it’s knowing which parts of it actually matter.
The Problem with More Data
When I started doing analysis work, I made the mistake most analysts make early on: I tried to show everything. If the data was there, I included it. More charts, more tables, more metrics. I thought comprehensiveness was the point.
It isn’t. Comprehensiveness is the enemy of clarity.
The person reading your report has a question they need answered. Your job is to answer it — not to demonstrate how much data you processed to get there.
Start with the Question, Not the Data
Before I open a spreadsheet or write a query, I write the question in plain English at the top of a blank document. Not “analyze sales performance.” Something specific: “Did the March promotion increase average order value for repeat customers in the Southeast region?”
That single sentence eliminates 80% of what I might otherwise spend time analyzing.
The Three-Filter Method
When I’m exploring a new dataset, I use a simple three-step filter:
- Is it relevant? Does this metric connect to the question I’m trying to answer? If not, set it aside.
- Is it reliable? Do I trust the data? Is the collection method consistent? Are there obvious gaps or anomalies?
- Is it actionable? Can someone actually do something with this insight? A finding no one can act on is just trivia.
Most data fails at least one of these filters. That’s okay. Filtering is the job.
The Specific Beats the General
“Sales are up” is noise. “Same-day orders on weekends are up 22% since we changed the checkout flow in November” is signal.
Specificity is what makes an insight useful. It tells you what changed, when it changed, and where to look next.
Conclusion
The goal of analysis isn’t to display data — it’s to reduce uncertainty. Every chart, table, and number you include should serve that goal. When in doubt, cut it. The best analysis I’ve ever seen could fit on half a page.
This is part of an ongoing series on how I approach data work. Next up: how I structure exploratory analysis when I don’t know what I’m looking for yet.