Seven hundred rows of columns. Every metric you could ever want, cleanly labeled, filterable, sitting in a dashboard nobody has opened since the week it launched. I’ve built that dashboard. Twice. Both times I spent days wiring connectors and naming fields, and both times the executive who requested it went back to asking me for numbers over Slack.
The problem wasn’t the tool. It was that I built a library when what they needed was a decision. Good digital marketing reporting starts with one question: what choice does this screen help someone make? If you can’t answer that in a sentence, you’re about to spend a quarter building shelfware.
So here’s the approach I use now. Three decisions, one page, then everything else expands from there. It’s slower in week one and dramatically faster by month three, mostly because you stop rebuilding the same view for every person who asks.
Nobody really wants a dashboard. They want relief.
When a VP asks for “reporting,” they’re usually describing a feeling, not a deliverable. They got surprised by a budget overrun, or a campaign underperformed and they found out late, or they sat in a meeting and couldn’t answer a basic question about pacing. The dashboard is their imagined cure.
Ask what went wrong the last time they felt blind. Nine times out of ten you’ll get a specific story: a channel stopped converting on a Tuesday and nobody noticed for two weeks. That story is your spec. Build the view that would have surfaced it on Wednesday morning, and you’re done. Everything else is optional.
I’ll take a stance here. Most teams should ship a smaller dashboard sooner rather than a bigger one later. A single page that answers three questions gets used daily. A twelve-tab report gets opened once and quietly abandoned, and the abandonment has nothing to do with quality.
What are the three decisions your data needs to support?
Write them down as verbs, not nouns. “Monitor channel performance” is a noun dressed up as a verb. “Decide whether to shift budget from paid social to email by Friday” is a real decision with a deadline attached.
Three is the right number for a first pass because it forces triage. When you cap yourself at three, you have to argue about what matters, and that argument is the actual work. Try it with the people who’ll use the thing:
- What would make you change what you’re doing this week?
- What do you get asked in meetings that you can’t answer fast?
- What do you check first thing Monday morning, if anything?
Their answers rarely match the metrics they originally requested. Someone asks for impressions and it turns out they actually care about how many of those impressions came from a market they can serve. That reframe is worth more than any chart you’d have added.
The Three Decision Test
Here’s the framework I run before designing anything, and it’s the part most guides skip. Take every metric someone has requested and put it through three gates.
1.The action gate.
If this number changed today, who does what differently? “Nobody, but it’s interesting” means it goes on a separate exploratory view, not the main one.
2. The freshness gate.
How often does this need to be true? A number that only matters quarterly shouldn’t load on your daily screen. Hourly numbers on a weekly decision just add noise.
3.The trust gate.
Can you trace this number back to its source without a phone call? If two systems disagree about the same metric, fix the definition before you build the chart. A dashboard that contradicts the CRM is worse than no dashboard, because people stop believing all of it, including the parts that are correct.
Run twenty metrics through those three gates and you’ll typically end up with five or six survivors. That’s your first page. I’ve watched this cut a proposed build from six weeks to nine days, and the shortened version got adopted faster because it loaded instantly and fit on one screen.
Build order matters more than tool choice
Teams spend weeks comparing platforms and about twenty minutes deciding what goes on the page. Flip that. The tool question gets much easier once you know you’re displaying six numbers with two dimensions each, because almost anything can do that.
Start with the decisions, then the metric definitions, then the visual, then the automation. Automation last, always. Hand-pull the numbers for two weeks if you have to. Watching yourself copy the same figures into a sheet teaches you which ones you actually look at, and it’s much cheaper than discovering that after you’ve hardwired them into production. According to a long line of research covered by Harvard Business Review, people consistently overestimate how much data they’ll act on and underestimate how much presentation shapes the decision. Build for the version of yourself that glances at the screen for four seconds.
A concrete example that changed how I build
A retail client kept asking why their monthly report felt useless. The report had forty metrics. The only person who read it was the marketing director, and she read it while waiting for coffee.
We scrapped it and built one page around three decisions: spend more or less on each channel, pause or keep any creative, and flag any region where cost per acquisition broke from its own recent range. Three blocks. Every number tied to a named action.
The interesting part came later. She started forwarding screenshots to her regional leads, and they started asking for the same numbers on their own screens. Demand grew from a single page, not from a catalog. Her teams wanted the same clarity, not a different set of tools. Broader context on how companies sustain this kind of internal habit shows up in reporting from Business Insider, which tracks how businesses adapt their operations around data they actually use.
Note what we didn’t build: no executive summary PDF, no anomaly detection, no mobile app. All fine things. None of them on the path to the three decisions.
A two-minute check you can run this week
Open your current dashboard. Time how long it takes you to answer this: is anything off track right now, and what should we do about it?
If you can answer in under two minutes, your dashboard is probably fine and you should leave it alone. If you’re clicking through tabs, cross-referencing a spreadsheet, or messaging someone for context, the problem isn’t your data volume. It’s that nobody wrote down the decision. Go back to the Three Decision Test, pick your three, and cut everything that fails the gates. You can always add tabs back later, and you almost certainly won’t.
One more thing worth knowing: research on how people scan screens, summarized by the Nielsen Norman Group, has long shown that most users don’t read pages in full, they scan for the first thing that looks relevant and leave. Your dashboard is a page. Assume scanning, not study.
What to do next
The teams that get this right aren’t the ones with the fanciest stack. They’re the ones who wrote down three decisions on a whiteboard, argued about them for an hour, and then built exactly that. The rest of it, the connectors and schedules and color coding, is downstream of a conversation you can have this afternoon.
So pull up whatever you have now and ask the two-minute question. If it takes longer than that, you already know what to fix, and it isn’t on the fifth tab.
Also Read :- The Role of Distributed Teams in Supporting Modern Business Operations
