Nobody Understands Your Sunburst Chart

Here's a test for whether a dashboard is any good: after you hand it over, does the stakeholder stop calling you?

If they keep coming back to ask what a number means, where a filter is, or "which line am I supposed to look at" — the dashboard failed. It doesn't matter how clean the code is or how clever the visualization looks. A dashboard that can't be read without the developer sitting next to it is not a tool. It's a dependency.

I bring this up because I keep seeing the same two mistakes, and they both come from starting in the wrong place.

Start with the questions, not the query

The most important work on a dashboard happens before you write a single line of SQL. Before the query, before the first chart, before you even open the tool — you have to know what questions the stakeholder actually needs answered.

That sounds obvious. It almost never happens. Most dashboards start from the data: "here's what's in the warehouse, let's put it on a screen." The result is a wall of charts that technically display information and answer nothing. The stakeholder scrolls, nods politely, and never opens it again.

Flip the order. Sit with the person who's going to use it and get their real questions out on the table. Not "what metrics do you want" — that just gets you a shopping list. Ask what decision they're trying to make. "How do I know if we're on track this month?" "Where are we losing customers?" "Which region needs my attention this week?" Those are the questions. The dashboard exists to answer them, in the order they'll be asked, and nothing more.

There is no universal checklist

Somewhere along the way, a myth took hold that every good dashboard answers the same small set of questions — three of them, usually. Build around those and you're done.

I don't buy it. A dashboard is tailored to its audience. A CFO, a warehouse manager, and a marketing lead are not asking the same three questions, and forcing all of them into the same template is the fastest way to make a dashboard that's useless to everyone. The CFO wants margin and cash. The warehouse manager wants throughput and bottlenecks. The marketing lead wants where the leads came from and what they cost. Same company, completely different dashboards.

Limiting yourself to a fixed set of questions doesn't make a dashboard focused. It makes it generic — and generic is just garbage with good spacing.

Clarity beats cleverness, every time

The second mistake is the fun one: fancy charts.

Sunbursts. Boxplots. Treemaps. Network graphs. They look impressive in a portfolio and they win design awards, but drop one into a real business meeting and watch what happens. Ninety percent of the room can't read it. They don't want to admit it, so they don't ask — they just quietly stop trusting the dashboard. A chart nobody understands isn't sophisticated. It's decoration.

The job of a chart is to make an answer obvious in about three seconds. A bar chart, a line, a clear number with context, a simple table — these are "boring" precisely because they work. If the person looking at your dashboard has to stop and decode the visualization itself before they can even get to the insight, you've added a layer of friction between them and the answer. And every layer of friction is another reason to call you instead of reading it.

Save the exotic charts for the rare case where they genuinely tell the story better and the audience can actually read them. That case exists. It's just far less common than dashboard portfolios would have you believe.

The real measure of a good dashboard

Strip it all down and it comes to this: a dashboard is good when the right person can answer their own question, on their own, without you in the room.

That means starting from their questions instead of your data. It means building for that specific audience instead of a generic template. And it means choosing clarity over cleverness every single time. Do those three things and you build something people actually use. Skip them and you build something beautiful that quietly gets abandoned.

The clean code, the clever visuals, the technical polish — none of it counts if the person it was built for has to pick up the phone. Build the dashboard that makes them put the phone down.