Should briefings be the new dashboards?

Every morning, people open a dashboard and do the same unpaid job: scan a wall of charts for the one that matters. I think that job belongs to the system, not the user. Dashboards should become briefings. Here's the case, and the design rules I'd hold it to.

The morning scan

Open Google Analytics on a Monday. You pick a date range, scan the cards, open a report or two, and try to answer a simple question: did anything change that I should care about?

The data is all there. The answer isn't. You have to assemble it yourself, and the effort grows with everything your product measures.

Why dashboards work this way

A dashboard has to serve everyone who might open it, so it shows everything. It can't predict what you need, so it hands you the full picture and lets you do the filtering.

That was a reasonable tradeoff when software couldn't judge what mattered. It's less a design flaw than a constraint of its time, and software can now do more of that judging, which makes the tradeoff worth revisiting.

Three jobs that should move to the system

The shift I'm arguing for is a change in who does the work:

• Selection. The system should decide what's worth your attention today and why, instead of showing you everything and hoping you notice.

• Composition. The right visual should be generated for each question, instead of a fixed widget with filters.

• Action. Each item should carry a proposed next step, not just a number.

Show me what's most important, help me understand why it matters, and tell me what I need to do about it.

Early signs, and where they fall short

Some vendors are already moving this way. Tableau Pulse sends a digest of insight headlines and summaries to Slack for people who rarely open dashboards. Microsoft has added Copilot summaries to report email subscriptions, so the summary comes to you.

But these are mostly assistants layered on top of dashboards, and many lean toward a feed: more summaries, no end. While I don't yet have evidence that users prefer them, it's a hypothesis I plan to test.

A briefing should not be a feed

A feed is endless and chronological. You scroll until you give up. A briefing should be the opposite: a small, ranked set of about five items, with a clear end.

Scrolling is for people with time to waste. A briefing tells you where to focus, then lets you move on with your life. Everything the system chose not to show collapses into a single line: "All clear: realtime traffic looks normal." The last line of the screen should say it's done.

What shouldn't change

A briefing still pushes information to you, so it still does a dashboard's job: keeping you aware of what you need to know. And people still need a predictable structure to read it.

Think of an inbox or a newspaper front page. The stories change daily, but you always know where the headline, the byline, and the "read more" are. The content flexes. The grammar doesn't.

And this isn't to say that dashboards don't matter; they do for exploration and can serve as pages for additional data. Briefings should just be considered in place of dashboards as the front door to the building.

Design principles for a briefing

Four rules I'd hold a briefing to, so the structure stays readable while the content changes:

• Fixed item anatomy. Every item has the same parts in the same order: severity, headline, change, evidence, action.

• Ranked sections. "Needs your attention," "Worth knowing," and "All clear" appear in the same order.

• A pinned strip. Three to five core metrics stay in the same spot, so people can verify the system isn't hiding something.

• A hard cap, with a reason. Limit the list, and let users tap "why is this here?" to see how it was ranked; transparency builds trust.

The risk to design against: invisible mistakes

A bad dashboard is cluttered. A bad briefing is worse, because it hides things, and users can't see what was left out.

That makes trust the central design problem. Show sources and confidence for every claim. Always show what was skipped. Make it easy to override the ranking. The "All clear" line and the pinned strip exist for this reason: they let people check the system's judgment instead of taking it on faith.

What this means for designers

The work would shift from laying out screens to designing the rules that generate them (a possible role for an agent): what gets selected, how it's ranked, which visual fits which question, and when the system should interrupt you.

That's a systems craft more than a visual one. The deliverable is a component system and a set of composition rules, and the screens are what they produce on a given day.

The question I'd leave you with: when your product decides what's worth someone's attention, how will they be fine with what it left out?