Which decision the dashboard supports

A dashboard with no decision behind it is wallpaper

A dashboard request almost always arrives as "we need a sales panel" or "we want to see everything about the product". Neither phrasing contains a single property you could check the finished work against. A dashboard is not a display case for data. It is an interface to a repeating decision: someone looks at the screen on a regular schedule and, depending on what they see, does something or deliberately does nothing. If that something does not exist, the panel enjoys three weeks of enthusiasm and then gets opened once a quarter out of inertia.

Three questions before the first chart

Which decision is made from this screen. Not "we keep an eye on things", but something concrete: move budget, call extra couriers onto the shift, stop the rollout, schedule a review.
Who exactly makes it. A name and a role, not "management". A decision has one owner, a dashboard has one primary viewer, and everyone else is secondary.
How often the decision comes up. Once an hour, once a day, once a month. Frequency drives everything else: history depth, refresh period, level of detail.

Dashboard passport (filled in before development)

Decision    : whether to call extra couriers for the evening
Owner       : city operations lead
Frequency   : daily at 15:00
Trigger     : evening order forecast above planned capacity
Action      : open shift slots and send out invitations
Cost of miss: a missed peak equals roughly 4% of daily turnover
Metrics     : orders actual/forecast, free capacity, late share
Freshness   : every 15 minutes, latency no more than 20 minutes
Death       : if no shift is opened from it for three months

Insight. A filled-in passport cuts the workload harder than any optimisation: half of the requested charts fail the Action line and get dropped before a single query has been written for them.

Decision frequency dictates the build

A daily decision needs freshness and a short history: the last two or three weeks, a comparison with the same week earlier, no annual trends. A quarterly decision is the opposite. It wants a horizon of a year or more, smoothed series, breakdowns by segment, and nobody cares whether the data is fresh within the last day. Mixing the two modes on one screen is pointless. The operational viewer never scrolls down to the quarterly cuts, and the strategic viewer drowns in the noise of daily swings.

Frequency also sets how much explanation the screen has to carry. A decision made every hour belongs to someone who already knows every metric definition by heart, so labels can stay terse. A decision made once a quarter belongs to someone who last opened the screen three months ago, so every axis, every threshold and every filter has to be spelled out on the page itself.

Common mistake. A dashboard for everyone is the most expensive way to build nothing. A panel that carries board-level revenue, marketing conversion and the support queue at the same time is opened regularly by nobody: each viewer needs 20% of the screen and sees 80% of somebody else's noise.

The one-month test

If a dashboard is older than a month, ask three of its viewers one question: name a decision you made from it in the last thirty days. Not "I open it every morning", but an actual decision with a consequence. If all three answer "nothing specific", the panel supports a habit rather than a decision. Such panels are more honest switched off than kept alive. Every living dashboard costs warehouse queries, time spent diagnosing breakages and, above all, attention: it takes up a slot in the list of things people treat as a source of truth.

Pro tip. Keep a Death line in the passport: the condition under which the dashboard gets switched off. Without a condition written down in advance, no dashboard is ever deleted. Nobody is willing to remove somebody else's panel without a formal basis, and the portfolio grows linearly with time.

Cheat sheet

A dashboard is an interface to a repeating decision, not a display case for data.
Before the first chart, fix the decision, the owner, the frequency and the cost of a miss.
Decision frequency sets freshness, history depth and level of detail.
No decision in a month from three viewers means switch it off.

1. What does the Action line in the dashboard passport check?

2. Why do daily and quarterly decisions sit badly on one screen?

3. What does an answer of 'nothing specific' from every viewer about the last month mean?

🔒 Answer the question correctly to move on to the next lesson.