Business

The numbers you check every Monday

Most business dashboards show a lot and change nothing. A number is only useful if someone can act when it moves: few indicators, updated automatically, each with a threshold and an owner. Before building one, the upstream problem has to be solved — where the data comes from.

29 Jan 2026 · 8 min read · updated on 27 Aug 2026

Almost every business dashboard we have seen shares a story: built with enthusiasm, watched for a month, then ignored. It sits there, full of colourful charts, and occasionally someone opens it for a meeting and closes it straight after.

The reason isn't that the numbers were wrong. It is that no action corresponded to them. A figure that doesn't change what you do on Monday morning is not information: it is decoration, and it still costs time to maintain.

The rule: one number, one action, one owner

Before adding any indicator to a dashboard, one question must be answered: if this number gets worse, what do we do, and who does it? If the answer is "we worry", the number is not needed. If it is "we call the supplier" or "we stop taking new work", it is.

From the same question follows a second thing: every number needs a threshold. "Stock is at 42" says nothing. "Stock has been below minimum for three weeks" says what to do. That is the difference between a dashboard that informs and one that asks for a decision.

If a number doesn't change what you'll do tomorrow, it isn't an indicator: it's an ornament with a maintenance cost.The filter we apply to every chart request

How many numbers: far fewer than you will ask for

Experience says five to eight for whoever runs the company, not twenty. The reason is practical: faced with twenty indicators nobody looks at the two that matter, and the typical consequence is not a wrong decision — it is no decision, because there is always a chart suggesting you wait.

People in a department will have others, specific to them, and that is fine: the production dashboard is not the management one. What doesn't work is a single screen trying to serve everyone and serving nobody.

Where to start: the three families

What is coming in. Orders, requests, open quotes. It looks forward and lets you react before the problem reaches invoicing.
What is slowing down. Cases stuck for more than X days, unfulfilled orders, tickets open too long. The most useful family and the least present in the dashboards we find: it says where to act today.
What cost more than expected. Jobs over budget, hours above estimate, returns. It looks backwards, but it is the only way to quote better next time.

Note that revenue is in none of the three. Not because it doesn't matter, but because you already know it and it doesn't tell you what to do: it is the result, not the lever.

The real problem is upstream

Before building the dashboard the uncomfortable question has to be answered: where does the data come from? If someone has to update a sheet every Monday morning, the dashboard is short-lived — because the first busy week that sheet doesn't get updated, and from then on nobody trusts it.

A dependable dashboard reads from the systems you already use, automatically. It is also why the "dashboard" project often becomes, after half a day of analysis, an integration project: the numbers exist, but they live in three places that don't talk.

If the data isn't there or isn't reliable, postpone the dashboard. Building it on shaky data produces something worse than not having it: decisions taken with the confidence a chart gives, on numbers that don't hold.

How it is built in practice

Start from decisions, not from data. What are the three recurring decisions you take every week? The numbers serve those.
One indicator per decision, with its threshold and the name of whoever acts.
Automate collection before polishing the looks. An ugly, true dashboard beats a beautiful one that is a week old.
Take it where decisions happen: a message every Monday with three numbers and the anomalies works better than a site you have to remember to open.
Remove what nobody looks at, after two months. It is the maintenance nobody does and the thing that keeps the tool alive.

The quickest test of whether your dashboard is useful: ask whoever uses it which decision they took last month thanks to a number they read there. If nobody remembers an example, you have a report, not a tool.

How often should the numbers update?

At the same frequency you can act. If a decision is taken weekly, real-time data is useless and costs more. Real time makes sense only where the reaction is immediate: stock, production, availability.

Do we need a dedicated tool?

Not at first. Many companies start perfectly well with the reports their business system already produces plus an automatically generated sheet. A dedicated tool makes sense when data comes from several systems and different departments need different views.

What if departments disagree on the numbers?

The most common case, and not a technical problem: two departments calculate "orders fulfilled" in two different ways and both are right by their own definition. The definition must be written and agreed beforehand, otherwise the dashboard becomes the arena for that argument.

How long does it take to build?

The visualisation part, little: days. The time goes into what sits underneath — finding the data, reconciling it, automating collection. If someone promises you a dashboard in a week, they are drawing charts on data somebody will still have to update by hand.