Management dashboard: why every screen in the business shows a different number

You asked how many active customers the business has. You got three answers.

The operations manager counted everyone with open work. Accounting counted everyone billed this month. The salesperson counted everyone who signed, including two who haven't started yet.

Nobody was wrong. Nobody had simply defined what an active customer is.

This is the main reason management dashboards fail to hold up in small and medium businesses, and it is almost never technical.

Management dashboard and BI system: not the same thing

The two terms are used interchangeably, and that creates a wrong expectation.

A BI system is analysis infrastructure. It pulls data from different sources, lets you cut it by any dimension, and is meant for someone who knows how to ask questions and dig into the results. It is a research tool.

A management dashboard is an operational screen. It answers a small number of fixed questions, in the same format, every day. It is not a research tool, it is a decision tool.

A business that orders BI when it needs a dashboard gets impressive slicing capability that nobody uses. A business that orders a dashboard when it needs BI gets a few numbers and then discovers it can't ask why.

The practical rule: if the question comes back every week in the same wording, it belongs in the dashboard. If it is asked once a quarter and differently each time, it belongs in analysis.

The problem is not in the display, it is in the definition

Before choosing any tool, three questions have to be settled. Without them, every screen will show a number, and no number will be reliable.

What counts as a customer

A customer who signed and hasn't started, a customer who started and hasn't paid, a customer who finished a while ago and still gets support, a customer who is actually several companies of the same owner. Each of these needs an explicit decision, otherwise every department will decide on its own.

When revenue is counted

On the signing date, on the invoice date, or on the day the money actually comes in. Three legitimate answers, and three completely different graphs. A business that presents revenue without stating which method it uses is presenting a figure that cannot be compared to the previous month.

Which time range

A calendar month, a trailing period, or a rolling quarter. Businesses with seasonality that present month over previous month see a drop in every seasonal period and draw wrong conclusions from it.

This is the position worth adopting before buying any tool: a dashboard does not produce truth, it publishes the definitions you have already settled. If you haven't settled them, it will publish the assumptions of whoever built it.

The practical rule: next to every metric, write who is counted, by which date and over which range. If you can't answer in one sentence, the metric isn't ready yet.

Five metrics every service business needs, and one everyone displays in vain

What is worth displaying:

• Inquiries that came in versus inquiries that received a response. The gap between them is the money you already paid for and did not realize.
• Average time between stages in the process. Not how long everything takes, but where exactly the time piles up.
• Jobs stuck above a defined threshold, with a name and a reason. Not a number, but a list you can act on immediately.
• Weighted revenue forecast for the coming period. By stage in the process, not by the optimism of whoever is reporting.
• Clients with no contact for more than a defined period. This is the metric identified with churn and the least displayed.

And what is almost always displayed in vain: the number of actions performed in the system. How many tasks were closed, how many calls were logged, how many records were updated. This is a load metric, not a performance metric. It rises when the team is busier, and there is no necessary connection between it and an outcome.

The practical rule: every metric must answer the question "what will we do now if it deviates?" If there is no answer, it belongs in a report and not on the main screen.

Why dashboards are abandoned after a month

The pattern repeats: you build, you get excited, you check every morning, and then you stop.

Three reasons:

The numbers do not add up with reality. A manager who sees a figure that does not match what he knows will stop trusting the screen, and rightly so. Trust is lost once and does not come back easily.

The screen displays a normal state. A dashboard that shows everything is fine does not generate action, and therefore there is no reason to enter it. A good dashboard should be quiet most of the time and alert when something deviates. If it is viewed out of curiosity alone, it is probably not doing the job.

The figure comes from manual entry. Every metric that depends on someone updating a field will become stale exactly when the load increases, meaning exactly when you need it.

The practical rule: a good dashboard does not compete with a report on the amount of information. It shortens the path between a deviation and a person who can handle it.

Where the data point needs to sit for the dashboard to be reliable

The dashboard is the last layer. It cannot be better than the source it draws from.

When data is scattered between a customer system, a spreadsheet, accounting software, and a calendar, there are two ways to build a dashboard. The first is to connect them all and sync. The second is for the data point to simply be in one place.

The first works, and it has a fixed operational cost: every connection is a point of failure, and a sync failure shows up as a wrong number that someone used to make a decision. On top of that, updating one definition requires updating every one of the connections.

This is why at alcyone14 we don't build a dashboard as a separate product. It's a view over the same system that holds the leads, customers, projects, and charges. There is no sync because there are no two sources, and when you redefine what counts as an active customer, the definition changes in one place.

The practical rule: before you connect another source, ask whether it adds truth or just another place where you can get a different version of the same number.

How to define a performance metric you can act on

A metric that doesn't lead to action is decoration. Three conditions separate the two:

• A specific someone is responsible for it. A metric owned by everyone is a metric owned by no one.
• There is a threshold that triggers an action. Not "improve response time," but "when response time crosses the defined threshold, a task opens." Without a threshold, the metric only documents.
• It is measured at a frequency that allows correction. A metric updated once a quarter teaches you about a quarter that has already ended.

And there is a fourth condition that sounds trivial and isn't: the metric must be one that cannot be improved without improving the business. Any metric that can be improved by changing how it's reported will be improved that way, and it will happen without anyone intending to cheat.

The practical rule: define an owner, a threshold, and an action for every metric. If one of these three is missing, don't present it as if it were a management metric.

Ready to see one number instead of three?

In the businesses we see, the problem isn't that there's no data, but that there's no agreement on what it says. In a 20-minute conversation we'll go over the questions you ask every week, and we'll say honestly whether what you're missing is a screen or a decision about definitions. Maybe what you need is an operational dashboard over one system.

Frequently asked questions

What's the difference between a management dashboard and a report?

A report is a snapshot of a given moment, built for a specific purpose and read once. A dashboard is a live view meant to be returned to, and therefore it must be minimal. A report can contain many rows. A dashboard that contains many rows is not a dashboard.

How many metrics should a management dashboard have?

A small number of metrics. This is not an aesthetic rule. A manager who opens a screen and sees too much data will scan rather than read, and the moment they scan they miss exactly the exception the screen was built for. What does not fit in the dashboard can live in a separate report.

Does a small business need a BI system?

In the businesses we see, usually not, at least not as a first step. A small business that suffers because every screen shows a different number will not solve it with a stronger analysis tool; it will get faster mismatches. First you unify the data source and define concepts, and then you consider an analysis tool.

Why don't the numbers in the dashboard match what I know?

Almost always one of three: a different definition of the same concept, a different time range, or data that comes from a source that was not updated. It is worth checking them in that order, because the first is the most common and also the simplest to fix.

Who should see the dashboard?

Not everyone, and not the same screen. A manager needs exceptions at the business level. A team lead needs exceptions at their team's level. An employee needs what concerns them. The same screen for all three creates exposure you did not want and noise you do not need.

How long does it take to build a management dashboard?

The technical part can be short. What takes time is the decisions about the definitions, and that is exactly the part that must not be skipped. Businesses that asked for a fast dashboard received a screen that displayed numbers quickly, and stopped looking at it within a short time.