Business intelligence for

A dashboard is not yet an insight

Over the years, business intelligence has become ever prettier.

We used to get endless lists on green-and-white paper from the printer. Then came Excel, reporting systems, OLAP cubes and eventually dashboards. Today you can practically have any key figure displayed in any colour you like. With speedometer, traffic light and world map, if you absolutely must.

That is progress. It is just that a dashboard mostly answers questions that someone already knew beforehand.

What is the revenue? How many cases did we have? How is the length of stay developing? Which cost centre is over plan? What is the utilisation?

Dashboards are excellent for that. It is just that the real analysis often begins exactly where the dashboard ends.

You see, for example, that the length of stay in a hospital has gone up. Good. Or rather not good. But now at least you know.

The interesting question is: why?

From one question to the next

Perhaps the longer length of stay doesn’t affect the whole hospital at all. So you look at the departments. One department stands out. So you look at the DRGs. Then at diagnoses. At wards. At doctors. At admission and discharge days.

And at some point you find out: this happens mainly with a particular case group, and there remarkably often with patients admitted toward the end of the week.

Now you have found something out.

The problem: when the dashboard was built, nobody knew you would one day ask exactly this question. How could they?

This is the point where classic reporting systems like to get a bit stubborn. For the next question you need a different report, an additional dimension or someone from IT.

Or you export everything to Excel.

After over 30 years of business intelligence, that has a certain comedy. But also a good reason.

In Excel the user can do with the data whatever they want. They can sort, filter, try things out, put two things side by side and follow an idea that has only just occurred to them. Because that is simply how analysis works. You don’t know the fifth question until you have seen the answer to the fourth.

Data is not individual tables

This is the part of business intelligence that has always interested me most.

Data is connected.

A customer orders a product. The order was handled by a salesperson. The customer is in Hamburg. The product belongs to a product group. The order was delivered in March and complained about in April.

Of course you can turn that into six tables and twelve reports. But you can also see it for what it is: a connection.

When I select Hamburg, I therefore don’t only want to see the revenue from Hamburg. I want to see at the same time which customers belong to it, which products these customers bought, which salespeople were involved, when it happened and which complaints are connected.

And then perhaps I also select a product group.

Everything changes again. Certain customers still belong. Others no longer do. Some salespeople remain. Certain months, too.

And suddenly I can move through the data. Not through reports. Through connections.

Association

The principle behind it is actually astonishingly simple.

The data is linked through its relationships. A selection doesn’t simply narrow down a report but changes the entire data space. You see what belongs to the selection, what is still possible and what has been excluded.

That means you don’t have to know beforehand which path an analysis will take.

I consider that a pretty important difference.

A report says: here is the answer to the question we planned for when building the report.

An associative system says: here is the data. Get started.

Of course you still need dashboards. At nine in the morning a managing director doesn’t want to spend half an hour exploring data just to find out what the revenue looks like. That is what key figures and standard reports are for.

But if something looks odd there, they should be able to get further. Without ordering a new report. Without SQL. And preferably without the by now traditional button: “Export to Excel”.

Because business intelligence shouldn’t consist of displaying as much data as possible as prettily as possible.

It should help you find something out. Sometimes even something nobody asked about beforehand.