Frequently asked questions about business intelligence
Frequently asked questions about business intelligence
A few questions come up with BI almost every time. And since pame, our own BI technology, joined the picture, a few new ones have come up.
Before talking about dashboards, a few things should be settled
Business intelligence often sounds simpler at first glance than it is in practice.
Almost every company would like better numbers, clearer key figures and less gut feeling. But the topic rarely starts with visualisation. Usually it starts with data scattered across five systems, with key figures that everyone understands a little differently, and with the question of whether you actually need a report or a way to really find something out.
That is exactly why the most important BI questions are often not technical but about the business. What is this actually about? Which decision should end up better? Who may see what? And how much of it can classic BI solve before you start writing AI on everything in a hurry?
And now there is a second layer: do you work with existing platforms such as Qlik? Do you build something of your own? Or is pame perhaps the more exciting direction for the specific case, because analysis, data model and our own BI technology come together there?
Here are the questions we hear most often. And the answers are sometimes a little less promotional than you would usually expect from an FAQ.
Questions and answers
What is business intelligence, actually?
Business intelligence is the attempt to make a picture you can trust out of distributed data. It is about making connections visible and basing decisions on more than gut feeling, not just presenting numbers prettily.
Is BI just a dashboard?
No. A dashboard is usually only the surface. The real work lies underneath: connecting data sources, clarifying terms, defining key figures cleanly, setting permissions and building a model in which the numbers really fit together. Otherwise you just have a good-looking misunderstanding.
When does Qlik make sense and when pame?
That depends on whether you want to work with an established platform or move closer to a BI technology of your own. Qlik makes sense when you are looking for a strong existing platform with a ready-made ecosystem. pame becomes interesting when associative analysis, a technological foundation of your own and more control over the model are meant to come together. It is not Qlik against pame, but about which tool fits which goal.
How long does a BI project take?
That depends on where the real problem lies. If the data is reasonably clean and there is a clear use case, you can see something useful in a few weeks. If key figures, data sources and responsibilities first have to be sorted out, it takes longer. The technical part is often not the hardest.
What does something like this cost?
That also depends less on the dashboard than on the foundation. Three pretty reports are quickly built. A reliable model with clean key figures, roles, data logic and sensible extensibility is something else. That is why we don’t like to bill BI by the number of charts. First it has to be clear what is actually supposed to be created.
Can existing data sources be connected?
In most cases, yes. Databases, ERP, CRM, Excel files, CSV exports, APIs or hospital systems are not an unusual problem to begin with. The more interesting question is usually not whether you can load the data technically, but whether you still understand afterwards what it actually means for the business.
Do we need a data warehouse for that?
Not automatically. Sometimes a lean setup is perfectly enough, sometimes a data warehouse or something comparable makes a lot of sense, for example for historisation, quality assurance or more complex models. Anyone who always says yes or always says no across the board usually makes it a little too easy for themselves.
How important is data quality really?
Unfortunately, quite important. BI can’t work magic with unclear or contradictory data. And neither can AI, by the way. If revenue means three different things in three systems, you may end up with a quick answer, but not yet a reliable one.
What exactly is pame in this context?
For us, pame is not another dashboard tool but an attempt to rethink BI technologically. We don’t want to put up finished reports, but to build a system in which data, relationships, analysis and, later, AI work on the same model. That is why pame now comes up quite naturally in our BI conversations.
What does AI change about BI?
The interface will change. People will ask data questions more often instead of just opening dashboards. But exactly because of that, clean models, clear terms and reliable permissions become more important, not less. An AI without a business foundation is above all one thing: eloquent.
Can employees adapt dashboards or analyses themselves?
Yes, if the system is built for it and the roles are set cleanly. Some users should only read. Others should filter, build their own views or analyse more deeply. That is often where you see whether a BI system is really meant as a working tool or only as a presentation surface.
How do you set up permissions sensibly?
Not with nicely hidden menu items. Permissions have to be attached to the data itself. Who may see which region, which customers, which HR data or which health data has to be defined in the model. And the same rules must apply when an AI later asks questions of this data.
Can we start with a pilot project?
Yes, and it is usually even sensible. A good pilot is not a toy but a real use case with a real question. Then you see fairly quickly whether the topic works in your own company or whether quite different things have to be settled first.