Definition: Martini BI – anytime, anyplace, anywhere BI.
I’ll let you into a secret. Yours is not the only organization struggling with business intelligence – whatever that term is supposed to mean. Over a decade ago I hosted an annual BI conference is Amsterdam. It attracted delegates from most of Europe. After the supplier presentations and the general ra-ra, delegates would talk about their problems. These were usually connected with data ownership, the cost of resources, data quality, performance, arcane technology – and so on. And guess what – nothing has changed. Someone who is still involved with a similar conference reports exactly the same problems. Just a moment’s reflection would reveal that if everything was rosy in the BI garden there would be no need for yet another generation of ‘agile, visual, easy-to-use’ tools – or whatever the ‘in’ words are.
Many BI problems stem from the organization itself – an unwillingness to share data being the most common. We then have technical problems associated with actually accessing data. Most large organizations have dozens of applications, maybe hundreds, often passing data via some hotch-potch of protocols, and accessing the data they generate is next to impossible. Of course the suppliers of BI technology overlook these ‘details’ and offer connectors to SalesForce, Oracle, CSV files – but not to that monster sat over there in the corner using a file format that is undocumented and someone’s idea of a bad joke. IT isn’t all nice and neat like we see in supplier brochures.
Then we have the question of what we want to do with BI. Is it OK for someone in accounts to run a BI query that accesses 2 TB of data and takes 48 hours to run – because if it isn’t they are going to be unhappy. Who acts as the BI police force – censoring what is essential from what isn’t?
The suppliers of BI technology are aware of all these issues. They tend to skip the delicate ones like ‘why won’t you share data?’, but offer solutions to the technical problems with in-memory processing, massively parallel processing etc etc. Not much use if you can’t get to the data in the first place.
So here is the solution – the difficult stuff that organizations prefer to ignore. Determine who owns what data. Determine who can use the data. Have a clear understanding of the data that can be reached and the data that cannot. Make someone responsible for quality of each data source – and give them a veto if it isn’t up to scratch. Above all set realistic expectations – anytime, anyplace, anywhere BI is not realistic despite the claims of suppliers. Martini BI is a dream and there are always plenty of suppliers happy to claim they can make it a reality. Once these issues have been addressed then maybe BI can deliver what it is meant to – organization wide views of activity in a well ordered environment.