Every one and his neighbor wants self service BI. What does it take to get it? Is it even reasonable to expect that significant parts of BI can be self service enabled?
For one, self service is loosely defined, at least in the BI world.
First – no enterprise BI tool is really designed with the user in mind from my limited experience. Invariably, they are designed with a developer in mind. This needs to change before self service becomes realistic. All BI vendors claim they are about self service, but if you look at the tools – it is hard to imagine them being people centric in design. If users need training to make this work – it generally won’t fly. Simple as that.
Next is the distinction of BI users as defined by software vendors to suit the products they have. It is chicken and egg – over time, one cannot figure out if different tools exist because different types of users/usage exists, OR different users/usage exist because BI tools are not consistent. Either way – any real self service is hard to pull off with such distinctions.
Ability to search is next. Organizations do not lack data – it is usually there somewhere. What is not possible is an easy way to find all of this data that might be sitting somewhere. So there needs to be a way to find this data in a form that can be used. What is possible today is to search for names of existing reports and column names. But since they are not usually well-defined – making sense of the search results is generally a complex task .
It appears to be obvious that a consistent data definition – or a common semantic layer in hip terms – is required to make this work. Tools can do this today to various degrees – but the process to get it done is nowhere near easy. Try asking 20 people in a big company to define “net profit”, and you will get an idea of what I am talking about.
Users won’t trust data they see unless it shows how current it is. Some of the users will also need to make sure they can see how the data flowed from its origin to the report they have in front of them, and what transformations happened en route. This is not a self service BI challenge – this is true in any BI scenario, but just that in self service this will become an even bigger issue.
Finally there is the question of support. When things don’t work – there should be an easy way for users to get support – via technology or via a human support person. Without integrating support in some comprehensive fashion, it is hard to sustain self service BI.
If you still want Self service BI, let’s go make it work. I am always game ! What are a few challenges between friends 🙂
In order to make it a selfservice BI, more efforts are to be made on training end users than on developing the BI tool. Because no matter how easy the tool is, users should be willing make an extra right click and see the available options, use basic drag & drop functionalities…etc
LikeLike