Is it BI or HI that Intelligence has divorced?
I like weblogs from Vijay Vijayasankar. I really do. I am waiting for his every single blog, because in my opinion they are they only truly blogs on the BI topic on SDN. Vijay’s blogs always have interesting topics and have author’s thoughts. Not just another pseudo “How-to …” post or SDN/TechEd announcement, but really some thoughts! What I like as well about Vijay’s posts is that they provocative, and are causing discussions. Well, discussions (if merit 😉 are the only ways to discover or learn something.
The last one from Vijay is “And “Intelligence” said to BI – I want a divorce !“. When I first started reading it, my first though was that it is about IBM acquiring SPSS 🙂 But then I realized Vijay is discussing the issue of WHY-HAPPENED analysis, not yet WHAT-IS-GOING-TO-HAPPEN. Funny enough PA (Predictive Analytics) vendors are blaming BI vendors for answering just questions of WHAT and WHY, while Vijay is saying that even WHY is not answered by today’s BI! And I do agree with him.
It is just WHY analysis are not only dependent on BI tools, although they play a central role. On the input side of BI are all the source systems – the breadth and depth of information captured and loaded into BI for analysis. On the output side of the BI are we – humans – with our ability to digest and analyze information to produce some insights.
Way too often I see that we think that solution to the WHY question is in loading more information and in more detailed form into DW and then give business tools for Ad-hoc analysis. Let them do roll-up and drill down and drill across and slice and dice and find answers to all their questions themselves. Let them swim in this ocean of information without drowning in the big hope that they know where they are, where they are heading, where it is they are heading to and how to get there.
In my opinion it is HI, or Human Intelligence, where “I” is often missing. This statement might sound strong, but I do not mean here that we are brainless. No. What I mean is that way too often in real projects we are missing one important link: Business Analysts. Limited by budgets we assume too often, that business users are able to describe their world on an abstract level in mutually exclusive and collectively exhaustive style, and that How many skills does BW consultant need? will be able to understand these descriptions and produce something that not only supports business, but as well optimizes the way business operates. Sorry, no bonus. You need true BAs with strong analytical skills who can help business and IT during implementation, and then can apply their business knowledge to use BI/DW solution for proper analysis that support business operations and management decision making.
This whole discussion initiated by Vijay leads to another interesting topic. So far we were talking about BI in general, while the follow-up should be how it is related to BI/DW solutions from SAP and SAP BusinessObjects that all SDN/BPX memebers are working with. But this is the topic for a separate blog.
In general we have a tendency to jump into predictive analytics even before we do a good and complete job on regular BI. If we can't explain the facts that already happened - it is kind of hard to believe that we can some how predict the future with decent accuracy:)
Your words on HI is an interesting observation. I, for one, cannot make out useful information from a big report. My brain tunes out if I see a lot of data. I need a system that tells me useful information without bombarding me with details I cannot fathom.
The tool question, especially SAP tools, is a valid follow up - something we should revisit very soon. Maybe you will take a lead on this?
Thanks.
1) Good business analysts in the companies who are able to communicate and educate their co-workers.
2)A BI industry (vendors) that understand the above.
I hope that SAP and other vendors reads this especially the marketing departments that all too often praise their products to be the solution that will reveal the truth about business and the solutions to be quick wins. Let’s face it: business understanding is the key to almost any use of BI.
With Kind Regards
Kristian Appel
Good to see your blog. My take on PA is that no matter how much your BI matures, theres always going to be a gap that needs "Human Intelligence" to make adaptive predictions, we dont want machines to make decisions too(or do we). And as far as adhoc analysis i think its the most abused word from a BW/BI perspective, each person have a seperate definition of what an adhoc analysis is, for some its ability to slice and dice, for some its more to do with data effectiveness, for some its the speed of information etc... but very few talk about these things when it comes to BI...so great to see people like you, vijay etc blog about it.
I have seen less BI projects that focus on presenting the health status of the business and this is a SAP BI specific. SAP BI a long time from the beginning of this century as an answer to missing reporting capabilites of the ERP system itself.
Data warehousing in most cases is an expensive approach, that lead to wide spreading reports over the whole company, trust or whatever. So the focus is on reporting and not on, very technical, making use of the OLAP processer as a calculation engine with the goal to retrieve knowledge from the data stored ... this would be one step closer to a structured approch into direction analytical application or data mining.
We should not argue HI is missing - the way into direction of IB (intelligent businees) instead of BI (naked truth of real ERP data vs. plan data).
People are not missing the I they are trained to take care of the very detail of every single transaction. They are brainfxxxed from ERP itself. The game is simple, the one with the more correct number is the better and not the one with the better overview over the business when we come to discussions... the larger the company is the more this evaluated to true >>How an we trust a system that differs on the second place beyond the decimal point ... when we want to find out why we have lost 10% of sales << ... this is not missing I, this must come from somewhere else.
OR is dead so is BI if it is only defined via the argument of reporting. I personally miss intelligent implementations presenting the overall picture.
As long as people take care about the consistency of the single record we will not come one step into this direction and this is of course a consequence from SAP BIs flexible ERP reporting (which is good) but takes away ressources for more sophisticated applications that add business value ... like they always say - lot's of place for improvement left, especially when looking at BI consultants that know the technology itself and don't care about the customers business ... on this level they are in harmony with lots of business users and this is the missing "I", maybe.
Mike