Editor's note: DMReview.com would like to welcome Gabriel Fuchs as an online columnist. He will write a bimonthly column entitled Reality IT in which he will address IT industry issues from a collective experience viewpoint, using common sense and a creative imagination. You'll want to read what he has to say and will probably want to engage him in further discussion. Your feedback is welcome.
At my job, we have decided to implement an operational data store (ODS) so that we can get real-time business intelligence (BI). Don't ask me why. Few of our decisions are made without proper reflection - or in real time. Also, if our real-time BI will tell us, for example, that we have not sold product X in the last ten minutes, we are unlikely to change our strategy or even call all our salespeople and tell them to get moving. Actually, real-time business intelligence will probably mean more IT resources in order to do the same reports and analyses as when our BI applications were updated every night at the most.
Oh yes, real-time BI has become the word of the day. Never mind that very few companies may actually need it. Today, the absolute majority of BI analyses are delivered weekly or monthly, with some quarterly or daily. Rarely minutely. When we do need data in real time, for production as an example, we do not call it BI. We simply consider it to be part of our operational activities. And once again, not even all our operational activities need to be real time. In some cases, we update our transactional systems with well-defined batch routines that run every night.
Personally, I believe that the minute we talk about real-time BI, it is not BI anymore. Rather, it has become part of operational activities, no matter where the data comes from. Whether it is an operational system or a data warehouse does not matter. In the case of a data warehouse - or rather an ODS - it has simply become an aggregated level of operations. The resulting data output is nevertheless most likely to be fed back into operational activities according to predefined processes.
Few of our older BI applications have defined processes or routines that create analyses as well as knowledge on how to act on the results. Instead, we gather and discuss these results and the appropriate actions to take. The discussions are done when needed in order to make so-called intelligent and informed decisions. If we already know what needs to be done when certain specific events occur, we automate these activities. I would say that they then become part of our operational process-based activities instead of being real-time BI.
As there are few standard solutions that can handle so called real-time BI, we are implementing a lot on our own. And funnily enough, most of the actual development is handled by our programmers who usually work with our operational systems. They are the ones who know how to query, extract and feed real-time data into different systems. For them, it does not matter if it is called operational activity or real-time BI. It all comes down to defining processes and systems that are performing well enough to handle it all.
Let's not forget that according to several surveys, up to 70 percent of data warehouse projects fail to produce the expected results. We have difficulties mastering projects for our classic BI solutions, i.e., it is already a challenge to create useful BI applications that add some serious value to our business. Now we hope that all this will be solved with something even more complex. Of course, we have no idea on how to calculate ROI on a real-time BI application. It is all so new and, therefore, we have little experience, meaning that any ROI estimation becomes difficult. What do we do? Well, real-time BI sounds good, right? It can never be wrong to have everything straight away. If we can achieve this (a big "if" here), we should automatically get a positive ROI. We think.
In any case, which is cooler: saying that we have a good operational system or that we have real-time BI? Actually, come to think about it, all our operational systems can be said to be real-time BI. We, or rather our automized systems, act on data and information according to predefined processes. Yes, we are actually more advanced than anyone else, as we have real-time or near real-time data in most of our systems. Whether it is intelligent information or not is always a matter of opinion.
It is only our older, existing BI systems that allow us to get non-predefined analyses and data extractions for our reports. These activities are not real time, and this is good, as we need to reflect on these reports and analyses before acting. Real time would be a bit too fast. We already feel stressed when it comes to acting, and too many decisions seem to be made without enough thought. Thinking slowly seems to be difficult for many people; thinking in real time even more so; giving the impression of knowing why one is acting in a certain way is even more of a challenge. Not to mention the follow up on the actions made. And all this should supposedly be done in real time, or even better, in no time. Oh yes, "no-time BI" - the next big thing once we have failed with our other BI applications!
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access