11 Replies Latest reply on Aug 24, 2012 4:55 PM by tzeng

    The many mysteries of BOBJ...


      Curious to know what other people are measuring and any challenges therein as BOBJ seems a mysterious beast at times. One example is a simple enough, at least at first blush, report on the most downloaded content. Getting such a report created showed some strange results - a document that had only 34 views had apparently been downloaded hundred of times for example. What we're now working on as a way around this is to count not the number of times that a document has been downloaded (BOBJ definition of DOWNLOAD appears to be very wide and counts more than just downloading a file), but to count how many people, i.e. distinct users, who have downloaded files and to then rank those documents in descending order.


      Similarly, trying to create a report showing the who was uploading the most content created charts with massive spikes for some users - turns out that BOBJ double counts photos uploaded to albums by adding the count of thumbnails too, so we had to filter them out (and we may even need to filter out photos too, to create a more level playing field of results).


      Wondering if others have come across other such conundrums and how they've overcome them?

        • Re: The many mysteries of BOBJ...

          Russell, you have dug into and identified two issues I have not discovered, but I have encountered a number of similar, baffling and frustrating odd results returned from BOBJ. I do not know enough to say whether it is the tool's fault or Jive's fault for building a flawed data warehouse -- or both.


          Perhaps, over time, with lots of people sharing in here we could uncover and solve the mysteries of BOBJ, as you and your team aptly have begun to do. Personally, I resigned myself years ago to not getting the kinds of information that I want which BOBJ can't readily return. I have neither the specialized skills nor the time to thoroughly investigate and resolve what are clearly inaccurate reports.


          For example, one that comes to mind is the Connections report that Jive supplies in the Jive Reports folder. We have no more than 6500 employees, yet many months show two or three times as many people as having "made connections."


          I'll get on my soapbox again: Jive needs to get the data right, and that's the first and most important step. If the data warehouse needs to be blown up and re-designed to accomplish this, then they should do it.

            • Re: The many mysteries of BOBJ...

              Hi Ted


              By measuring activities only, it's not a really data warehouse, it's more an activity counting house...


              In many ways BOBJ is conceptually flawed in the way that it measure only activities - so for example, as you probably know, it can't tell you how many Followers a person has, it can only record how many times the action of Follow has happened, when mapped to the distinct user. This make it very vulnerable to any break in transmission or data saving - it's like it's counting in its head. It also over counts - it seems know one has looked in detail at exactly what constitutes an Activity and what specific measures it will trigger, (at least none of that is documented). This makes it very difficult to get data that matches what you see in the site - the 1st check users do, esp if its about them. It also makes it as you've pointed out, a right royal pain in the proverbial.



            • Re: The many mysteries of BOBJ...

              Russell, thank you for starting this thread. We have the same challenge with getting a report for downloaded content. We would also like to see how many times an attachment has been downloaded. I don't think there is a way for us to do this using BOBJ, and we are not reporting on it via other means.


              The other report I tried for in the past was 'who has the most number of status points'? There is 'Top participants' but not actual number of points.