Q&A with BI Expert Penny Silvia on Self-Service Analytics (transcript)
I recently moderated a web forum with BI expert and IBM consultant Penny Silvia on building sound strategies for self service analytics.
For the full Q&A, you can view the questions from Insider Learning Network members and Penny’s responses in the BI Forum, or read excerpts from the transcript of the Q&A, below.
Avoid the common pitfalls of self-service analytics:
An exclusive Q&A today with BI expert Penny Silvia on building a sound strategy for self-services
Kristin Bent (Moderator):
Welcome to today’s forum on BI and self-service analytics with Penny Silvia
Penny, thank you for joining us today! Before you respond to questions, I'd like to start with one that stems from your session at the most recent Reporting and Analytics conference:
In general, is there a particular type of information that lends itself to self-service reporting and, conversely, is there a particular type of information that is not appropriate for self-service reporting?
Thank you, Kristin! To your question:
Self-service definitions and what will versus won't work - at least to me - is more based around the user and their requirement more than the business area. For instance, providing a user a list of reports they can execute (rather than having an automated email generated with the results of all of those reports) would be considered self-service to a casual user but would be consider highly restricted to a power user.
For our attendees ... do your organizations allow self-service for analytics within your environments? And if so, to what level or degree?
Our organization is trying to encourage self-service by empowering users with the required tools and knowledge. It is very difficult as we have nearly a thousand BI reports and just getting an accurate inventory of those reports and what they represent and the business processes that they support is challenging, but that is one of our starting points.
We are pretty wide open on letting users create there own queries but it has come down to about a dozen users who actually write their own and everyone else just uses those reports.
This is quite common, Elizabeth. There will be a natural evolution of those people who are more comfortable creating and then optimizing queries. For those people who do create queries within your organization - can they also create calculations?
We do offer different types of self-service analytics, all in WEBI, of which the most common are:
- Giving our users lots of input controls to play with
- Letting a select group of users query on the universes
I also recently
built a report in which the user can drag and drop usage metrics into the report. Our limitation right now is that our end users have limited knowledge of Business Objects, but we are starting training now.
Daniel - this is a great approach. WebI is a good tool that is comfortable for a cross-section of users.
When you train your users are you going to train them on how to create WebI reports for themselves? Will you also train then on how to develop Universes or are those centrally developed?
A bit part of any successful Self-Service program is making sure all of the data people will need is accessible and easily understood. Not always a simple matter, of course.
Penny - The users will be trained primarily on working within a WEBI report which has been pre-defined. A subset will be trained in how to use specific universes. We will not train users to develop their own universes, which will be our role within the BICC.
Sounds like a solid approach, Daniel. This gives a degree of self-service but within a guided (and guarded) area.
I think there is a common mistrust between IT and the business users. In our case, IT insists that it controls the data and that only IT knows the 'right' way to analyze it. So they don't aid the users in building agile solutions, let alone providing data discovery.
Danny - that issue of trust between business and IT will create challenges not just in a self-service program but in any technology effort, as you know. It's easy for me to say but there needs to be a culture of respect (of what each 'side' offers the other) and shared work in order to be successful. This lack of trust will hamper the organizations ability to take advantage of the tools and the potential of
discovery that these tools enable.
I understand that, but since I am only working as a service provider for the IT department it is difficult to change this from within. I suppose that is generally true if you work for a consultancy and you start "overstepping" your statement of work.
Danny - I know it can be tough to try to be an advocate for change within an organization. I'm not understating that effort at all. While you are with the IT department, I would say you could start small - identify key business users who are technically savvy and maybe work on small projects with them. Hopefully once others see how effective and beneficial the relationship can be they will be more open to it themselves.
In our case, we have the luxury of being in a BI Competence Center, which started June 1st. We are the bridge between IT and the business (in fact, we are sitting together with the business), and when it comes to analytics it's all in our own hands without interference from IT, and we have a good sense on what the business needs. IT just handles the data warehouse stuff.
Daniel - this is a GREAT approach. A BICC can be an excellent bridge - and translation engine - between these diverse groups.
It's difficult here as the IT is central and the business is so diverse that it would be a challenge to get them all on one table to set up said BICC in the first place. Combine that with a "strong" IT department and business driven innovation loses against IT driven innovation.
Yes, I definitely know how difficult it is. We were lucky actually that we started out as analysts for a specific Customer Base Management department, because the manager then didn't want to rely on IT (alt
hough we ended up doing reports for the entire marketing department). So when I started I got into the situation that I was actually part of the business and were able to do a lot of experimentation without necessarily having to wait for approval first, and this happened in various parts of the business.
The disadvantage was that we created massive inefficiencies and multiple versions of the truth. However, the advantage is that each separate team gathered a lot of business knowledge in their specific areas.
I realize such an approach is probably not realistic in many other organizations concerning the structure, but that's more or less how it evolved for us.
I think the critical part with the success of "Agile BI" and self service is to wrap the results into a change management process. If you have users creating their own queries and nobody follows up on that, you have a growing mountain of queries, nobody understands or uses. You may have a few gems in there, but they are difficult to identify.
I think the feedback process from the business back to IT to integrate the created reports and definitions into the growing corporate repository is the dark art in self service. Otherwise I miss the continuous improvement of the existing solution and processes.
Danny - this is a great point! Change Management is key to this type of information evolution. It doesn't do an organization any good for everyone to be running around creating their own version of the truth. Change Mgmt will help direct people's activities and thought processes by socializing not just the "how" do something but also "what" is considered acceptable and why. Socializing the accepted metrics and criteria will help everyone focus their attentions.
Feedback and coordination between business and IT is indeed an art (hopefully not too dark) and has involv
ed many things in my experience - forums, CRPs, pizza & beer and more. These solutions we are discussing are meant to enable the business - and if the solution isn't dynamic and able to adapt to business changes then it will become stale and ineffective. Communication and feedback is the key to keeping that line of conversation - and development - open.
So, then the 1-million dollar question is how to you get the business to feed their "creations" back into a coordinated process (possibly BICC), in which these creations get sanctioned by corporate and implemented in the stable EDW so that all users can benefit from these analysis and drive the business forward.
Today, change management is mainly a barrier build around your IT systems to control what goes in. But in a more agile scenario you need to encourage the business to put time and effort into the problem, even if they have already solved it themselves.
Do you know of any way to encourage the business community to contribute to the development and adoption of best practices beyond their department? I would think this is difficult to drive it out of IT and it would need the right incentives for the business to invest in the advancement of the corporation as a whole.
Danny - some of this will boil down to whether you are enabling Self-Service for Ad Hoc. What I mean by that ... will your organization be allowing creation of queries/reports in Production - the ultimate in self-service?
If so then the business users who are identified for this program - and it should realistically be MAYBE 10% of the base - would be given a set of structures for what is created ad hoc and what gets created ad hoc but then is determined to be 'corporate quality' (for instance) and is brought back down to a Dev system to work jointly with IT to optimize.
One way I've see
n the business engage in this contribution and back-and-forth is to show them that IT can add value to what they've built by using technology to make it stronger and faster (the Six Million Dollar Query - man, I'm showing my age). IT can optimize the cubes, indexes, aggregates that will make the business' queries run faster and therefore make that business user look better to their peers and bosses. Always a good trick!
Excellent point about the added value of an IT supported implementation.
However, I would push the envelope even further than the 10% of the user base. I believe there are people out there who may not be in that group because they just joined or don't bother with requesting access, but they may have great ideas. These resources need to be tapped as well.
So, here is my suggestion: Collaborative BI. Create something like a social network around your reports and (somehow...) monitor activity in that network. Create incentives for people to share their analysis and then look at the reports that run often or by many people. If you selected these reports talk to the creators and involve them in an implementation.
Problem 1 is that you need to create a culture of sharing (and come competitive strive for getting all information out of your data) and problem 2 is to create the tools to support you getting all the information you need to find the analytical gems in your company.
Or maybe I am just delusional again... :)
Hi Penny - Is the reporting tool dictated by type of user such as executive or power user who will be using the report?
Hi Laura! - The reporting tool is really more a matter of what the requirement is than who the user is - although it may seem to go hand in hand. Executive users, for instance, tend to want summarized information and key metr
ics that are organized in dashboards. That tool would be Xcelsius. Casual users or information consumers tend to want list based reports - rather than dashboards or analytics - and that tool would be Crystal. That is not to say that executives would not get Crystal reports, however. It really is based on the information need.
I try to bring down to four points:
1. The Right Information
2. In the Right Form
3. To the Right Person
4. At the Right Time
How would you address the problem around data training? Most users can be trained relatively easy to understand and use the tool, but when it comes to making sense of the data it would usually fail.
My problem is really with the business users not being able to interpret their own data. How would you address that problem?
Danny - this is a tough one. Data Mining is an advanced technique and process. I would first ask you why your organization is looking at data mining and what are you trying to achieve or enable - and is data mining, in this case, really just giving people access to the full data set so they can create their own queries?
If you have business users who cannot interpret their data I would question if they are the right people to be using such advanced tools. The matter of understanding that field XYZ actually means "Material Cost" - or something like that - should be an easy enough matter. That is just understanding data elements. The MEANING of data is really something that the business should be best equipped to understand. If those business people can extract the data but can't interpret the results ... I would say you might want to look within a different user community - maybe analysts or the people who are
doing all the data collection and consolidation today.
With self service I was actually focusing on the BO Explorer - I suppose I am now a victim of SAP marketing :) But you are right; other tools can be used too, with a more defined navigation path (WebI or building custom queries in BEx).
Danny - I wouldn't say you are a VICTIM (yet). :-) Explorer offers an interesting set of capabilities but it is not as 'google-like' as some might want it to be. There still needs to be a good understanding of data and relationships. You might want to start your users on something that is more guided until they become more familiar with the data. Still a ton of capabilities for them!
Also, you would have to have the BWA in place for Explorer - I don't know if you have that in place already. Then it's a decision of what data to load into it and things can get more complication (for some users) when you start bringing in non-SAP data directly into BWA.
I played around with BO Explorer on HANA early this week. It is a very interesting perspective, if you have a client who can afford to buy HANA and has a business case to use it on. But the main problem remains - the users have to be damn well educated to slice and dice through their data on that level.
Danny - you are correct on all fronts on this one. HANA and Explorer are not for the light of heart right now. It is a GA tool but still a 1.0 release and you need very savvy users right now who know how to be a little creative and know very well how to manage their data. HANA will not change the nature of the data - it will "just" give you the opportunity to look at billions of pieces of it all at once. Better know what you are diving into before you tackle THAT much info.
Hi Penny -- Ken Kiefer i
ntroduced us at the SCM 2011 conference. How are you?
I have two questions: Are there any technical prerequisites for using the SAP self-service reporting tools? And are there any particular products that must be in place or are there any products that must be at a particular release level? Thanks
Hi Scott! - Hmmm ... to your first question on technical prerequisites. This will really come down to making sure that however it is that YOUR organization defines self-service has those tools in place along with the appropriate authorizations for people to be able to execute.
For products at certain levels ... not really. Any tool can be considered part of a self-service strategy if it meets the goals of that organization. Any of the current products and/or releases would fit a need - some may just be more user friendly than others based on the use case.
Penny - A question about tooling.
Currently we are using WEBI for probably 90% of our activities, including our standard reports (although we want to start using Xcelsius more). However, I heard multiple times throughout the BI2011 conference that WEBI is primarily an ad hoc tool. Are there other tools out there which you would recommend to set up interactive and detailed standard reports?
Daniel - WebI is indeed a good Ad Hoc tool - but Ad Hoc for 'light' users. What I mean by that is that it is not a full-blown OLAP tool. If you need fully interactive reports that allow for multi-dimensional analysis then you want to look at ye ole standby BEx or the new Analysis (formerly known as Pioneer). Analysis is the full-blown OLAP tool that will also deliver complete Ad Hoc capabilities should you so desire.
For detailed standard reports I might direct you to Crystal - depending
on the nature of the report and how it will be used.
Thanks, I heard a lot about BEx and Analysis during the conference as well. However, I never heard of these tools before, so are there any resources you could recommend to get a better grasp for the functionalities and whether it would suit our needs?
Is there a general rule of thumb in determining which reports should continue to be created in BEx Analyzer and which should be created in BusinessObjects Web Intelligence?
Oh Bette! Such a loaded question you ask! :-)
BEX is an OLAP tool - full blown multi-dimensional analysis, slice & dice, switch and swap 'em capabilities.
WebI is an Ad Hoc analysis tool, not an OLAP tool. If you are 'simply' looking for drill-down (rather than swap this with that and change axes, etc) then WebI is a good and easy (easier) option.
I would also say it depends on how much data we are talking about. WebI does have some issues with large returns of data sets.
Hi Penny - Danny raises some interesting questions in the area of business users and best practices. Do you have any broad-brush advice for accelerating adoption of new BI tools by the business side? Do you recommend a big-bang push for all tools or a phased approach with maybe standard reports first, followed by dashboards, etc.?
Definitely a 'start small' approach, I would say. Find people within the organization who are willing to experiment and do some test cases with them. If all goes well, you will be able to prove the added value to a wider group of people. In my experience, the business is easier to convince if you have something tangible to show, as most of them will have trouble grasping what you mean when you talk mainl
y in concepts and ideas.
On top of that, you'll get a better idea on how to roll out new tools because you've done it before, thereby minimizing the risk you have.
Hi Davin! - Big Bang can be a little scary to business users - too many choices! Overload!!! :-)
I would roll this out in a phased approach - either by business area or by tool. And deliver a complete solution during the phase (even if it's with multiple tools) so they can see quickly how each part of the solution addresses a need for them.
IF you are going Big Bang I would say you want to do it with BO4 - as it has a much more consistent look, feel and toolset across the tools. It will make it easier for training.
Thanks to all who posted questions and followed the discussion!
A full summary of all the questions will be available here in the BI/BW Forum and in the BI Group and the Reporting and Analytics conference group on Insider Learning Network. I encourage you to join these groups for ongoing information and additional resources.
Lastly, we’d like to wish our BI expert, Penny Silvia, a happy birthday! Thanks for taking the time to chat with us on today!
Thank you, Kristin for this opportunity! And for the birthday wishes!
As a reminder, Penny is a featured speaker at numerous SAPinsider conferences, including this fall’s upcoming Reporting & Analytics conference. More details about this annual conference and ongoing updates about the sessions that will be presented are available here.