Old BI and the Challenge of Analytics
I recently spoke at an event in Melbourne, Australia that was sponsored by Innogence, a consultancy that exclusively supports SAP business intelligence implementations. My presentation was followed by two others: one of Innogence’s customers described his experience of turning around an SAP Business Objects (hereafter referred to simply as SAP) implementation that was failing, and an employee of SAP previewed coming attractions.
The fellow from SAP didn’t describe anything new in the realm of data visualization, but referred to Crystal Xcelsius (recently re-branded into two products, SAP Crystal’s Dashboard Design and Presentation Design) and Business Objects Explorer as SAP’s current datavis offerings. Both of these products demonstrate SAP’s lack of expertise in this area. What he featured most was a new in-memory technology called HANA, which promises to speed up access to data considerably, but at significant cost. While he was describing HANA, I couldn’t help but fear that it might at best enable a faster trip to nowhere.
Big, old, traditional BI companies are good at producing technologies that enhance the infrastructure of business intelligence—more and faster—but not the actual use of data in ways that lead to greater intelligence. Being big, focused primarily on technology from an engineering perspective, and devoutly sales driven makes it difficult for companies like SAP to develop useful tools for activities that support decision making: data exploration, sensemaking, and communication. To meet this challenge, they must shift their focus from technology to the humans who use it—our needs and abilities—and expand their perspective to embrace design. They must commit their efforts to what actually works, rather than silly, shiny features that fill their existing products with smoke and mirrors.
It will be hard for large organizations to turn their ships against the tide of tradition into unfamiliar waters. If it can be done at all, it will take time. Will SAP and other big vendors find their way into the analytics age? If so, will they do it in time, or will analytics become the exclusive realm of smaller and more agile vendors, leaving traditional BI companies in the back room to maintain the infrastructure (data collection, transformation, cleansing, warehousing, and production reporting)? Only time will tell, but I recently received a glimmer of hope that SAP just might discover the path forward. I had lunch three weeks ago with John Armitage, who has been tasked with improving the user experience of SAP’s information visualization offerings across the board. I’ve known John for a few years, and although I don’t know him well, I believe he has the perspective, experience, and skills that are needed to give this big ship a new bearing. John is a designer with years of experience working in the field of usability. He’s one of the guys inside the company who quietly thank me when I point out flaws in SAP’s data visualization capabilities and challenge them to do better.
If big, old, traditional BI companies are going to find their way from the back room to the living room and kitchen where people live, they’ll need people like John Armitage, and they’ll need to listen to them. John’s role can’t be token; it must help drive the organization.
Despite the acid that often drips from my tongue when I speak out against SAP’s contrived and clueless attempts so far to support data visualization, I’m not rooting for them to fail. That is, unless they continue to promote hollow promises about dysfunctional products, in which case they’ll deserve to fail and I’ll gladly bid them goodbye.
Take care,
10 Comments on “Old BI and the Challenge of Analytics”
Steve,
great insights. I am coming at this issue from years of “institutional research” and program evaluation, where we applied data to various organizational questions. It has always been about as you say, the living room where people live. I have been trying to put myself professionally at this intersection of BI tools with client needs for BI they can use. It’s not been fruitful. The IT firms all approach this from an IT perspective, wanting IT people. I think to play that role you mention, IT needs to open the gate and let a few of us non IT folks into the party, because we have lived in those decision support trenches since the time IT was struggle to have stable networks. Thoughts, advice? Many thanks, Michael
Michael,
As you suggest, BI companies have traditionally been primarily aligned with IT departments. They usually deal with IT departments when selling their products. This is appropriate for products that handle the back-end BI infrastructure, but not for data exploration, sensemaking, and presentation products. People who actually work with data to support better decisions are the ones who are in the best position to understand analytics. Most BI companies, unfortunately, don’t fluently speak the language of analytics. They throw around terms like business analytics, data mining, data visualization, performance management, etc., without any real experience in doing these things. As a result, they don’t understand them. This lack of understanding produces dysfunctional decision support products, which is perhaps the primary reason why business intelligence has never come close to fulfilling its promise. Frustration with the failures of BI will soon reach a tipping point. It already has for many organizations that have looked beyond traditional BI companies to products such as Tableau, Spotfire, and SAS JMP for solutions. As more and more people who work with data become exposed to better tools like these, the tide will turn. It’s only a matter of time. In the meantime, those who understand analytics should show the world how much better we can wring value from our data when equipped with the right skills and tools.
I think we can compare it to a sales person trying to sell a car… he/she will probably find plenty of reasons to buy a certain car but trying to sell a car to a racing driver who is after a “good car” is a different story; the sales person will have to find really the right car “compatible” with the racing driver otherwise they will fail miserably.
Unfortunatelly businesses usually buy cars for their racing drives so they usually agree with sales person more then with the driver ;)
Regards
Emil
Organisationally, I think part of the challenge is understanding where the responsibilities of IT (should) finish and the responsibilities of the business (should) start. I’m engaged in several discussions with clients at the moment over master data, data quality and so on, where IT has traditionally been handed these problems as if they were technology issues whereas in fact the relevant business SME needs to be empowered (and charged with responsibility) to look after these kinds of information assets.
I see a parallel with the whole area of “back end” data and infrastructure (the DW, the ETL milieu, the core data model) and the front end – the kind of apps that allow for exploration, analysis, sensemaking. In my view a DW and BI initiative should always be business driven, with clear business requirements (questions to be answered; subjects and hypotheses to be analysed) defined from day 1. However, IT has to actually build most of this (hence the ongoing conversation becomes very important to the eventual success of the initiative) and then support it. At some point – somewhere between data mart data modelling, the BI semantic layer, or the layout and design of multiple front-end tools – responsibility for iterative design and improvement needs to pass to the business, with IT continuing to play the appropriate supporting role. Then, if the design skills don’t exist in the business SME camp, the relevant business information customer needs to engage resources to assist – perhaps from a kind of BICC pool of talent (depending what kind of BICC model is operative). This team should have a swag of tools available, from visualisation tools to hard-core analytics, so that the right tool can be provided for what the business is trying to achieve; as you point out, it’s the niche players who seem to be scratching where it’s itchy here, so IT departments may need to reconcile themselves (most already have anyway although they may not like it) with supporting a range of apps for different kinds of visualisation and analysis.
There are many layers of questions that this throws up which deserve exploration in their own right; I guess the point I’m making is that getting the org structure and roles & responsibilities right is another key factor in actually getting the right result at the front end.
in my experience, part of the reason BI fails is that BI in a corporation often does not have a proper business model and a realistic measurement of system benefits. Business people see BI tools as threat to their job security, providing little if any useful support; IT struggles over data integrity and completeness and never quite present a reliable replacement to manual processes. BI vendors over-promise BI benefits in a corporate culture that’s not quite ready to integrate it. In order to sell BI, they resort to flashy charts, faster access and processing power. perhaps the basic business needs for BI are not been addressed, even before we may care about data visualizations. However I do believe in choosing the right visualizations to make impact in presentations for change.
Great post Steve. Thank you for demonstrating the importance of data and how it is used in the BI environment. Isn’t that what it’s all about anyway? We have a community for IM professionals and have bookmarked this post for our users. Look forward to reading your work in the future.
Great post Steve. I would have been very interested to hear you speak at that event in Melbourne. Did you focus at all on implementing your best practices for visualisation in the SAP products or did you steer clear?
David,
As I was preparing the presentation, I weighed the value of a direct assault on SAP’s poor data visualization practices vs. letting this fact take shape in the minds of the audience on its own from the content that I presented. I chose the latter, but was prepared to address SAP’s current poor data visualization quality directly in response to audience questions and comments following my presentation, if the occasion arose (it did not). I think my choice was wise in this case.
Hi Steve,
Thanks for sharing your experience. I have pondered the question of Business Objects based on multiple client and student questions and comments. I ultimately think that it would require a “transformational” event for a company like SAP to address the needs of the average analyst versus the lengthy BI “checklists” that often drive tool selection.
More disturbing was a simple analysis I performed based on input from these interactions. The average spend was almost $12,000 per person (first year fees) based on actual users of Business Objects. While they may sell 500 seats at a company, perhaps no more than 50-100 are utilized at most companies I have encountered.
Thinking of even basic training in the principles you so eloquently write about and teach combined with Excel ($100 per seat, already in most companies) would appear to deliver more value in many of my client reviews. Even better, add your training, Tableau and a well-designed data warehouse built around key subject matter needs at $2,000 average cost per user for a phenomenal change.
This would leave plenty of money to eventually enhance and enrich the data with the work of a central team of expert analysts who could add forecasting, data mining and other advanced analytic techniques to the overall analytics culture. Advanced tools from SAS, JMP, R and SPSS could fill the bill for this area.
Business Objects and many other traditional BI vendors are stuck in the centralized rut; I have yet to see strong evidence to the contrary. However, based on recent marketing from these companies, I suspect that many realize the threat of empowered end-users to their traditional architecture and mode of operation.
Best,
Stephen
Hi Stephen,
We who work closely with people who struggle to use traditional BI tools successfully in their work know all too well the high costs and deep pain associated with their use. I, like you, have noticed that the big BI vendors are showing signs of concern that they’re customers are becoming fed up as they attempt to venture into analytics with ill-begotten tools. Unfortunately, in their attempts to find a quick fix, they try to emulate tools like Tableau without taking the time to understand them and the reasons that they work. They’ll have to do better if they hope to survive.