It seems to me that advances in the value that BI can add must now be to do with how it is used, rather than technological progress. I think that many (not all, many) of the tools that you need to drive BI adoption in organisations have been there for years. I managed to reach 90% of my target audience (50% of the organisation) and achieve 93% user retention back in 2003 with plain old PowerPoint 7.

As with any mature technology, it now has to be about adding value, rather than being sexy. Maybe this is why the vendor presentations lack sparkle – it’s like MS telling us how much better Excel 2007 is than its predecessors.

I’ll try to stay more up-to-date with your blogging in future!

Peter

I don’t think the blame can always be laid solely at the vendors’ doors for lack of progress, innovation and pervasiveness of BI.

Users need to exhibit enough creativity themselves in order to drive the success/adoption/ubiquity of BI forward.

I can’t say if that is a lack of education/training on what may be possible, or simply a raw materials issue. You can’t make a purse from a sow’s ear.

Regarding Andy’s point in the comments : “Goes without saying that the value of BI is in the answers, not in the questions. Then again, there are significant issues even in defining the questions”

I think that the impetus (and the resulting acceleration of adoption) for BI is dictated far more by the questions themselves, more than the answers.

The refinement of the question scope is a secondary issue if the creativity and drive to ask a large number of incisive questions is simply not enough.

Of course, these users may be weary (and wary) of asking questions over years that the BI vendors have not been able to help them answer, leading to cynicism and apathy about any claimed improvements in capabilities and ease of use.

Vendors could throw users a rope by providing more interesting case studies and usage scenarios, but in the end, the needs of individual businesses is so diverse, it can serve only as inspiration, rather than a template. Again, it is up to the users to project that inspiration on to their own business requirements – unless you prefer the external professional services route.

As Mark Madsen says, BI is everywhere, and maybe highlighting more of these creative examples would inspire users to do more with the technology they have invested in.

Maybe more organizations like DunnHumby will spring up, specializing more in asking BI questions rather than selling BI technology.

The BI tools we look at are the same thing we had at my first job writing report-bursting code off 9 track tapes. We went from greenbar on paper to greenbar in EIS tools to greenbar in a browser. I agree that there’s a relevance and accessibility problem – navigating and finding information is not as easy as it needs to be.

I think the IT-biz duality you mention with BI will evolve away as IT takes care of data infrastructure and new tools and models for data delivery and use develop.

I speak at BI conferences, but I don’t attend them. The interesting new things are happening in other fields in academia and in the IT market. Go to an emerging tech conference, or academic conference, or web developers event and you see BI everywhere. It’s just that people don’t call it BI and just plug into data.

Spend a little time with design people (in the traditional, not software) sense and you’ll see a whole range of new things people are doing with data. Heck, look at what Radiohead did with digital data in a music video: http://code.google.com/creative/radiohead/

As Dan said, storage may be idiosyncratic, but that could be addressed by employing a more useful representational framework for metadata, such as ontology and reasoners.

Decision support systems were by and for business people until data warehousing came along, then it became on IT affair. That’s why most BI tools today (though it is changing) are still based on manipulating and reporting data, not creating collaborative models.

If you include spreadsheets and personal databases in the BI tally, you’ll find that the market share of the largest BI vendors is about 0.1% and you will also find that BI penetration is closer to 75%, not 5-10%. Knowledge workers voted with their feet because a spreadsheet solves their proximate problems.

People shun BI because they don’t understand the meta-models and too often, what is presented to them isn’t relevant enough for their day-to-day work.

The solution is to resolve this IT vs “the business” duality that was created by IT (you won’t find anyone in an organization who says “I’m the business”) and to lobby for software products that make data smarter. We have the resources to do it, someone just needs to jumpstart it.

-NR
twitter: nraden

Kidding aside, BI doesn’t work without “building things”. While I agree that BI will be consumed more and more through specific applications that contain some BI functionality, the question is, who builds these things… Your “packaged BI applications” need to be built, too. Granted, not by the end user, but by programmers, Java or C++ geeks, that have enough knowledge what the end user needs. I see first signs of “model-driven apps development” by which the end user can assemble semantic BI and performance management components into an application, without the need to dive down into code. But that’s still early stage.

To do the above, you would still need to “build” the models, “build” the data warehouse, “build” necessary transformations. So, the fact that building stuff is hard is largely irrelevant, as BI wouldn’t work without “building things”.

Goes without saying that the value of BI is in the answers, not in the questions. Then again, there are significant issues even in defining the questions. Because of a consistent lack of definitions or metrics you may get multiple answers to the same question, depending on the individual interpretation of the question. “What is revenue?” “What is net sales?” “What is a delivery date?” If those questions and their corresponding metrics aren’t clearly defined, the answer can’t be trusted.

BTW, that Google comparison doesn’t work so well. Google does not give you “the answer” either. What Google does is pointing you to a list of documents from Google’s large but limited index that may (or may not) contain what you were looking for in the first place. That’s like asking a BI system about “net sales of iPods last month in Stamford” and the result is a list of all relevant credit card receipts, online orders, etc, throwing in some Stamford Advocate newspaper reports about iPod sales plus a few links to some Stamford sales guy’s iPod videos on YouTube. Is that the answer?

One of the fundamental problems in BI is the fact that quite a few people approach it as if it is was easy. Those vendor demos showing how to quickly build funky dashboards by click-here, click-there, drag, drop, and voila, done, is putting more oil on the fire. BI has become a highly complex organism with a lot of moving parts. Maybe that’s why users get stuck in “reporting” because that’s what they can grasp.