Andy on Enterprise Software

A long journey

January 5, 2007

An Accenture study:

http://www.informationweek.com/research/showArticle.jhtml?articleID=196800921

quantifies how much time middle managers in enterprises waste seeking out information, and comes out at two hours a day i.e. a quarter of an average working day. When they find it, half of the information turns out to be of no use. This sounds about right to me, and ilustrates just how far BI really has yet to go in being genuinely useful, and also shows just how bad the true state of information is in large companies.

The issue is not only that technologies are insufficiently intuitive. In my experience there are a number of factors that come into play:

- no culture of sharing information
- inconsistent data definitions
- poor data quality
- inability to locate appropriate data sources
- insufficient understanding of how to use BI tools effectively.

If you set out to produce a useful new report in some area and succeed in doing so, what incentive is there for you to make this easily shared around the company, and to help others find it? In most companies this would be pure altruism, and so people just keep the information on their hard disk, and indeed may gain kudos from the “information is power” syndrome. Overcoming such cultural barriers is hard, and few companies succeed. I should say that Accenture themselves do as good a job as anyone I have seen, where their consultants are actively tasked with documenting project lessons and storing these, with appropriate keywords, in an internal knowledge management system. However I have not seen this in other consultancies to anything like the same extent.

The other problems are all too familiar to people working in BI. Inconsistent data definitions and poor data quality are the heart of what MDM is all about, and we know how immature that is. Yet without fixing this then accurate and easy to obtain information is still elusive. A further problem which some technologies are starting to address is the sheer job of finding an existing report. Ironically there is an excellent chance that if yoiu want some partioular report, then someone else did too and has already built it. The troiuble is that may be in an Excel spreadsheet on a hard drive, or sitting on a shared server but you simply have no easy way of finding that it is there. It is ironic that Google allows us to search the whole internet in moments, yet finding a report within our own company is a much tougher proposition. Enterprise search vendors like Fast and Apptus, as well as Google itself, are beginning to apply smart technology to the problem, but here it is still early days.

Finally, most end users either don’t have access to create a new report easily, or are not trained in making best use of BI tools, or simply don’t have time to learn. This is why Excel is so popular; it is familiar and ubiquitous, and so people would rather get data into Excel and play with it there than learn a new BI tool.

I believe that these are mostly quite intractable problems, only some of which lend themselves to new and better technology. So anyone with a magic bullet e.g. “the answer is SOA” is talking nonsense. It is only by addressing the organisational, cultural and data ownership issues in combination iwth enterprise search and better tool training that a company can improve that two hours a day per person. It will be long, hard slog, and buying the latest trendy tool is not enough, whatever the salesman tells you.