There is a survey this week about current BI issues and trends:
I always tend to be a little sceptical about such surveys since you are never sure how representative the sample base is, but with that caveat there appear to be a few interesting themes. Broadly spending on BI is expected to go up by 10%, and this is in line with other sureys.
One opportunity for the industry is that just 4% have a BI tool on a subscription basis, but 30% would be interested in one. The success of salesforce.com suggests that if existing vendors hesitate too long then a competitor could get an advantage here (just ask Siebel).
73% of respondents reckon they are reducing the number of BI tools in their shop, with 29% hoping to get down to one. While some of this may be wishful thinking it certainly reflects the recent trend towards consolidation e.g. Hyperion buying Brio, and then Oracle buying Hyperion.
Just 18% of respondents are “evaluating” open source BI tools, presumably with considerably less having actually deployed one. This is a useful wake-up call to the open source zealots. At least in the BI sphere there seems to be little impact so far.
However I found the most revealing data point to be the satisfaction levels with existing solutions. In eight categories (ETL, data warehousing, various packaged apps) the second highest scoring category (ETL) had just 43% of people claiming to be “successful”, and this feeble rate drops to around 30% or so for analytic apps that are packaged with financial, CRM or supply chain systems. 44% of people claimed success with packaged BI reporting tools, but again this is hardly a number to be crowing over if you are a BI vendor: “we are slighty less dismally unsuccessful than packaged analytical apps” isn’t something you want to put on your advertising campaign. Of course, as with all such surveys, there is the question of how the question was phrased i.e. what does “successful” mean, but even so it hardly paints a picture of contented customers.
This level of satisfaction does not strike me as odd. Large companies have mostly not solved their basic management information problems. Many of the issues come back to data issues: poor master data and dismal data quality are rife. One large (and highly successful) company I know has been trawling through its supplier data recently: four out of five supplier records turned out to be duplicates, and just in case you assume their problem is historical, one in three of even of new suppliers set up in the last 12 months turned out to be duplicates also. With such fundamental issues of data classification and quality, you can have all the pretty reporting tools you like and yet you are still unlikely to be solving too many real management information problems. You will get numbers, but they are not good numbers
I am always surprised by how little attention that data quality gets when it clearly undermines the quality of the reports that management rely on. Maybe people just feel comfortable seeing a pretty chart: it is reassuring. Knowing that the data it is based on is junk is perhaps too uncomfortable for a lot of organisations to wish to confront.