Guesstimates are often essential for market sizing and trending. To be useful, especially where primary data are lacking, they demand a valid conceptual framework.
Like you, I’m looking forward to the responses to Boris Evelson’s quick Web-based survey, which you can access from his most recent blogpost. It’s always a challenge to assess how truly pervasive BI is—and pervasive it could potentially become.
To generate a valid first approximation, Boris scoped his blog comments and quick survey to “traditional BI” applications (i.e., historical reporting, query, dashboarding). He scoped his estimate only to large enterprise and midmarket firms (i.e., those with 100 or more employees) and only to BI usage in the US.
In order to keep this task manageable, Boris excluded some use cases that are often included in the “traditional BI” category: spreadsheets and other “homegrown” analytics apps; BI embedded in line-of-business apps; and non-interactive, static, published BI outputs. He leveraged both public and Forrester-gathered primary data to gauge how many actual and potential BI users there might be.
Scoping it as he did, Boris estimated that slightly more than 1.5 million people in the US are using traditional BI applications, which is between 2-3 percent of the employees of BI-implementing firms. He suspects the actual percentage might be as high as 6-8 percent of employees, but he’s not sure. That’s why he’s running the Web-based quick survey.
On a related note, I’m curious how many actual and potential users there are for advanced analytics applications. Of course, that’s entirely outside the scope of Boris’ focus, but very much part of the “BI writ large.” If traditional BI is far from pervasive in the business world, then advanced analytics—which I scope to predictive modeling, data mining, text analytics, in-database analytics, and complex event processing (CEP)—is a mere drop in the bucket.
Interestingly, the Q4 2009 Forrester survey gives us useful fodder for such an estimate. It shows that, where reporting tools are concerned, 74 percent of enterprises either already use them or are planning to implement them in 2010. For predictive analytics and data mining (PA/DM), the equivalent number is 31 percent; for text analytics and CEP it’s 19 percent; and for in-database analytics, 7 percent. Considering that PA/DM is the most widely adopted advanced analytics solution category, I’d estimate that perhaps 25 percent of the companies that responded use or plan soon to implement one or more of these types of solutions.
That would mean approximately one in three companies implementing traditional BI also uses advanced analytics. In other words, if, say 3 percent of employees in BI-implementing firms use traditional BI, that would correspond to 1 percent of those firms’ employees using advanced analytics (for the time being, let’s not concern ourselves with the overlap, within each firm, among users of traditional BI, on the one hand, and advanced analytics, on the other).
As a back-of-envelope mental calculation, it appears that Boris’ numbers correspond to roughly 300 traditional-BI users per company for those that deploy the technology. From statistics gathered for my recently published Forrester Wave on PA/DM Solutions, I know that there are an average 10-20 users (i.e., statistical modeling specialists, primarily) per PA/DM-implementing enterprise. Let’s take the midpoint on the latter range: 15 PA/DM users per firm. Hence, we’re looking at roughly 5 percent of employees in “BI writ large” user companies that use PA/DM. If we adjust this to include the smaller numbers of text mining, CEP, and in-database analytics users, we might guesstimate 6-7 percent of employees in “BI writ large” user organizations are also doing advanced analytics.
How does that look to you? I haven’t done a Web survey on this. I’m not yet sure that I would be able to define and frame the questionnaire in such a way as to elicit valid responses.
As I said, I’m waiting for the results of Boris’ more narrowly scoped survey before deciding how to proceed on this.
Jim also blogs at http://blogs.forrester.com/james_kobielus.