Business intelligence thinkers can take a leaf from the books of grammarians, who speak of “tense” as defining the time over which an event occurs. And as someone who has long criticized BI for being mostly concerned with predicting the past, I thought it would be instructive to consider the convoluted way we often find ourselves describing the work we do in BI activities.
Simple tenses are just that: “we sold 10017 widgets last month and our margin was 21.3%” is simple past. Great stuff, and we did that in the earliest days of batch programming. After the transaction program ran, we ran a report program that told us what it did and sent it to print on green bar paper. (If you don’t know what that is, ask one of the older guys.) And, of course, we still do – this stuff never goes away, and shouldn’t.
Perfect tenses are more rich: for example, past perfect indicates that an action was completed (finished or “perfected”) at some point in the past before something else happened. And future perfect is very cool: it tells us about something that will have been competed at some point in the future. For example, “by the time this quarter is done we will have raised our margin to 22.45.” This is today’s predictive analytics in a nutshell, applied simply.
Progressive tenses tell us that something happens continuously (like most business processes we care about.) So future progressive describes something that will be happening: “by Q3 we will be improving our margin at a steady pace as these new efficiencies kick in.” Just the thing for business performance management.
Finally, if we compound these we can get to something called future perfect progressive, which indicates a continuous action that will be completed at some point in the future: “by the time we finish rolling out this program, we will have been improving margins every quarter for 2 years.” This is a richer way of doing predictive analytics: describing a process with a reasonably well-understood endpoint and describing its likely outcome in terms measured in values we can communicate in business language.
So, if we’re going to be tense about BI – and who isn’t these days? – at least we should be future perfect progressive tense. The next challenge is whether the BI can be directly integrated into the process in question, so that the effects can be had without requiring human intervention. You know, every BI demo you ever saw used to end with “now we can update the forecast.” Not good enough! Why not let the analytics be used in an automated way? Use rules to avoid transactions with a low margin, for example, in favor of ones with better margins. That is not a simple lookaside table lookup; it will require real-time analytics coupled with fresh data: commodity prices, creditworthiness, or other continuously variable facts.
That’s where we need to go next. Call it the analytic application, where business processes are connected (maybe using SOA) to real time analytics, with rules engines or other mechanisms that change the execution of the business process to drive a desired outcome.
Have you got examples of products you can buy today that operate that way? Have you implemented one? Please let me know by leaving a comment here or emailing me at merv@itmarketstrategy.com. I plan to discover and document those stories in the months ahead, as this next step in applications emerges.
Tense is what Bill (our resident data analyst) gets whenever I mention BI. He has long been comfortable answering questions and sees no need to further automate. Moreover, management is reluctant to invest gobs of money in imperfect systems for the purpose of generating “sexy” output when they have the simplest interface and the best results generator in place today. His name is Bill.