You Can Only Do Two Things With BI

I get a kick out of how vendors tout Business Intelligence. It’s usually something like, “Get more out of your data” OR “Don’t you want to be data driven?” OR “All the really cool companies are doing it, shouldn’t you?”  The problem is that this doesn’t say a bit about what you can really do with BI. And frankly, becoming data driven is kind of an amorphous (i.e. stupid) reason for spending a lot of time, money and political capital, isn’t it? I mean, where’s the payback in being more data driven? It’s probably similar to the payback associated with “Paint your walls a more attractive color.”

So, why invest? Well, in truth, business intelligence can provide one, or both, of only two things: Insight and Process Improvement. Two potentially powerful things but, still, just two.

Insight

Looking to learn something about your business that you don’t already know? Then, you’re looking for insight. People looking for insight don’t know what they don’t know but they’re determined to find it. 

For example, we’re working with a client right now who has a definite need for insight. They have a mass of data, we’re talking in excess of 40 billion transactions per year, coming from two separate sources. This organization employs an analyst who is sure that he can find out new things about their customers, and what their customers need, if he could just run various models and scans on this data. He’s great at coming up with hypotheses, now he wants the tools with which to test them. 

BI for Insight has some interesting characteristics: 

  • ROI: You don’t know what you’re going to find, or if you’ll even find anything, when looking for insight.  Thus, the ROI is virtually impossible to predict.  However, if you’ve got good hunches, and you’ve got good analysts, you might just find huge opportunities. 
  • USERS: As insightful as we’d like everyone to be, in truth the number of insight-focused users in any organization (unless the organization is a university statistics department) is tiny.  It takes special skills, special tools, a special mindset, enough time, and a mandate from management to find insights.   How many organizations really have all of these in ready supply? 
  • DATA: While we can build great data warehouses with all the controls, and refresh architectures for long-lived stability to support our insight users, should we?  Really?  Think about it, these guys need data from which to work but are buying patterns, for example, really going to change that much between Monday and Tuesday?  Probably not.  So, in many cases, it can be very cost effective to just get these guys a data set and then let them play with it.  Don’t spend a lot of money keeping it clean and refreshed.  Next year, or a few years from now, the underlying patterns may have changed and you can grab some new data.  Next week, though?  Not worth it.

Process Improvement

Process improvement is the other use for BI. Process improvement focuses on using data to do what your folks already do, but doing it better.  For example, perhaps your sales team calls on each customer each month. But, if they could see which customers are suddenly buying less than they did last year, they might focus on those first, trying to head off problems. So, perhaps you give them a sales call planning tool that suggests the customers on which the salesperson should focus based on pre-established criteria. 

This sales call planning tool is a BI application but not what I call an ‘insight’ application.  In this case, you want to change the business process – to do what you already do, but do it better. Sales people aren’t scanning through billions of records looking for patterns, they’re changing their basic business processes.  See the difference? 

Things to consider about process improvement BI include: 

  • ROI: It’s sometimes easier to predict the ROI associated with process improvement applications because you’re redesigning a business process with a goal in mind. For example, you might be able to say something like, “Given what I know about our sales, heading off 25% of customer defections should protect $1.3m in revenue.”
  • USERS: Process improvement applications are typically used by larger populations, folks whose ‘day jobs’ don’t include data scanning.  Management dashboard users, believe it or not, fall into this category. They’re looking for interesting patterns but they’re inside the box that was drawn by the person who built their dashboards. 
  • DATA: Unlike insight uses, process improvement almost always requires refresh architectures. We’re enabling repeatable business processes, right? I guess, if you’re always using the same data, the process is 100% repeatable but, on the other hand, it also quickly becomes 100% counterproductive.  In other words, the new business process requires a constant data refresh. 

Are there grey areas between insight and process improvement? Sure. Perhaps a sales person looks at each of his or her customer’s histories to see what they’ve done in the past.  Kind of looking for insight. But, in real life, this probably doesn’t happen all that often. Remember, these are people who already have 40+ hours worth of work to do each week. They are tasked with selling, not with analyzing data. 

On the other hand, an analyst can certainly query off of data collected to support business process improvements. In this case, a data warehouse would support both modes of use. 

So, why does this matter? Because too many folks fail to understand what they’re aiming for before they shoot. Rather than building BI to give people data, first figure out what end result you’re really trying to get – and then the reward will be much higher returns on your BI investment. 

My Recommendation

To maximize the returns on your BI investment, consider the following… 

  • Get beyond the seductive idea of being “data driven” and on to what you really want to do with data and how that new thing will pay off. Then, build to support that real, business need. 
  • If what you’re really trying to do is support “insight” users, consider forgoing the time and expense of building a traditional data warehouse. Consider instead newer data technologies like Hadoop, MongoDB and other tools that allow you to quickly set up and query data sets. 
  • In rare cases, forward-leaning organizations are starting to blur the line between “insight” users and “business process” users.  These organizations employ data-savvy folks who execute processes but who also have the skills to look for insights. This trend is very young and this kind of employee is hard to find but, if you’re forward-leaning, this might be where you should be leaning, too.
Scroll to Top