Homepage Rankings and Research Companies Channelcast Marketing Matters CRNtv Events Acronis #CyberFit Summit 2021 Avaya Newsroom Experiences That Matter Cisco Partner Summit Digital 2020 Intel Partner Connect 2021

4 Keys To Improving Business Intelligence Projects

Data analytics projects often get bogged down. Read how one CIO learned to avoid the pitfalls.

Niel Nickolaisen

For a long time, we IT types have installed all kinds of transactional systems: ERP, CRM, sales force automation, e-commerce and warehouse management systems, among many others. Those systems have been generating piles of transactional data that can provide business insight. This idea becomes even more intriguing when we consider all the non-transactional system data we can now collect from social media, mobile devices and other sources.

But if we want to get real value out of our analysis of this data, we have to start doing things differently. For too long, our business intelligence projects have been more about reporting and visualization than insight. It seems we hope that by getting a prettier picture of the data, the epiphanies will follow. In order to get the intelligence out of business intelligence, here are some lessons I have learned:

1.) Listen to the data and ignore opinion. We have a natural human tendency to express our opinions. Sometimes those opinions filter out the data that shows us the real cause-effect relationship. To avoid this, we need to suspend our opinions and let the data do the talking.

2.) Translate data into dollars. With so much data, it's easy to pursue analytical hobbies. When planning an analysis or an analytics project, focus on where the money is -- what are the activities that drive growth, customer retention, profit margins, etc.?

3.) You don't need perfect data. Too often, analytics projects slow down and get distracted by the quest for perfect data. We consume resources writing, testing and rewriting data conversion scripts. When it comes to data cleansing, particularly given the piles of data we have, close enough is good enough.

4.) Things move fast; so should your analyses. If it takes months to create and validate an analytical model, you should find a better way to do the analysis. In this dynamic, competitive environment, you need both accuracy and speed.

Recently, we had a really interesting analytics opportunity. We thought we had the data to identify the factors that drove customer retention. Could we use this data to create analytical models that would let us identify "at-risk" customers and trigger the specific intervention that would retain that customer? If so, this would quite possibly be the best work I had ever done. Given my checkered history with business intelligence, I made sure that I followed the above guidelines as we developed our customer retention model.

We started by not making any assumptions about cause and effect. We put the data through an advanced business intelligence tool and let the data tell us what drove customer abandonment. We also let the data tell us what specific actions (email reminders, offers, telephone calls from customer service agents, etc.) would make the biggest difference in retention.

The data told us which customers would respond to specific actions. The data also told us when we should do the interventions. Knowing this, we could figure out the relationships and make sure the interventions were worth the costs.

Rather than spending time perfecting the data, we figured we had enough data to generalize the results; we figured that if 80 percent of our data was good, so was the analysis.

Finally, our advanced modeling tool allowed us to complete analyses in hours, not months. That let us do lots of hypothesis testing. We could try an intervention, analyze the data and update our models and our intervention plans.

When we started this project, I hoped it would be some of the best work I have ever done. And I think it was. We identified the common elements of customer retention success and failure. From this came the characteristics of customers trending toward leaving us. This allowed us to build meaningful profiles of "at-risk" customers. As soon as a customer started to develop an "at-risk" profile, we took action. Depending on the type of "at-risk" profile, we knew which actions would keep them from straying.

When all was said and done, customer satisfaction and retention went up more than 15 percent, an increase worth millions of dollars to the bottom line. Like I said, some of the best work I have ever done.

Niel Nickolaisen is the chief information officer for Western Governors University, a nonprofit, entirely online university. He holds a M.S. in engineering from MIT and a B.S. in physics from Utah State University. He is the author of "Stand Back and Deliver" and one of the founders of Accelinnova, a think tank focused on improving organizational and IT agility.


Back to Top



    trending stories

    sponsored resources