From looking at various successful case studies where big analytics have been employed, it is clear that success depends on a good understanding and definition of the business problem, and the ability to embrace a wide variety of relevant data sources, which might be quite diverse in nature. Predictive models then need to be managed through their own cycle so that performance can be monitored and changes made when necessary. These case studies are courtesy of FICO, the sponsors of this report.
- A leading North American supermarket chain wanted to target and interact with customers earlier in the purchase cycle. To achieve this it would be necessary to increase the depth and speed of analytics. Time-to-event (TTE) models were already used that pinpoint when a customer is likely to purchase particular products – often resulting in a lift of 150% in average visits for redeemers over nonredeemers. The business engaged FICO to see whether a combination of descriptive and predictive analytics could improve matters further. The descriptive technique used established similarity based customer groupings (called “archetypes”). This allows the retailer to reach out to customers with greater accuracy, and significant improvements in customer interactions have been realized.
- As the telecom industry increasingly focuses on precise management of risk and reward, one global player in this industry determined to address the marketing efforts which resulted in problematical customers. It has used big analytics to move beyond traditional credit classes to more granular analytic segmentation that separates populations by credit risk and customer lifetime value. The big analytics initiative uses a wide range of data to not only predict customer behavior, but also balance all the key elements of risk and reward in its decision to prescribe the best action for maximizing discounted cash flow over time. This has resulted in a significant fall in valued customer attrition rates, knowing when to apply leniency and when to act.
- A large North American bank employed big data analytics to address several issues – new product development, strategic segment growth and customer retention. Big analytics was used to help develop a new credit card for a particular target segment. This involved rigorous research, strong analytics and alignment of sales and marketing plans. Advisory services for affluent customers were improved by embracing a much broader data set and applying analytics. Customer retention was significantly improved after the bank engaged FICO to help with analytics. This involved analysis of thousands of characteristics and the creation of predictive models which resulted in much greater take-up of new offers. The primary lessons learned were that disparate data sources create accurate predictive attrition models; that timeliness is very important, and that large data volumes were critical to the effectiveness of analytics.
- A manufacturer of autos wanted to improve sales of one of its models. To achieve this a mobile marketing campaign was to be launched with high impact messages. This business decided to use FICO’s Big Data Analyzer to gain insights and build models from a broad range of data sources including Twitter, and to employ text analytics. Sentiment analysis was carried out with extraction of high frequency product differentiators. This resulted in a highly successful mobile campaign.
What is most noticeable in these and many other case studies is the diversity of business problem and the uniqueness of the solutions. It is also clear that in-depth understanding of the business domain is essential, and to this end the use of data exploration and visualization techniques is essential. The multi-discipline nature of building and using predictive models means that an integrated decision management environment is crucial.
The previous article in this series is Big Analytics Applications
The next article in this series is Big Data Analytics Strategy