Machine Learning Times
Machine Learning Times
Re-examining Model Evaluation: The CRISP Approach
 The performance of prediction models can be judged using...
An Agile Approach to Data Science Product Development
 Introduction With the huge growth in machine learning over...
Wise Practitioner – Predictive Analytics Interview Series: Haig Nalbantian at Mercer – BIZ
 By: Eric Siegel, Founder, Predictive Analytics World for Business...
Is What You Did Ethical? Helping Students in Computational Disciplines to Think About Ethics
 In addition to this article, Dr. Priestly will also...

4 years ago
Mobile Analytics-Mining the Visit Experience of the Customer


Mobile technology as part of the Big Data discussion is enabling data miners/data scientists to conduct analytics on information on the customer’s experience within a given location which otherwise was unavailable up until a few years ago. At the very basic level, a given company puts in routers at its various retail locations. As a visitor enters a given location with their mobile phone, which is Wi-Fi, enabled, the data from their cell phone is “pinged” to the router. The router then yields the following information:

  • Strength of signal
  • Time of entry to the location
  • Time of exit from location

With only these three pieces of information, the data scientist can begin to derive very powerful information. This powerful information is all collected at a unique individual level as each cell phone has a unique MAC ID which is the critical match key when summarizing information up to an individual level. At a static level, information can be derived that relates to duration of a given visit. Yet, it is the longitudinal view or dynamic view of visit behavior overtime that is going to provide the very rich long-term visit experience of the customer. For example, listed below are just a few key pieces of historical visit behavior   that would be extremely valuable to marketers:

  • How does visit duration change overtime?
  • Does my presence within the location change overtime based on the strength of the signal.?
  • Is my overall visit behavior declining?
  • When do I go to visit (morning,afternoon,night) and is this changing overtime?
  • How many locations do I visit and which location is my preferred location and how is this
    behavior changing overtime

The above just represents a few ideas but many variables can be created which would look at standard deviations, minimum/maximum values as well as trend behavior for each above metric. In effect,
it would be very realistic that the data scientist could easily create over a hundred variables from just the three pieces of information outlined in the first paragraph.

Value-based retention models could then be created which predict whether or not a highly engaged visitor is not likely to come back within a pre-defined time period. At the same time, once these models are created, individual retention scores can then be summarized to a store level. Each store now has an average customer retention score, which can then be used to segment stores into high, medium, and low performing stores.

But beyond the visit behavior, the real Holy Grail is the ability to integrate this information with existing customer demographic and purchase behavior data. In fact, the genesis behind most loyalty programs was to gather this type of customer-rich data as marketers recognized the great value of this information. Mobile data now takes analytics to a new level as we can now draw inferences on the actual visit experience in conjunction with this rich base of non-mobile customer information. From a data science perspective, many of the developed solutions using just non-mobile customer data produce excellent results. However, there are situations where this data yields a marginally performing solution. Having access to Wi-Fi visit behavior, new variables can be created that could cause improvement on an otherwise marginally performing model. Once again, the key to improvement is integration of this visit behavior to both the demographic and purchase behavior of the individual. As stated before, math is important but DATA reigns supreme.

Big Data discussions and what to do with all this data are not necessarily new concepts for most seasoned marketers. The focus of marketing has always been to use data to obtain that “one view” of the customer. The use of mobile technology and the resulting data by itself is not a panacea in creating CRM type solutions. Yet, if the goal of marketing is to optimize that one view of the customer, then mobile data needs to be a critical component within the overall customer data infrastructure.

Author Bio:

Richard Boire, B.Sc. (McGill), MBA (Concordia), is the founding partner at the Boire Filler Group, a nationally recognized expert in the database and data analytical industry and is among the top experts in this field in Canada, with unique expertise and background experience.

Mr. Boire’s mathematical and technical expertise is complimented by experience working at and with clients who work in the B2C and B2B environments. He previously worked at and with Clients such as: Reader’s Digest, American Express, Loyalty Group, and Petro-Canada among many to establish his top notch credentials.

After 12 years of progressive data mining and analytical experience, Mr. Boire established his own consulting company – Boire Direct Marketing in 1994. He writes numerous articles for industry publications, is a well-sought after speaker on data mining, and works closely with the Canadian Marketing Association on a number of areas including Education and the Database and Technology councils. He is currently the Chair of Predictive Analytics World Toronto.

Leave a Reply

Pin It on Pinterest

Share This