Digital Marketing – Machine Learning

Home / Case Studies / Digital Marketing – Machine Learning

Background

One of the top four wireless carriers in the U.S. with millions of prepaid subscribers was looking for ways to reduce churn and increase revenues generated from their customer base. They primarily relied on SMS and email channels to message their customer base and were seeing response rates to their marketing efforts slowly decline over time.

The Challenge

Due to the size of the customer base, the myriad of segments and the relatively small marketing team, it was difficult to deliver customized messages that its customers would find both, relevant and compelling. The resulting one-size fits all marketing approach drove results but the over-messaging involved was expensive, drove list fatigue and contributed to customers opting out of receiving future messages.

The Solution

The carrier selected a big data company and worked with Real Marketing Solutions partners to implement and manage a solution leveraging marketing analytics and machine learning to proactively deliver the right message to the right customer at the right time.

Step One: Data Sourcing

The first step was to determine the customer behavioral data that resided at the carrier that could be used to feed the big data marketing analytics machine. A vast array of data from multiple sources was reviewed and hundreds of data elements were selected to be delivered to support the effort. As part of this process nearly a year’s worth of data was provided for set up and then daily data feeds were implemented to continue to provide signals and campaign results to the big data machine.

Step Two: Model Training, Offer Development and Content Creation

Once the initial data load was completed, the big data company began its data analysis efforts and started creating models based on signals they were able to read from the data, specifically aimed at churn reduction. At the same time a library of offers (and the email and SMS content containing those offers) that would be used as marketing treatments during the pilot phase of the program was built.

Step Three: Pilot Phase

A geography representing about 10% of the customer base was selected with half of the market continuing to receive business as usual communications while the other half was sent messages as dictated by the big data company. To support this part of the effort daily campaign feeds from the big data company to the carrier’s CRM partner were built as well as contact history feeds back to the big data company from the CRM partner to close the feedback loop and support the ongoing machine learning effort.

The Results

In a typical rollout additional geographies would join the program with a measured approach while results were continually monitored to ensure that the lift shown in the pilot phase applied to the newly added segments. But in this case, the initial test results were so positive that the carrier’s senior marketing management decided to accelerate the rollout dramatically to reach the entire customer base much more quickly than originally planned.

Based on the results of the pilot phase the big data company was projecting an increase in annual revenue as a result of the program of more than $30 million with additional benefits including large savings in marketing deployment costs and lower opt-out rates due to sending about 20% of the messages that were previously being delivered.

Can we help you improve your installed base marketing results by implementing big data marketing analytics and machine learning? To find out more call us at 425.243.4109, visit us at www.realmarketingsolutions.biz or write to info@realmarketingsolutions.biz.

Leave a Reply

Your email address will not be published. Required fields are marked *