This is the second of the two-part series analyzing the white paper “Unsubscribe Predictor.” The first installment describes the concept of unsubscribe rate and how it impacts marketers.

This week we published the white paper Unsubscribe Predictor, which teaches brand, agency and product professionals how to maximize mobile marketing customer retention using various factors such as industry, time of day and message content. Today’s installment of our two-part Cliffs Notes version, a “Marketer’s Guide to Unsubscribe Predictor”, reveals our research approach and findings.

Research Approach

Having developed our thesis of how unsubscribe rate impacts customer lifetime value, we set out to develop a research approach. We started by trying to think about the common characteristics across each outbound mobile marketing blast. After some boardroom spitballing, we determined that all characteristics fit into two general categories:

  • Broadcast-specific: those characteristics innate to the broadcast itself. Examples include the time the broadcast was sent, the content of the broadcast, the frequency and “recentness” of previous blasts.
  • List-specific: those characteristics representative of the recipients of a broadcast. Examples include the age of the list, the size of the list and subscriber demographics based on factors like carrier and length of time subscribed.

To clarify with an analogy, think about how we decide on the proper value of a house. On one hand, we have to consider factors about the house itself, such as square footage, number of rooms, types of window dressings, colors of the blinds and the condition of the roof. On the other, we have to think about external factors, such as school district, city, proximity to grocery stores and neighborhood crime rate.

Math is math of course, but the general concept isn’t overly intimidating. We know that each of these factors has some impact on house price. Some of those have a huge impact, such as size, whereas something like window dressing type will only have a minimal – if any – effect. The mathematical model simply quantifies the impact for each characteristic so that once you input all the parameters you can accurately calculate the house’s value.

Straight forward, right? That’s exactly what we did with mobile marketing. The coolest part is that the models we used, which included multinomial mixture models, topic models, lasso logistic regression and random forests, allowed us to investigate unsubscribe rate without knowing a priori which characteristics had the largest impact. We let the data tell us what was most important, so that we could pass this information on to our clients and other marketers (you can read about our process in detail in the section “Feature Selection”). Across all our analyses, we averaged an 85% success rate, demonstrating the power and accuracy of our predictive analytics.

Our Findings

You can find our complete findings in section 5.3 of “Unsubscribe Predictor,” but most interesting to us are the insights about industry, call to action and list age.

The Impact of Industry on Unsubscribe Rate

Across all industries, the political vertical is least prone to unsubscribes. Logically, this makes sense. For better or worse, people’s emotions connect much more emphatically with their political views than their brand tastes. Companies in the political vertical use mobile not only to spread “Rock the Vote” style messaging, but also as an organizational tool for unions, NGO’s and lobbyist groups. Thus, for marketers in politics, capturing mobile subscribers has enormous value creation potential. Marketers outside of the politics can compare their unsubscribes to politics to understand and approximate the degree to which their customers demonstrate head, hand or heart loyalty (#Nordhielm).

The QSR industry, on the other hand, is most prone to unsubscribes. Again, logically this makes sense. Due to a lower cost of acquisition, QSR customers’ allegiance to a brand is less firm. QSR marketers should take away three lessons from this fact: one, unsubscribes should cause particular concern for QSR marketers when overall list size growth slows down or ceases. Second, QSR  marketers should launch tagging campaigns that isolate the most loyal customer groups. Finally, QSR marketers should execute in the face of competition, as customers seem more willing to switch brands.

An analysis of the retail industry revealed a less transient subscriber base, meaning that retail subscriber lists are less prone to decay over time. The largest unsubscribe events come as a result of delivering less than optimal content. So, as far as takeaways, retail marketers need to make sure that they preplan their ongoing marketing messaging upon launching an initial campaign. Furthermore, the follow-up directly after an acquisition campaign has a huge impact on the degree to which retail marketers can positively impact customer lifetime value.

The Impact of Creative Calls To Action on Unsubscribe Rate

The data reveal that content containing URL’s, phone numbers and email addresses increase the likelihood of unsubscribes. When we looked closer, we concluded that customers across all verticals look for calls to action that are creative and self-contained. Unsubscribes came as a result of customers feeling like brands tried to use mobile to forward them through to another channel.

Thus, marketers who want to use mobile to acquire customers for other channels need to make sure that they make these attempts within a mobile conversation. Mobile campaigns that ask customers to click a link or submit an email address – without any more context – have a negative effect on customer loyalty. To secure long-term customers via mobile, make sure to create personalized engagement that uses customer data to enhance targeting and incite action.

The Impact of List Age on Unsubscribe Rate

To be clear, “list age” describes an average of the number of weeks each individual customer has spent on a mobile marketing list. Younger lists denote a newer campaign, whereas older lists show campaigns active for a longer period of time. Of course, list age has a dynamic value, as an influx of new customers will make a list younger overnight.

What’s fascinating is that the data show that older lists are much less likely to have major unsubscribe events than younger ones. What this means is that, although established campaigns never escape the inevitability of losing customers, older lists escape the growing pains of large droves of customers simultaneously opting out. So, again, marketers need to make sure that they preplan the first follow-up messages of new campaigns to ensure the retention of the largest amount of customers as possible. By the same logic, marketers with older lists can experiment with new products and services, as remaining customers will be (and remain) loyal to the brand. All in all, knowledge of the state of your list, especially in terms of message count and list age distributions, provides pivotal insight to inform an engagement strategy.

The Future Of Unsubscribe Predictor

Here at Waterfall HQ, we’re really excited as to how we can run predictions based on these types of results. Now and going forward, we can instantaneously analyze any broadcast as it’s being created.  With our data insights, we will inform clients with greater accuracy which broadcasts seem to evoke a higher than normal level of unsubscribes. An actionable recommendation engine will empower our clients to maximize customer retention across all their mobile marketing outreach. And that, in a nutshell, is what makes us most excited: using the power of data to help marketers become even better and more strategic about their respective roles and responsibilities.

To download the “Unsubscribe Predictor” white paper free of charge, please go to click here

Similar Blog Posts You Might Like:
SMS Marketing Vs. Email Marketing for Your Loyalty Programs
Predicting Customer Retention – A Marketer’s Guide To Unsubscribe Rate Predictor
Top 3 Pain Points for Mobile Marketers

Share This