November 4, 2019
Maximizing customer retention
Mass adoption of text messaging started about 15 years ago, when American Idol was a top-rated television show. To vote for your favorite singer you texted a short code from your mobile device, which was fitting since mobile phones had 12 numbered buttons but no keyboard. Data services on cell phones were still in their infancy. Voting now has expanded to online and even an app, but in the subsequent 15 years since it started with text….
- The first iPhone would launch 3 years later
- The app store launched
- Mobile devices developed an open browsing format
- Mobile streaming video became ubiquitous
- U.S. Wireless penetration grew from 2/3 to over 120% (Source: CTIA)
…and 15 years ago, monthly subscriber churn rates in the wireless industry were around 2%. Wireless carriers were setting aspirational targets of 1.5%.
Fast forward to 2019 and 3 of the 4 national carriers has churn below 1%. Networks are faster and more reliable, and end users are consuming more mobile data than ever. In addition, the elimination of the phone subsidy model has customers keeping devices longer, delaying or foregoing churn events.
Postpaid Phone Churn: 1Q16 through 1Q19
Source: Top 4 National Carrier quarterly earnings reports, weighted by subscribers
Now that churn has been reduced to record levels, how much lower can it go? And how do carriers continue to reduce churn by incremental basis points?
Mature industry economics suggest it is less expensive to retain a customer than acquire a new one, with acquisition costs estimated to be 10-20 times that of retention due to lower response rates. With industry churn at such low levels, if a company loses a customer it may be unlikely that they ever get them back. Then why is so little spent on retention?
The problem begins with the inefficiency of retention. Offering promotions to 100% of the customer base - when 99% are not going to disconnect – wastes valuable marketing spend. The more precisely the vulnerable base can be targeted, the better the efficiency of the retention program and the ROI. With acquisition proving to be more difficult as there are fewer potential prospects and shoppers in the universe, it makes sense to model addressable retention targets.
There are several ways companies can find valuable insights to extract those additional basis points of churn:
Make use of data along the customer journey
- Has the customer submitted negative feedback via a survey or social media?
- Do they call customer service frequently, implying their issue is not getting resolved?
- Were they dissatisfied with a recent retail visit?
- Have their usage patterns declined, indicating a potential lack of engagement?
- Has the customer disconnected one or more lines, but still has an account, suggesting trial of another provider?
Deliver an excellent and consistent experience
- Provide predicable monthly billing to avoid surprises
- Ensure a consistent experience across all customer touchpoints – retail, customer service, digital, billing, etc.
- Acknowledge and show appreciation for the customer’s business across all touchpoints
Propensity to churn modeling can aid in identifying vulnerable customer segments that may be responsive to retention offers. Using data from the customer journey, modeling usage patterns, tenure, and other factors can play a role in predicting future churn.
This modeling forms the basis for scoring the customer base placing them into deciles of risk, then predicting likelihood to churn using a holdout sample. Once the base is scored, marketing teams can target the most vulnerable decile(s), rather than offer to 100%.
Customer marketing treatment would then be conducted in an omnichannel testing environment, where marketers create and trial offers, and report churn results compared to a relevant control cell. Since churn does not happen immediately, these tests need to be measured over a defined period. The offers should have varying levels of discount or added benefit - in the case of mobile customers it could be bonus data or a short-term monthly discount, or some enticement for the customer to upgrade and extend the relationship.
Another offer construct might be a marketing message of network investment and improved performance in the customer’s local area, highlighting increased speed and reliability. Marketing trials would then inform the effectiveness of the predictive model and guide future retention offerings.
1. As wireless networks have improved, customer churn has declined
2. The next step down in churn can be achieved through examining the customer experience
3. Companies have the data available to them to make incremental changes
4. Predictive modeling can help target the most vulnerable segments of the customer base
BLEND360’s data science team has extensive experience in building, executing and testing targeted acquisition, retention, and maximization models for the telecommunication sector and many other industries. Find out more about our capabilities at www.blend360.com or email us at email@example.com