Reduce Churn with Targeted Resurrection Emails

Please note: This post was originally written by Bob Colner, my colleague at Boomtrain, where he is our lead data scientist. I was so impressed with the article that I wanted to share it here on LinkedIn so others could learn more about this topic. Thank you, Bob, for a great article.

Churn hurts. Every instance of customer churn represents the loss of both past investment and future revenue, which is why marketers lie awake at night trying to find new ways to reduce turnover/attrition/defection or whatever euphemism your industry has coined for it. But don’t feel too bad. Eliminating churn completely is like having a carbon footprint of zero: nice to imagine but not realistic. What really hurts (and what you maybe should feel bad about) is churn that could have been prevented.

The opportunity of tomorrow is using real-time data to predict what kind of resurrection email will trigger churned/churning users to re-engage.

Stop churn before it starts

Your best opportunity to prevent churn is when the customer relationship is at its healthiest. Every email you send is a strategic retention lever—not just the ones you send after you suspect a customer might leave. Once your users give you permission to send, you have the great power to connect with them in whatever manner and at whatever pace you choose. You also have the great responsibility to make sure you’re always delivering value to them through relevant and timely information.

To this end, you should be A/B testing key design and editorial elements regularly, but you also need to test and evaluate big picture campaign elements like cadence, frequency, timing, and behavioral triggers, so make sure your individual emails and your overall communication strategy are optimized for maximum performance. (Maximum performance = minimal churn. See our blog post measuring retention)

You also need to know when and how to diversify your communication strategy. You will naturally (and correctly) want to focus on optimizing your primary newsletter or email campaign first. If done right, you’ll see rapid improvement, but as the low hanging fruit gets picked over, performance gains will become harder and more incremental. It’s important to realize when you’ve reached the performance limit for a single newsletter or campaign so that you can start segmenting and targeting narrower audiences before their interest plateaus.

But, you might be asking, as the hosts of this party, what do we do about the guests who—despite our best efforts at hospitality—seem to be inching (or even running) towards the door?

Identifying Churn-Risk Users

Most users will slip away quietly, so you need to be alert to the subtle shifts in behavior that may indicate a churn risk, keeping in mind that different users will give different signals. Finding complex relationships in large and incongruous sets of data is difficult to achieve at human scale, which is why most sophisticated marketers turn to vendors who employ advanced machine learning technology to gauge an individual user’s churn risk. These tools analyze user behavior and extract patterns that are likely to precede churn, enabling you to more accurately and effectively engage high-risk users. (The best ones also enable you to rank those users based on how valuable they are likely to be so you can decide who you really want to invest in keeping.)

the ‘4×4 Rule’: A user who has visited 4 times in the last 4 months but has not returned in the last 4 weeks should be flagged as a churn risk.

Even if you’re not ready to invest in advanced data science, you can use some basic analytics tools that help identify possible churners. We recommend starting with a method called the ‘4×4 Rule’: A user who has visited 4 times in the last 4 months but has not returned in the last 4 weeks should be flagged as a churn risk. If you want to get a bit more sophisticated, go back through historical data and see what percentage of your 4×4 users churned compared to your overall rate, then look at 4x4x6, 6x4x4, and so on. The more you can hone in on what predicts attrition in your customers, the more effectively you can target them with retention notifications.

When all else fails: 5 steps to better resurrection emails

When a user crosses the threshold into churn-risk, you still have some levers to pull, but you want to pull them very strategically. At this stage, irrelevant emails are even more likely to trigger an unsubscribe, and then you have no choice but to return that user to the top of the funnel.

Offer the right kind of value

A bit of a no brainer, but one worth digging into. Offering value at this stage means something different. “We miss you! Come visit” may have worked earlier in the lifecycle, but now we need to think more along the lines of “We miss you! Here’s a plane ticket.” These users have become habitualized to not thinking about your brand, and you need to break them out of that pattern with something special. Coupons, promo codes and special offers are the go-to incentives at this stage, and they have proven effective.

Know your limits

It’s not worth spending $10 to retain a $5 customer. Increasing the size of the incentive will increase the likelihood of re-engagement, but you need to consider the lifetime value of the customer. This, again, is an area where machine learning and predictive analytics can help you gauge the lifetime value of a customer and cross reference that with their likelihood of churn so you can calibrate your investment in retaining or resurrecting them.

Supplement financial incentives with content-based incentives

The problem with using financial incentives alone (aside from the fact that this may not be an option for content publishers) is that they reward a behavior that you don’t want to continue to encourage. You can also incentivize engagement with content. Highlight your best and newest, consider sneak-peek content, or even “what you missed” content to try and provoke a fear of missing out.

Tailor your message and content to the individual

Personalization is vital to the resurrection experience. (In fact, there is a high cost associated with NOT personalizing your communications.) If you want someone to be a part of your tribe, you need to show them that you know and value what’s unique about them. Emails should be personalized with messaging and content that’s most likely to engage that individual recipient based on past user behavior and real-time data about content performance.

Automate & Optimize

To achieve the best results, you want to create a set of rules for recurring churn resurrection emails. A good practice is to identify and email at-risk users on a monthly basis. A better practice is to tailor each individual email with content that’s relevant to that user. A best practice is to automatically trigger personalized resurrection emails based on individual behavior, and time them to arrive when that user is most likely to engage.

The technology available to marketers today enables you not only to set rules about resurrection emails, but to continually optimize those rules through machine learning. Using retroactive data to predict churn is a huge win, but predicting churn is only half the battle. The opportunity of tomorrow is using real-time data to predict what kind of resurrection email will trigger churned/churning users to re-engage. Then maybe we marketers can finally get some more sleep.

post originally appeared on the Boomtrain blog.