Reducing B2C Churn by 30% Via UXR, AI-Infused Engagement, and No UI .
- Alan Mills
- Jun 10
- 5 min read
Introduction
A common business objective is to acquire and sustain sales or subscriptions to reach a particular milestone. After a year of enhancing TTV metrics and reducing friction in converting customers, we made significant gains by more-than-doubling our activation rate. However, this incoming traffic must be sustained in order to easily reach a quantitative business objective (ie. "Sustain 15k paid subscriptions"). In this piece, I walk through how we've identified at-risk users and reduced churn with an AI-generated email that re-engages them with progress tracking, as well as providing them with guidance to find relevant, alternative content that might be more compatible or compelling for them to consider.
Prior to April, we were essentially gaining as many users as they were leaving. While our sales intake increased, we couldn't celebrate our primary goal of sustained MRR from subscription sales. In an earlier PLG audit, I've already pointed out the opportunity space where we lacked any re-engagement strategy that taps into the user's current activity. So I looked into a solution that brought our churning traffic down significantly in early April, allowing for us to increase of subscription count, as desired by our CEO (with the exception of losing a few high-volume, B2B sales from not updating our B2B UX).
Understanding and Identifying At-Risk Customers
All product teams can ethically prevent churning customers like we've done by helping at-risk users rectify barriers they are experiencing and help them find value in their investment. The first step is to understand how to identify these at-risk customers. Usually, these are customers that begin to drift away, disengaging from the product / services of your team's value proposition. They've either struggled to make use of these services, and potentially found solutions elsewhere - due to barriers in ability, awareness, or motivation from external sources. This disengagement can be tracked through analytics tools.
We can start to identify these users by flagging customers with decreased activity. At Pragmatic Works, I've estimated that if a user fails to re-enter the platform over 7-days, this is a disengaging customer that will potentially cancel their subscription. This was calculated by studying the trends of who was offboarding from our platform. In HotJar, I'd assess:
When was their last session?
What were they consuming last?
Did they watch multiple modules of a course or struggle to lock in on one particular piece of content?
Did they watch the entire piece of content?
What is the average time spent on the platform?
How much is spent consuming content vs searching?
Was there negative feedback or engagement with customer support?
This gave us a picture in our B2C customer personae, allowing us to better predict who was (and who wasn't) getting sufficient value in their investment / purchase.
This was assessment was shared with our product team. And in result, we decided to create a system on the backend of the platform that marked each at-risk profile with a "profile attribute" (only visible to us) for future steps of reporting and/or mitigation.
Engaging At-Risk Customers with AI-Infused Content
From my earlier, generative research, I was able to identify common barriers our customers were facing in finding or accessing value from the platform. These barriers were summarized into the following:
Not finding relevant training content
Content wasn't compatible to environmental circumstances
Simply not perceiving growth / value from engagement with platform.
Having this understanding, I presented a workshop for stakeholders and posed the question...
"How might we re-engage with at-risk customers, and support them to learning of more potential value of the platform that is relevant to them so that they are less likely to cancel their subscription?"
In result of this, we built an automated application that consumed newly-marked "at-risk" customers, and created a formatted email with natural language processing (NLP) that provided relevant information that could preserve the users' interest in the platform by:
Reminding them of the progress they've made on the last course content - and the potential benefits of learning this content.
Alternatives to this content that may be more compatible or relevant to the users' desires.
Reinforce the community-first business principle to show how we care about our customers.
Anatomy of Our Solution
I worked closely with the dev team to align on what defines an at-risk user, and how do we handle their profile annotations when the user returns to the platform. This collaboration was to also reduce undesirable circumstances, such as a person receiving too many emails within a short period of time. This would overwhelm a customer, easily driving them to consider offboarding. We aligned on a conversational design of the email, defining the voice and tone and layout of the copy, as well as a simple format in how we address the goals of the email campaign.
Providing Alternative Options
During the redesign of our catalog page, we updated the I/A of our content data model to support better resolution to the metadata about each content group. One of these updates involved vectorizing relations between courses, topics, and contexts. This came handy in powering an MVP in guiding the user to discover new (to them) content.
Measuring Success
Prior to this rollout, we tracked a baseline, average metric that 7 accounts would display intent to cancel their license each day, by initiating the cancellation flow. Many of these users determined that they couldn't make use of the subscription. After launching our campaign that reminds the user about their investment. These furthermore, suggest relevant training content the recipients could enjoy. This reduced the average to 4 accounts considering cancellation a day.
Lessons Learned
While I was confident in the need of multi-modal engagement to resuscitate at-risk personas, I was very impressed by the immediate impact this gave to our metrics. Not only did this reduce churn, but allow us to come closer to our CEO's goal of sustaining 10k subscribers by lengthening the average lifespan of monthly subscriptions. This got me inspired to look into wondering
How might we ethically encourage monthly subscribers to upgrade to an annual plan?
What the leadership team has recently learned from this was:
they began to feel the impact from recently increased activation rates
delaying the need to address poor UX of the B2B buyer/advocate personae can cause high-volume churn
Conclusion
Thank you for your time and interest. This case study was very internally rewarding, so that I could help people make the best with their investment in themselves. Obviously, I'm also pleased that I could help the team get closer to their goals in faster time. Understanding user needs is so important when designing a great product / service. And sometimes, there doesn't need to be a flashy interface.
