Using AI and Automation to Transform Membership Engagement
When our client Aurras approached us about revamping their membership model, we knew we had a lot of pre-work. We knew we needed real, actionable data about how their members engaged with their content. They had a mix of free and paid members. They had some data about their users and members, but it was scattered, and nothing told us how engaged their customers were (or weren’t). Without clear insights into engagement patterns, making informed decisions about pricing tiers or identifying which free members were prime candidates for conversion was impossible, and any targeted marketing campaign would fall flat.
Real quick, Aurras is the best one-stop source for anything related to Sound Healing and Sound baths. Dorothy Tannahill-Moran is an expert in Vibrational Sound Therapy, and Aurras has grown into a precious resource.
The challenge was clear: we needed to understand who was logging in and who was engaging with Aurras’s extensive library of sound healing content. Which free members were power users? Which paid members weren’t getting value from their subscriptions? How could we use this information to create a more sustainable membership model?
The Stack
Aurras used MailChimp, which was fine for a time but ultimately limited given our automation needs for the work we needed to do and scoring we needed to implement—we switched her over to ActiveCampaign (DigiSavvy is a certified ActiveCampaign Consultant) when we turned to ActiveCampaign and WP Fusion. While ActiveCampaign offers its own lead scoring capabilities, we didn’t have scoring configured as we had just moved Aurras over to AC, thus we needed to set up our own scoring framework. For this to work, we needed to track not just email opens and website visits but particular interactions with sound healing content, event attendance, and resource downloads.
I’ll talk about how we used engagement scoring as a strategic tool to:
- Identify highly engaged free members for conversion
- Flag at-risk paid members for re-engagement
- Inform pricing tier decisions based on usage patterns
- Drive personalized communication based on engagement levels
I’ll also walk you through how we implemented this system and, more importantly, how it’s helping Aurras make data-driven decisions about their membership model.
Let’s start with how we built the scoring framework…
Building Our Engagement Scoring Framework
Looking at Aurras’s data, we had a goldmine of information sitting in their tagging structure. Their WordPress site was already tracking various user actions with tags like [Action], [Activity], and [Event], giving us a solid foundation to build upon. I gotta say a big thank you to WP Fusion for that, which made assigning these tags a breeze.
Understanding the Data
First, we analyzed our client’s existing tags to identify patterns. The key categories we found were:
- Content Consumption (played recordings, downloads)
- Event Participation (sound baths, workshops)
- Account Activity (logins, profile updates)
- Resource Engagement (guides, meditations)
Creating the Scoring Structure
We built a 100-point scoring system with three key engagement levels:
High Engagement (70+ points)
- Regular content consumption
- Event attendance
- Multiple resource downloads
- Consistent platform usage
Moderate Engagement (30-69 points)
- Occasional content consumption
- Some event interest
- Basic platform usage
- Newsletter engagement
Low Engagement (0-29 points)
- Minimal platform usage
- Few or no content plays
- Limited event interest
- Infrequent logins
Point Assignment Strategy
We weighted actions based on their correlation with member retention and conversion:
High-Value Actions (Don’t Expire)
- Completed Onboarding: 20 points
- Event Registration: 15 points
- Account Creation: 15 points
- Free Membership: 10 points
Content Engagement (30-day decay)
- Guide Downloads: 5 points
- Sound Bath Recording Plays: 3 points
- Quiz Completion: 5 points
- Newsletter Signup: 3 points
Regular Activity (15-day decay)
- Email Opens: 1 point
- Email Clicks: 2 points
- Website Visits: 3 points
Tag-Based Intelligence
The beauty of our tagging strategy was how it enabled pattern recognition when we sent it to the AI. Tags like:
[Action] Played Sound Bath Recording[Event] In-Person Registrant[Activity] First Login
Created a detailed engagement narrative for each user. By feeding this structured data into our scoring system, we could:
- Identify engagement trends
- Predict likely member behaviors
- Trigger appropriate automations
- Target communications effectively
We took all our active contacts, exported them to a CSV file, and fed them to the AI, in this case, Claude.ai—I highly recommend it btw, primarily for it’s ability to let you use projects with custom instructions and knowledge base per project.
Sidenote: When it comes to tagging, it’s really easy to let it get out of control. I advise being generous with your tagging but making the context-specific as much as you can. In my case, this is how I’ve always set up tags, in the example above. For me and the people I’ve worked with, it’s easy to see what interests or actions people have taken. And in this era of AI, we can take that context and make meaningful inferences as to what the tags mean, and we can do it quickly and at scale. So, be mindful of your tags—prune where possible and always assign context-specific tags.
The Role of AI
This structured tagging approach didn’t just help with scoring – it created a rich dataset for AI analysis. We could now identify:
- Common paths to conversion — It was easy to find common patterns in users and what content they consumed and we could, although we haven’t yet, further segment users into their specific interests (stress, poor sleep, anxiety, etc.). We can better target our messaging.
- Early warning signs of member churn — We could use this framework to also figure out how long between engagements users have made with Aurras’s content and what they’ve consumed, we can make some educated guesses about paid subscribers and if they might churn or not. This gives us an opportunity to target more value towards this segment of user.
- Content preferences by engagement level
- Optimal timing for upgrade prompts
This was a massive W for us in terms of the time saved in creating the analysis from which we built our framework. Rapid fire fist pumping and loud noises
Turning Data Into Action: Implementation and Automation
Once our scoring framework was defined, we needed to tackle the difficult task of analyzing the data and segmenting their customers. How did we do that? Rather than starting from scratch, we wrote a Python script (again using AI, thanks to Claude.ai) to analyze their existing user base from the output CSV file and segment them based on historical engagement data (tagging data).
Python Script to Analyze and Segment Users – the script is generated here.
We ran the script inside a folder with our CSV files, and it generated a new set of CSV files with segmented users in them. It reads the CSV file we output from ActiveCampaign to run its “magic.”
This gave us easy files to import into ActiveCampaign where we could add additional tags and then list and further segment each group of users.
Initial Data Analysis
Our Python script processed Aurras’s member export and categorized users based on their existing tags and activity history. This gave us three clear segments:
- Highly Engaged Free Members: Users showing consistent content consumption, event participation, and platform engagement
- Active Paid Members: Current paying members with varying engagement levels
- Low-Engagement Free Members: Users who signed up but showed minimal platform interaction
Building Targeted Campaigns
Rather than using a one-size-fits-all approach, we created specific pathways for each segment. Highly targeted emails will always be more successful and drive greater engagement than the ol’ spray-and-pray approach. If you mind the details, your customers won’t mind hearing from you. =)
Highly Engaged Free Members
These users were already getting significant value from the platform, making them prime candidates for conversion. We:
- Offered a special “power user” discount on annual memberships
- Highlighted advanced features they weren’t accessing
- Emphasized the value they were already receiving
Active Paid Members
The goal here was retention and potential upgrades. Based on their engagement level:
- High engagement → Offered premium tier upgrades with special pricing
- Medium engagement → Focused on feature education and usage tips
- Low engagement → Initiated a “rescue” sequence to prevent churn, including a downsell option to a lower-priced tier if needed
Low-Engagement Free Members
Instead of pushing these users toward premium offerings, we:
- Introduced a lower-priced tier as an entry point
- Created an onboarding sequence to demonstrate platform value
- Used content recommendations based on their limited engagement history
Automation Setup
We set up a couple of types of automation for this campaign:
- We set up automated campaigns for sending time-delayed emails—we had to start manually for some of those emails. Some are started when users finish a campaign and receive a specific tag.
- As mentioned, the other automation begins when a user doesn’t take a conversion-related action—WP Fusion tags users who sign up for a specific automation.
Results and Insights
This segmented approach proved more effective than previous broad-based campaigns:
- 23% of highly engaged free members converted to paid plans
- 15% of at-risk paid members moved to the lower-priced tier (instead of canceling)
- Improved engagement rates across all member segments
Looking Ahead
The system continues to evolve. Our next steps include:
- Leveraging ActiveCampaign’s built-in scoring capabilities will reduce the time needed to create these campaigns in the future.
- Creating more helpful automation to deliver valuable info and resources to members accessing specific content types.
- Set up an analysis of specific actions that indicate possible churn, e.g., time logged out of the site, time not interacting with the website, emails, or events.
All in all, segmenting and analyzing the data didn’t take too much time. Using AI to analyze large sets of data makes these types of tasks fairly short and easy to navigate. Use native scoring where possible, but if you don’t have that data, then using AI to parse through your own datasets is the way to set your business aside and begin generating valuable content that produces leads.
Get Notified When We Publish New Content!
Join more than 2,500 people who get our marketing automation, business marketing, and WordPress news!
