Member Health is now live across eligible Novi AMS sites. After 10+ months of research and development, two data scientists, a patent application, and invaluable feedback from Alpha and Beta testers, Member Health answers the question most teams ask: “Which members are most likely to renew - and who needs attention?”
Rather than a DIY “member engagement points” system, Member Health uses statistically tested machine‑learning models trained on anonymized historical renewal outcomes to classify each current member as On Track, At Risk, or New Member. You get simple labels in your workflow and a transparent report of the key signals behind each label.
Watch the Recording
Due to the proprietary nature of the feature, you must log in to watch the recording as it is available to Novi Customers only. If you are a customer and need assistance with your Novi HQ login, let us know!
In this recording, we discuss:
The journey to Member Health
How the model works
What signals are included, and what's not included
Review of the features available to you, today
Roundtable discussion of real-life use cases and integrating Member Health into your day-to-day work
Global Data Trends
Roundtable Recap
Why We Built It
Association Executives told us NPS is a great feature for understanding members' sentiment, but their staff still needed a way to focus their limited time before renewals. Member Health bridges the gap by predicting renewal likelihood and surfacing the why behind each prediction so you can act with confidence.
How It Works
Multiple models: There’s a global model, an association specific model (your data only), and a per‑member‑type model (because your suppliers behave differently than your regular members, for example).
Blended Ensemble: These models are blended together uniquely for each association, with carefully tested weightings. This captures different dimensions of member behavior and ensures each association has the highest accuracy possible, with its current data.
Trained & backtested: Models train on ~1.2M member records across 79+ data points, then are backtested against prior years to verify performance.
Recalculation cadence: Statuses currently refresh every 30 days to emphasize big picture trends over day‑to‑day noise.
Security: Models run on Novi; your data never leaves Novi servers. This is not an LLM like ChatGPT.
Key Data Points
What’s Included?
Membership Profile
Membership Dates & Renewals
Events
Transactions
NPS
Census
Committees
What’s Not Included?
Industry factors
Macro-economic factors
Custom & 3rd Party Activities
Website Analytics
Qualitative Data (like open NPS feedback)
Offline conversations
Manipulated Data (manually manipulating member dates & dues invoices)
Where You’ll See Member Health in Novi
List Views: Enable the Member Health column via the gear icon. Use the filter to focus on New, At Risk, or On Track.
Groups: Build dynamic groups from Member Health status to power campaigns in Mailchimp/Constant Contact or other tools.
Member Record: A status badge appears beside NPS for current members. Click it to open the Member Health Report with the top healthy and warning signals for that record.
What the Member Health Report Signals Mean
Signals are correlations the model has found across your data, not guarantees and not hand‑tuned “weights.” A single signal rarely tells the story; it’s the unique combination that points to risk or health. Use them as conversation starters and checklists, not verdicts.
How Teams are Already Using Member Health
Real‑time insight on calls: When a record shows At Risk or New Member, stay on the phone longer and dig into the signals to resolve issues in the moment.
Daily micro‑outreach sprints: 15–20 minutes each day calling At Risk members; log timeline notes; use the report to guide a friendly “I’m updating your record…” conversation.
Pre‑renewal head start: Divide the list of At Risk members among staff & volunteers for personal check‑ins and resources before invoices are sent.
Budget preparation: Use aggregate Member Health to sanity check renewal projections for upcoming fiscal planning.
Calibrated response: Treat statuses as guidance, then validate through regular touchpoints before spinning up full campaigns.
FAQ
Org + Individual membership? We've got you covered. Models run per member type, understanding both company and individual memberships, so hybrid associations are supported.
Grace periods & late payers? If a member lapses then renews, they'll either be treated as new or as a renewal based on your Member Type Settings; future work will factor “days to pay” as an additional signal.
New members feel fragile. Can the model judge earlier? Today, “New Member” covers the first 12 months (where data is most sparse, these members are inherently "At Risk" already). We plan to explore predictions in the second half of year one as reliability improves.
Can we see how our accuracy is trending? We’re tracking history internally today and exploring ways to expose change logs and org‑level accuracy dashboards.
How often do Member Health statuses update? Currently, once a month.
What’s Next
We’ll continue expanding inputs (e.g., last login, days‑to‑pay, selective third‑party data), improving explainability, and exploring the right way to display the history of status changes. If your results look “off,” reach out! Data nuances matter, and we’re happy to review.
Have feedback or a use case to share? Ping us in Intercom (Blue Bubble), that's the fastest way to help shape the roadmap.
Thanks to our alpha and beta testers for invaluable feedback throughout the development process, and to everyone who shared strategies during the roundtable.