UnionVision ML Research
AI/ML platform for trade union research: member churn prediction, remittance outlier detection, and contractor performance analysis. Built 2023 to 2025.
Trade unions steward large, diverse memberships whose engagement signals are scattered across dues systems, event attendance, and communications tools — with no single lens on who is disengaging.
Trade unions needed to understand member engagement patterns and identify at-risk members before attrition became irreversible.
We built a research and analytics platform that applies machine learning to member engagement data, surfacing churn risk and actionable outreach signals.
Early intervention workflows contributed to a 22% reduction in member churn.
Union member data is small, sensitive, and personal: every model output touches a real member or contractor, so results have to be explainable to the leadership who act on them. And the signal is buried in simple records: each transaction row holds only a member number, a contractor number, a work date, hours worked, and amount paid. Churn risk hides in the patterns across those rows, not in any single field.
The models read the data the union already had rather than requiring new collection: 500,000 work-transaction rows accumulated since 1999 in the union's member hour-bank system, a system Mindfulware's founder originally built. Twenty-seven years of continuous, consistent records is a training corpus most organizations cannot buy, and it existed because the 1999 system was built to last.
The platform computes a risk-of-churn percentage for each member, giving leadership a heads-up on who needs a call so issues get addressed before they escalate and a member leaves the union.
Key results
- Predictive member engagement modeling
- Churn risk scoring and alerts
- Outreach prioritization dashboard
Technology stack
Want a similar outcome for your firm?
Book a 30-minute consultation with our founder to discuss your data, compliance, and AI needs.