Case study · Fitness

Know who's about to leave, while you can still help

Membership businesses lose most members quietly: visits drop, then the cancellation arrives. This is how we predicted which members are at risk from attendance and billing patterns, and trigger retention actions whose effect is measured properly.

Start a project like this See more of our work

Project overview

Churn prediction with measured retention.

Industry: Fitness & membership
Client: Confidential
Solution: Churn prediction
Platform: Azure ML + CRM
Timeline: 8–14 weeks
Team: 3 people
Signals: Visits, billing, classes, tenure
Actions: Coach outreach, offers, reminders
Measurement: Control groups

Expected business outcomes

What the client set out to achieve

Only retention measured against a control group counts.

2–6 pts Lower monthly churn in targeted groups
At-risk members contacted early with the right action.

Target range, measured against a control group
Weekly Updated risk scores
Every member scored weekly from recent behaviour.

Automated scoring pipeline
Measured Effect of each retention action
Actions compared with a held-out group.

Randomised control groups

Business challenges

Where the client started

Churn is noticed after it happens.

Signals

Declining visits are visible in data nobody looks at.

Retention

Offers sent to everyone or to nobody.

Clubs

Staff don't know which members need attention.

Measurement

Retention campaigns judged by feel.

Economics

Winning back a member costs far more than keeping one.

Proposed solution

What the build covers

A prediction model is only useful with actions and measurement attached.

Data pipeline

Visits, bookings, billing and tenure prepared weekly.
Risk model

Churn risk score per member with main reasons.
Club action lists

Club staff see at-risk members with suggested actions.
Automated journeys

Reminders and offers triggered for specific risk groups.
Control groups

A share of at-risk members held out to measure effect.
Retention dashboard

Churn, risk and action results by club.

Project requirements

What the client asked us for

The request was to predict churn. The value comes from what happens next: the right action for the right member, measured against a control group.

Reasons matter: a member who stopped visiting needs a different action from one with a failed payment.

What the project had to deliver

01Weekly risk scores
02Reasons behind each score
03Actions for clubs and journeys
04Honest measurement

Key challenges & considerations

What made it harder, and what we did

Data spread across access, booking and billing systems

What we did

Access, booking and billing data were joined into a single member view.
Seasonal patterns (January, summer)

What we did

Seasonal features were built into the model, so January joiners were not flagged unfairly.
Clubs with different engagement

What we did

Models were checked club by club.
Offers that simply discount loyal members

What we did

Offers were tested against a control group, to avoid discounting members who would have stayed.

What we watched for

Always keep a control group
Show reasons, not just scores
Don't discount members who'd stay anyway
Account for seasonality

Development process

Predict, act, measure

Each action is tested before it is scaled.

Stage 1
Data

Visit, booking and billing data joined and cleaned.
Stage 2
Model

Risk model trained and validated on past cancellations.
Stage 3
Actions

Club lists and journeys designed with operations.
Stage 4
Test

Actions tested with control groups in pilot clubs.
Stage 5
Scale

Effective actions rolled out to all clubs.

Technology stack

Machine learning with CRM actions

Scores flow into the tools staff and marketing already use.

Machine learning

Python
scikit-learn / LightGBM
Azure Machine Learning

Data

Microsoft Fabric or Azure SQL
Feature pipelines

Serving

Batch scoring
REST API
ERP/CRM integration

Monitoring

Accuracy tracking
Drift alerts
Retraining schedule

Quality assurance & deployment

Measured against a test set before launch

AI quality was measured, not judged from a demo. A test set of real examples with agreed answers decided whether the system was ready, and the same set was re-run after every change.

A test set of real examples with agreed answers
Accuracy thresholds agreed with the business before launch
Safety checks: refusals, permissions and prompt-injection tests
A person reviewing low-confidence output
Cost and response time measured at real volumes

Deployment

01TestActions tested with control groups in pilot clubs.
02ScaleEffective actions rolled out to all clubs.
03Pilot, then staged rolloutA small group of real users first, with results reviewed weekly before widening access.
04Monitoring and warrantyAccuracy, drift, usage and cost tracked after launch, and six months of free bug fixing.

Final deliverables

Retention that's measured, not hoped

Every member has a weekly risk score with reasons, clubs know who to talk to and automated journeys handle the rest.

Control groups show which actions actually keep members.

Weekly churn risk
Club action lists
Targeted journeys
Measured retention

Talk about a project like this

Handed over to the client

01Production AI serviceDeployed in the client's own cloud tenant
02Source code and modelsIn a repository the client can access
03Test set and resultsThe evidence behind the launch decision
04MonitoringAccuracy, usage and cost
05Six-month warrantyFree bug fixing after delivery

Success metrics

How success is measured

Measure	Outcome	How it is measured
Lower monthly churn in targeted groups	2–6 pts	Target range, measured against a control group
Updated risk scores	Weekly	Automated scoring pipeline
Effect of each retention action	Measured	Randomised control groups

Verified reviews

What clients say about our custom software work

Verified reviews from our clients on the same kind of work, published on Clutch.

5.0 27 verified reviews

5.0 Custom Software Development

Web-Based Coupon App Development

The work produced by the team was of exceptional quality.

Jon Thies CTO, Model Rocket Feb 2022 – Mar 2023
Verified review on Clutch (Jon Thies, opens in a new tab)
5.0 Custom Software Development

Mobile & Web Platform Development & Design

Silver Scintilla was responsive, respectful, and enjoyable to work with, quickly addressing any questions or requests.

DeShawn Brown CEO & Founder, Lithios Jun 2021 – Nov 2022
Verified review on Clutch (DeShawn Brown, opens in a new tab)
5.0 Custom Software Development

Web-Based HRM System Development

Silver Scintilla provided the best solution with scalable functionality, personalized features, and robust security.

Suheb Khan Director, Collaborative Insight Technologies Jul 2020 – Jun 2022
Verified review on Clutch (Suheb Khan, opens in a new tab)

See All Testimonials Read All Reviews on Clutch

Know who's about to leave, while you can still help

What the client set out to achieve

Where the client started

What the build covers

Data pipeline

Risk model

Club action lists

Automated journeys

Control groups

Retention dashboard

What the client asked us for

What made it harder, and what we did

Data spread across access, booking and billing systems

Seasonal patterns (January, summer)

Clubs with different engagement

Offers that simply discount loyal members

Predict, act, measure

Data

Model

Actions

Test

Scale

Machine learning with CRM actions

Machine learning

Data

Serving

Monitoring

Measured against a test set before launch

Retention that's measured, not hoped

How success is measured

What clients say about our custom software work

Services we used on this project

AI & ML development

AI proof of concept

Gym management software

Have an idea? Let's turn it into AI-powered software.