WallaPredict is an AI-powered churn prediction tool built to help you spot when clients may be slipping away—before it affects your business.

Instead of waiting for someone to stop booking altogether, WallaPredict keeps an eye on key engagement signals and flags clients who might need some extra attention. You’ll see clear, actionable recommendations you can use to reconnect and build stronger relationships.

Think of WallaPredict as an early-warning system that adds context—not a replacement for your experience or your other reports.

At the core of WallaPredict is a predictive model developed in partnership with Amazon Web Services (AWS). Walla Predict looks at a focused set of data points we know are linked to churn, such as:

When a client meets at least one of the defined risk criteria, WallaPredict flags them as At Risk and suggests next steps you can take right away.

Good to know:

If something doesn’t look right or seems inconsistent, using the feedback button is the fastest way to let us know. Your input helps improve accuracy over time.

Default access is determined by user role and location permissions. You can update these as you'd like.

Key Data gives you a clear snapshot of each client’s activity and purchase history—all in one place. This makes it easier to decide if you need to reach out or take other action.

Shows whether the client currently has an active membership or pass, along with usage details.

Shows whether the client has left any feedback.

Note: If no data appears, it just means they haven’t rated a class or left a review yet.

Provides the client’s original lead quality when they joined.

WallaPredict is designed as an early-warning system, not a historical report you can recreate line by line. The AI model uses patterns and probabilities based on a defined set of engagement signals—not every piece of data in your system.

What this means:

If you’re comparing a churn risk to historical data and it feels off, that doesn’t always mean the prediction is wrong—it just means the AI is seeing patterns across many signals that may not be obvious at first glance.

While churn risk shows who may be leaving, the Lead Score highlights how promising a new client is overall.

Each lead is automatically assigned a score out of 100, based on four main factors:

Example: A lead who lives nearby, has a high-income ZIP, strong fitness affinity, and was referred by a member will typically score 80–100.

Lead Scores are meant to add helpful context—not replace your personal judgment. If you know someone personally or have a strong connection, you can absolutely prioritize them regardless of the score.

To generate a Lead Score, each lead’s profile needs to include:

If this information is missing, no score will be generated. The more complete your lead profiles are, the more accurate the scoring will be.

When a client is flagged, you’ll see recommended next steps in their profile. These suggestions are designed to help you take action quickly.

Available Actions:

Note: Clients don’t see any of these labels or scores—these insights are for your team only.

AI predictions are a powerful tool, but they aren’t perfect. The system learns and improves over time based on real usage and your feedback. If you see something that doesn’t look quite right, sending feedback is the best way to help the model get smarter.