WallaPredict · AI Retention

AI that knows which members are about to leave, before they do.

Half of new members quit within six months. You usually find out from a refund request. WallaPredict scores every member's risk daily, surfaces the ones who need a call this week, and tells you what to say. Built with AWS, trained on boutique fitness behavior.

Member at riskSample · fictional studio
SK
Sarah K.
Unlimited Monthly · 14 months tenure
92
Risk score
Why she's at risk
Booking cadence dropped 70% in the last 21 days
Last visit 18 days ago, was a 3x/week regular
Plan renews in 14 days · auto-charge scheduled
10 weeks agoThis week
Recommended action
Personal text from Tess K. (her favorite teacher). Mention Wednesday's Slow Flow.

Retention without the spreadsheet.

Stop running a Friday-afternoon report to figure out who's slipping. WallaPredict surfaces the at-risk list daily, ranks who matters most, and tells you what to do about it.

See risk before it costs revenue

Daily-updated risk scores per member based on booking cadence, plan proximity, teacher patterns, and engagement velocity. The signals that actually predict churn, not the ones that look impressive in a slide.

Get the next step, not just the alert

Every at-risk member gets a recommended action. Personal text from a favorite teacher. Re-engagement offer. Win-back sequence. WallaPredict tells you what works, who to send it to, and when.

Built on AWS, trained on boutique

Enterprise machine learning infrastructure from AWS, trained on boutique fitness behavior, not generic SaaS churn data. The retention math your studio actually needs, with the reliability your data deserves.

Risk scores you can explain to a member.

Every score is a sum of explainable factors. Booking cadence dropping. Teacher pattern broken. Plan renewal approaching. You can see exactly why a member is flagged, which lets your front desk make the call with confidence.

Behavioral signals
Risk scoring
Cohort comparisons
Daily updates

The signals that actually predict churn

Booking velocity, last-visit recency, favorite-teacher patterns, plan proximity, cancellation history, social signals from the app. The features Walla trained the model on, surfaced as readable factors per member.

Scores that map to action thresholds

Scores from 0 to 100. Above 70 triggers an outreach prompt. Above 85 is critical, ranked at the top of the day's call list. Below 30 is healthy and stays out of the way.

Compare risk patterns across cohorts

See risk patterns by plan type, teacher, modality, and tenure. Which intro cohort retains better, which teacher's clients show the lowest churn. The cohort data that used to mean exporting from Studio Management, surfaced natively.

Updated every night, ready before coffee

The risk model runs nightly on the previous day's behavior. By the time you open Walla in the morning, the at-risk list is current. No manual refresh, no waiting for batch jobs.

Risk scores you can defend to a member who asks.
Risk factor breakdownSarah K.
Total risk score
Updated 6:02 AM today
92
Critical
Booking cadence drop
From 3x/week to 0 visits in 18 days
+38
weight
Last-visit recency
18 days since last class, longest gap in 14 months
+28
weight
Favorite teacher absence
Skipped 4 Tess K. classes she'd usually book
+16
weight
Auto-renewal proximity
Plan renews in 14 days, at $189
+10
weight
Sample breakdown · fictional member

Ask the questions you used to spend Friday afternoons answering.

WallaPredict turns the retention question into a conversation. Type what you need, get a ranked answer, and the recommended next step.

Ask
Show me at-risk members
Who should I call today?
Which intro members didn't convert?
Top 3 at-risk members this week23 total flagged
SK
Sarah K.
Unlimited Monthly · 14 mo tenure · Booking cadence dropped 70%
18 dayssince last visit
92
JR
Jamal R.
10-class pack · 8 mo tenure · Skipped 4 favorite teacher classes
12 dayssince last visit
87
ML
Maya L.
Unlimited Monthly · 6 mo tenure · Auto-renewal in 14 days, no recent attendance
22 dayssince last visit
76
Recommended next step
Send personal texts from each member's favorite teacher this morning. Past 7-day data shows 64% re-engagement on this type of outreach.
Priority outreach for today3 calls suggested
SK
Sarah K.
Mention Wednesday's Slow Flow · "We've missed you, miss the bench by the door?"
Risk 92renews in 14 days
1
JR
Jamal R.
Offer to switch to Unlimited · He's been hitting class cap on his 10-pack
Risk 87value upgrade signal
2
ML
Maya L.
Free pause month offer · Lower-touch outreach since she's only 6 mo tenured
Risk 76soft re-engage
3
Recommended next step
Start with Sarah K. Her plan renewal is the closest action window. The longer the gap, the harder the save.
Intro-offer members who didn't convert8 from last 30 days
DK
Devon K.
Used 8 of 12 intro classes · Loved Tess K. Slow Flow · Didn't see a plan that fit
$79 introended 3 days ago
68
PT
Priya T.
Used 5 of 12 · Bookings clustered on weekends · 4-pack or weekend-only plan fits
$79 introended 7 days ago
62
LR
Luca R.
Used 11 of 12 · Hit class cap mid-intro · Likely Unlimited candidate
$79 introended 10 days ago
81
Recommended next step
Send Luca R. an Unlimited Monthly offer with first month at 50% off. The intro attendance pattern is the highest converting signal in your data.
Sample interaction · fictional member data for visualization. The Ask interface is in beta.
Built on AWS · Enterprise machine learning infrastructure trained on boutique fitness behavior
Risk-triggered journeyMember crosses risk 70
0h
Personal text from favorite teacher
"Hi Sarah, it's Tess. Saw you've been MIA, all good?"
SENT
3d
Re-engagement email with class recommendation
Mentions her favorite class times, includes one-tap booking link
QUEUED
7d
Pause month offer (if no booking yet)
Lower-friction save than a full discount. Preserves plan value.
CONDITIONAL
Track outcome & learn
If she rebooks, journey closes. If she churns, win-back sequence starts at 6 weeks.
EVERGREEN
Sample journey · fictional member

From alert to action in one click.

Knowing who's at risk only matters if you act on it. WallaPredict connects to your Walla automations so the right re-engagement message ships to the right member at the right time, without a separate marketing tool.

Automated journeys
Manual outreach
Re-engagement offers
Win-back sequences

Risk-triggered automated journeys

When a member crosses a risk threshold, trigger a sequence: personal teacher message, then a re-engagement offer, then a final outreach attempt. Built into Walla's Marketing Suite, no separate automation tool.

Daily outreach list, prioritized

The front desk gets a ranked list of who to call or text personally today. Each member arrives with their recommended talking point and the right teacher's name attached. The same list rolls into Reporting, and when a call is the right move, Walla Voice can place it.

Re-engagement offers that target the right members

Send a 20%-off month to members who fit the "would respond" profile. Don't blast everyone, don't train your members to wait for promos. The right offer to the right segment.

Win-back sequences for canceled members

When a member churns, WallaPredict still tracks them. Six weeks later, the right teacher reaches out with a "we've missed you" message. The win-back rate compounds over a year.

Alert plus action plus tracking, in one platform.
This is a game-changer! WallaPredict is replacing our complex spreadsheet KPIs. You all rock!
Justin RandolphOwner, Sol Seek Yoga

Pairs well with the rest of Walla.

WallaPredict is sharper when the platform behind it is connected.

What you'll need.
Included on Core ($320/mo) and Pro ($599/mo)No additional cost · no per-member fee"Ask WallaPredict" interface in beta

Common questions, plain answers.

How is the risk score calculated?
A weighted blend of behavioral signals trained on boutique fitness data: booking cadence trend, last-visit recency, favorite-teacher pattern, plan renewal proximity, cancellation history, app and email engagement. Each member's score breakdown is visible in their profile so you can see exactly why they're flagged.
How often does it update?
Risk scores recalculate nightly based on the previous day's behavior. By the time you open Walla in the morning, the at-risk list is current. No manual refresh, no batch jobs to monitor.
What does "trained on boutique fitness" mean?
WallaPredict was trained on real attendance, plan, and churn data from Walla's customer base of boutique studios across Yoga, Pilates, Lagree, Barre, Cycling, and HIIT. Not generic SaaS churn models or gym-chain data. The factors that predict churn in boutique fitness are different, and the model reflects that.
Is the data secure?
Yes. WallaPredict runs on AWS infrastructure with enterprise-grade security. Member data stays within your studio's account. The model trains on aggregate patterns, not identifiable member data, and your records are never shared with other studios.
Can I act on the risk list without leaving Walla?
Yes. Each at-risk member's record includes a recommended next step, and you can trigger an automated journey, send a manual text, queue an email, or assign the outreach to a teacher, all from inside Walla. No CSV export, no separate marketing tool.
Does WallaPredict cost extra?
No. WallaPredict is included on Core and Pro plans at no additional cost. No per-member fees, no usage tiers, no separate AI add-on. Retention intelligence is part of the platform.
What is the "Ask WallaPredict" interface?
Ask WallaPredict is a conversational interface where you type questions about your members ("Who should I call today?", "Which intro members didn't convert?") and get ranked answers with recommended next steps. It's currently in beta with a subset of Walla customers and rolling out gradually.

Stop finding out from the refund request.

Book a Demo →