Churn Reduction with Braze

Braze gives you the infrastructure to build a real churn reduction system — one that identifies at-risk users in real time and intervenes before they leave. Here's how to set it up.

The Foundation: Events That Signal Churn

Before you build any Canvas in Braze, you need the right custom events flowing. Generic events like "session start" aren't enough for churn prediction.

Set up these custom events:

• feature_used with properties for feature name and depth of usage
• subscription_status_changed — tracks downgrades, pauses, and cancellations
• support_ticket_created — frustrated users churn at higher rates
• payment_failed — involuntary churn signal
• usage_milestone — positive signal that the user is getting value

Plus custom user attributes:

• days_since_last_active — recalculated daily via a data pipeline or Braze webhook
• activation_status — whether the user has hit your activation milestone
• engagement_tier — power user, regular, casual, at-risk (updated via Braze segments)

Building the Churn Reduction Canvas

Flow 1: Early Warning Intervention

Trigger: User has been active in the past 30 days but shows declining engagement (fewer sessions this week vs. last week).

Step 1: Wait 48 hours to confirm the trend isn't a blip.

Step 2: Send a push notification with a personalized value reminder. Use Connected Content to pull in something specific — their progress, a recommendation, or a feature they haven't tried.

Step 3: If no activity in 72 hours, send an email with a different angle. Don't repeat the push message. Offer a new reason to return.

Step 4: If still no activity after 7 days, move to the re-engagement Canvas.

Flow 2: Pre-Cancellation Save

Trigger: User visits the cancellation page (tracked via custom event).

Step 1: Fire an in-app message immediately with a save offer. This could be a pause option, a plan downgrade, or a limited discount. In-app is critical here — the user is literally on the cancellation page.

Step 2: If they proceed with cancellation, send a follow-up email within 1 hour asking for feedback. Include a "changed your mind?" link that restores their subscription.

Step 3: Enter the user into a win-back Canvas set to fire at 30, 60, and 90 days.

Flow 3: Dunning Recovery

Trigger: payment_failed event.

Step 1: Immediately send a push notification letting the user know their payment failed. Keep the tone helpful, not alarming.

Step 2: Send an email within 2 hours with a direct link to update payment info. Braze Connected Content can check if the card was updated and skip this step if so.

Step 3: Retry payment at 3, 5, and 7 days (configure this in your payment processor, not Braze).

Step 4: Send a "last chance" email 1 day before the grace period ends.

Braze-Specific Features to Use

Intelligent Timing: Let Braze optimize send times per user. For churn reduction specifically, this matters because you want to reach users when they're most likely to re-engage, not when it's convenient for you.

Frequency Capping: Set a cap of 1 push + 1 email per day for users in churn reduction flows. Over-messaging an already-disengaged user accelerates churn.

Content Cards: For at-risk users who are still opening the app, Content Cards are less intrusive than in-app messages. Show a personalized card highlighting what they're missing.

What to Measure in Braze

• Canvas conversion rate — percentage of users who re-engage after entering the churn reduction Canvas
• Save rate — percentage of users who visit the cancellation page but don't cancel
• Dunning recovery rate — percentage of failed payments recovered
• Incremental retention — compare churn Canvas recipients vs. control group

Frequently Asked Questions

How many messages should a churn reduction Canvas have?

Keep it to 3-5 touchpoints over 7-14 days. After that, if the user hasn't re-engaged, move them to a longer-cadence win-back flow. Bombarding disengaged users does more harm than good.

Should I use Braze segments or filters for churn prediction?

Use segments for broad targeting (e.g., "at-risk users") and filters within Canvas for more granular logic (e.g., "at-risk AND hasn't received a push in 3 days"). This keeps your Canvases manageable.

Can Braze predict churn automatically?

Braze has predictive features (Predictive Churn) that use machine learning to score users by churn likelihood. It's useful as a starting signal but shouldn't replace your own behavioral triggers. Combine Braze's predictive score with your custom event data for the best results.