Braze Customer Engagement Design
How to design lifecycle and cross-channel engagement in Braze with Canvas, Campaigns, channels and BrazeAI, using the names and statuses Braze documents.
Braze gives you a lot of channels and a capable journey builder. The hard part is deciding which message should reach which person, on which channel, at what point in their lifecycle, without flooding them. This page explains how Braze structures engagement, which channels and AI features its documentation lists, and how to design a lifecycle programme that uses Salesforce data where it helps.
The business problem
Engagement programmes tend to grow one campaign at a time. After a year there are dozens of overlapping messages, no shared frequency rules and no clear owner for each lifecycle stage. Customers get a push, an email and an SMS for the same thing. The fix is a lifecycle design that decides stages, entry rules, channel order and exits before anyone builds a Canvas.
How Braze structures engagement
Braze describes Canvas as "a unified interface where marketers craft campaigns with multiple messages", a visual journey builder. A Campaign engages users "with a single message step", and Braze's own guidance is that "most multi-step user journeys are better created as Canvases." Both support three delivery and entry types: Scheduled, Action-based and API-triggered.
Canvas components include Action Paths, Agent, Audience Paths, Canvas Audience Sync, Experiment Paths (with Winning Path), Content Optimizer, Context, Decision Split, Delay, Feature Flag, Message, Send to Destination and User Update.
Channels listed in the Braze documentation
| Group | Channels |
|---|---|
| In-product | In-app messages, Content Cards, Banners |
| Out-of-product | Email, Transactional email, Landing pages, LINE, Live notifications, Push (including web push), "SMS, MMS, and RCS", Webhooks, WhatsApp |
| Also in the docs sidebar | KakaoTalk |
A lifecycle design, stage by stage
| Stage | Typical trigger | Braze tools | Salesforce input |
|---|---|---|---|
| Welcome and onboarding | Sign-up or first session | Action-based Canvas, Delay, Message | Account tier or owner, if relevant |
| Activation | Key action not yet taken | Audience Paths, Decision Split | Contract or plan details |
| Habit and growth | Usage patterns | Segments, Content Cards, Catalogs for recommendations | Products held |
| Retention | Churn risk | Predictive Churn, Intelligent Timing | Open cases, to suppress promotional sends |
| Reactivation and win-back | Inactivity window | Scheduled Canvas, Experiment Paths | Opt-out and complaint status |
Cross-channel design rules
- Consent first. Subscription groups "control which users can receive messages from a specific set of sending resources within a channel". SMS, MMS and RCS, WhatsApp and LINE require groups; email groups are optional. SMS has no global state, only per-group status.
- Choose the channel deliberately. In-product channels suit people who are active; push and email reach people who are not. The Intelligent Channel filter can pick a channel per user where you are happy to let it.
- Personalise from governed data. Braze supports "Liquid 5 from Shopify". Catalogs enrich messages from CSV files and API endpoints, with up to 500 MB on the free tier. Connected Content calls must answer in under 2 seconds and return no more than 1 MB.
- Exit rules matter as much as entry rules. Remove people from promotional journeys when CRM shows an open complaint or a purchase.
BrazeAI features (names as Braze lists them)
BrazeAI features and what to check
| Feature | What it does | Check before relying on it |
|---|---|---|
| Operator | Dashboard assistant; generative AI is accessed through it | Data access and review process |
| Braze Agents | Run through the Agent Canvas step | Braze says Agents need Message or Action Credits |
| Optimize with BrazeAI | Optimisation features | Which apply to your package |
| Winning Paths | Sends the winning variant after an experiment | Test size and duration |
| Content Optimizer | Content testing | Labelled beta |
| Decisioning Studio Go and Pro | Go is self-serve, email-only and tuned to clicks; Pro supports any channel, data source and metric | Pro is set up with Braze's AI Decisioning Services team |
| Intelligent Timing and Intelligent Channel filter | Per-user send time and channel | Enough engagement history |
| Predictive Suite | Predictive Churn and Predictive Events | Event volume and definitions |
| Item recommendations | Recommends catalogue items | Catalog quality |
| Braze MCP server | Lets AI tools reach non-PII Braze data | Generally available per Braze (23 Jul 2026); see Braze MCP integration |
Data inputs, security and failure handling
Salesforce data reaches Braze through an integration layer or a warehouse sync; the patterns are on Salesforce and Braze integration. Keep API keys server-side and use Braze's per-endpoint key permissions. When a Connected Content call times out, the message needs a fallback, so design one for every external lookup. Webhook steps that write to other systems should log failures somewhere a person will see them.
Deployment, testing and monitoring
Build in a test workspace, then promote. Test every Canvas branch with seeded test users, including opted-out users, users with no CRM record and users who meet two entry rules at once. After launch, watch entry and exit counts, conversion against a control group and unsubscribe rates by stage. Our journey QA checklist applies to Canvas as much as to Journey Builder.
Prerequisites and limitations
Feature access depends on your Braze contract: Braze says a Currents connector is included in many pro and enterprise packages, Catalogs Pro adds storage tiers, Braze Agents need Message or Action Credits, and Decisioning Studio Pro is set up with Braze's AI Decisioning Services team. Content Optimizer is beta. Check current Braze documentation for anything your design depends on.
Implementation stages
- 01
Map the lifecycle
Agree stages, entry and exit rules, and owners.
- 02
Set consent rules
Define subscription groups and frequency limits per channel.
- 03
Design the data
List attributes, events and Salesforce fields each stage needs.
- 04
Build Canvases
One Canvas per stage, with Experiment Paths where you want to learn.
- 05
Test
Seed users, check every branch, then send to a small live audience.
- 06
Review
Compare each stage with its control group and retire what does not work.
Related
Braze consulting
Lifecycle and platform advice
→Braze implementation
Build Canvases, data and channels
→Welcome and onboarding journey
A lifecycle pattern to adapt
→Re-engagement and win-back
Reactivation design
→Questions we hear
Should we build journeys as Canvases or Campaigns?
Is Content Optimizer generally available?
Do Braze Agents cost extra?
Can Braze use Salesforce data in messages?
Sources
Design your lifecycle programme
Share your lifecycle stages and channels. We can review your Canvas design and the data each stage needs.
