Braze Operations Workflow

Proposed owner: Prospect Agent

What It Does

Sam monitors and audits FuturHealth's Braze email and SMS operations. This includes checking canvas status, analyzing flow entry counts, auditing subscriber data, and investigating gaps in communication coverage.

Critical Discovery (March 10 & 15, 2026)

Sam conducted two major Braze audits that revealed significant revenue leaks:

Audit 1: Cart Abandon Flows (March 10)

• Cart abandon CVR: 5.6–6.5% (above 3–5% industry benchmark ✅)
• Email click rates: ~1.5–2.5% (below 2.5–4% benchmark ⚠️)
• SMS coverage gap: #1 revenue leak — phone numbers not collected in quiz funnel
  • Phone is only captured at checkout/intake AFTER quiz completion
  • ~1,800 cart abandoners/day get ZERO SMS recovery messages
  • Estimated value: $50–100K/month opportunity

Audit 2: Critical Data Gap (March 15)

25% of purchasers are invisible to Braze. This is the biggest finding.

Metric	Value
Purchasers in audit sample	44
Invisible to Braze (no match)	11 (25%)
Root cause	User ID format mismatch

Root cause breakdown:

User ID Format	Broken?
Short hex IDs (futurhealth_legacy)	100% broken
Long Firebase UIDs	0% broken

All broken users = MICRO tirzepatide path users. This is a systemic ID integration issue.

Canvas health check (March 15):

Status	Count
Working canvases	2
Canvases with 0 entries for 7+ days	11

The 2 working flows:

1. New Wegovy Pill Flow — 475 entries/day
2. Old SMD Welcome Eligible — 72 entries/day

Everything else: 0 entries. Broken.

Impact:

Problem	Revenue Impact
25% purchaser gap	Unknown — no onboarding for these users
11 dead canvases (prospect + onboarding + post-purchase)	Recoverable: $460–770K/month
1,800/day cart abandoners with no recovery	$50–100K/month

Proposed fix (3-layer approach):

1. Catch-all canvas (immediate) — broad canvas that catches all purchasers regardless of ID format
2. PostHog→Braze webhook (robust) — event-driven sync that handles both ID types
3. Segment rescue (backlog) — retroactively add broken users

This was posted to sam-chat with Christian and Alexandra tagged.

Automation Scripts

Sam has written 4 Braze audit scripts in the workspace:

Script	Purpose
braze-audit.js	Initial canvas and flow health check
braze-audit-full.js	Comprehensive audit with subscriber analysis
braze-check-fields.js	Check specific subscriber field values
braze-check-fields2.js	Updated version with additional checks

These are standalone Node.js scripts that run against the Braze REST API.

Cart Abandon Flow Details

Wegovy-Pill has a dedicated cart abandon flow:

• Entry rate: 871 entries over 2 days
• Conversion rate: 7.7% (67 conversions)

Other paths (SEMA, MICRO, NUTRITION) — no dedicated abandon flows. May be on legacy system.

Recommended improvements identified by Sam:

• Faster SMS (trigger at 30 min, not after delay)
• Include 58.5% better results stat in emails
• Objection handling copy
• Social proof additions
• Team is A/B testing these recommendations

Routing: Who Gets Which Flow

From March 18 audit task (Braze lead audit for 1,200 prospects):

User medication	Braze Canvas
wego-pen	New Wegovy Pen Flow
wego-pill	New Wegovy Pill Flow
All others	Copy of Feb2026 Cart Abandon

Gap found: Expected 1,200 entries from March 17 → Actual 1,077 (Pill=113, Pen=43, Cart Abandon=921). ~10% gap. Requires user-level matching to diagnose.

Why This Belongs in a Dedicated Agent

Problem with current setup:

• Braze audits are complex, multi-step operations requiring many API calls
• Currently mixed into Sam's context with funnel data, creative work, and Slack monitoring
• Scripts exist but aren't packaged as reusable tools

Dedicated Prospect Agent benefits:

• Deep Braze knowledge without contamination from analytics/creative context
• Can maintain subscriber health state across sessions
• Can proactively monitor canvas health daily (not just when audited)
• Can manage the PostHog→Braze webhook integration once built

---

See also: Prospect Agent | Braze Integration | Workspace Scripts