Monitor Braze transactional emails
API-triggered Canvas journeys fire on external webhooks, send within seconds, and depend on real-time entry conditions. When a password-reset webhook fails or an entry property mismatches, the Canvas logs the event as received and the email never fires.
Telltide confirms the email reaches the inbox, under two minutes from trigger to alert.
What breaks quietly
Why transactional Canvases fail without alerting
Canvas analytics show API call volume and send rates. They do not alert when a single webhook never reaches the entry step, or when an entry condition rejects a payload that looks valid upstream.
Webhook signature validation fails after key rotation
An API key is rotated for security. The Canvas still references the old signature. Incoming webhooks are rejected at the entry step. Braze logs the rejection. No alert fires. The customer clicks password reset and waits.
Entry property schema changes upstream
A Canvas enters users when a webhook includes a reset_token property. The upstream service renames it to token_id. The Canvas entry criteria no longer match. Zero users enter. Canvas logs zero entries as normal state.
Rate limits exhaust during traffic spikes
A promotional email includes a password-reset link. Thousands of users click at once. The Canvas hits its API rate limit. Later webhooks are queued or dropped. Braze logs the throttle event. No user-facing alert fires.
Liquid template references a connected data source that lags
A Canvas pulls user details from a connected database. The webhook fires before the database write completes. The template renders with blank values. Braze logs the send as delivered. The customer receives an email with no reset link.
Exception event ejects users before the email step
A Canvas includes an exception event that exits users who complete account verification. The event criteria are broader than intended. Users who request a password reset after verifying exit immediately. No email fires.
Timezone miscalculation delays send to the next day
A Canvas includes Intelligent Timing on a step that should fire immediately. The timezone calculation resolves to a future send time. The email queues for 12 hours. Canvas logs the step as scheduled. The customer assumes the reset failed.
Real breakage pattern
When a late arrival is not the same as a broken send
A broadcaster's Saturday newsletter once failed to send for the first hour of the window. The platform reported success. The inbox stayed empty. When an email arrives in the first 30 minutes but later than expected, the cause is usually a scheduling or timezone issue. A genuine breakage shows up as no arrival at all.
Late arrival is usually scheduling, not breakage
When a transactional email arrives 10 minutes after the trigger, the likely cause is queue congestion or a misconfigured delay node. The email still arrived. A broken Canvas shows up as no arrival within the entire expected window, typically two minutes for password resets.
Heartbeat mode catches absence before customers notice
The symptom is not an error event. It is the absence of the expected send. Scheduled monitoring fires a test webhook on a cadence. If the Canvas does not send within the arrival window, an alert fires. You know the journey is idle before the first support ticket.
Canvas analytics will not flag rejected webhooks as broken
When a webhook payload fails entry validation and Canvas entry drops to zero, Canvas treats that as valid state. There is no threshold for alerting on abnormally low entry counts. Inbox-side monitoring fills the gap by confirming a test profile enters and receives the email.
How Telltide fits
A monitored profile for every transactional path
Telltide runs alongside Canvas, not inside it. You add a test user to the entry audience. Telltide watches the inbox for the sends Canvas says it made.
Add the monitor address to your Canvas entry segment
Telltide gives you a unique inbox address per monitor. You create a user profile with that address, assign the entry properties the Canvas expects, and let a scheduled test event fire the same webhook a real password-reset flow would call.
Set the arrival window to match expected latency
For a transactional Canvas with no delay, the window is typically two minutes. For a Canvas that includes a short delay before sending, add that duration to the window. For Intelligent Timing, set a wider buffer to account for send-time variation.
Get alerted when the inbox disagrees with Canvas
If the email does not arrive in the window, an alert fires. If it arrives twice, an alert fires. If the content deviates from the reference template, an alert fires. Canvas might still report the step as healthy. The alert tells you what actually reached the inbox.
Monitoring specific transactional components
Webhook triggers, connected data sources and send-time logic
Each Canvas component has its own monitoring considerations. Here is how to set up Telltide for the components that break most often in API-triggered flows.
Fire test webhooks from a scheduled harness
For a Canvas that enters on an external webhook, trigger the same API endpoint from a scheduled script. The monitored profile receives the webhook, enters the Canvas, and Telltide confirms the first step fires. If the webhook fails or the entry criteria reject it, the monitor alerts within the configured window.
Watch for race conditions with connected data sources
If a Canvas pulls data from a connected source, set the monitor to check both arrival time and content. A race condition where the Canvas sends before the data write completes will render with blank or stale values. Telltide compares the arrived email against the reference and alerts on structural deviation.
Monitor each send-time variant separately
If a Canvas includes Intelligent Timing or a delay based on user timezone, create monitors for each expected send-time bucket. Each monitor gets a user profile with the timezone or preference that qualifies it for one path. If one bucket stops sending, you know which logic broke.
Set arrival window tighter than retry logic
Most password-reset flows include a retry button visible after 60 seconds. Set the monitor arrival window to 90 seconds. If the email does not arrive by then, the alert fires before the customer clicks retry and assumes the system is broken.
Canvas observability vs native analytics
What Canvas Analytics shows, and what it cannot
Canvas Analytics are detailed. They show every webhook, every entry, every conversion. What they cannot show is whether the email that Canvas logged as delivered actually arrived in the shape you intended.
Canvas reports delivery, not inbox arrival
When Canvas logs a send as delivered, it means the receiving mail server accepted the message. It does not confirm inbox placement, spam filtering, or correct rendering. Inbox-side monitoring closes that gap.
Webhook rejections are logged, not alerted
When a webhook payload fails entry validation, Canvas logs the rejection reason. It does not alert you that the rejection happened. If the rejection was caused by a schema change upstream, you will not know until you actively review the entry logs.
Liquid errors render silently
When a Liquid template references a missing attribute or a connected data source that times out, Canvas renders the block as blank. The email is logged as delivered. The customer receives broken content. Telltide compares the arrived email against a reference and alerts on structural deviation.
Pair it with
Concepts and related monitoring guides
The pages below cover the broader transactional monitoring context and how it fits with other journey types.
- Monitor Braze: the parent guide covering all Canvas send surfaces.
- Canvas monitoring: step-by-step confirmation for multi-step journeys.
- Triggered flow: how entry events and conditional logic interact across platforms.
- Webhook trigger: API-based entry mechanisms and their failure modes.
FAQ
Common questions about monitoring transactional Canvases
What causes API-triggered Canvases to stop firing?
Webhook signature validation failures, API key rotation without Canvas update, entry property mismatches after schema changes, or rate-limit exhaustion during traffic spikes. Braze logs the API call. The Canvas never enters the user. No email fires.
How do I monitor a Canvas that fires on external webhook triggers?
Set the arrival window to match the expected webhook-to-send latency, typically under two minutes. If the email does not arrive in that window, the webhook either failed to reach Braze, or the Canvas entry criteria rejected it. The alert fires before the customer retries the password reset.
Can Telltide catch race conditions where the email sends before the database write completes?
Yes. A monitored profile enters the Canvas at the same time as a real user. If the Canvas pulls data from a connected source that lags behind the trigger event, the email might render with stale or blank values. Telltide compares the arrived email against the reference and alerts on content deviation.
Do I need to trigger test API calls to monitor transactional Canvases?
Yes. Scheduled test events fire the same API endpoint a real password-reset flow would call. The monitored profile enters the Canvas. If the endpoint fails or the Canvas rejects the entry, the monitor catches the missing email within the configured window.
Start watching your transactional Canvases
One monitor free. Paid plans from $49 USD per month. Set up takes about two minutes.