Delivery Stack
Messaging Stack Reliability Guide
Updated September 15, 2025
Mission
Guarantee that every Cuculi nudge fires on time, selects the best channel, and delivers measurable feedback into the Insight Control Tower.
Core Components
- • MongoDB Triggers: Change streams publish user/event deltas to AWS.
- • AWS Lambda: Evaluates trigger recipes, enriches context, and routes payloads.
- • Twilio SMS: Short-form urgency nudges with delivery + click tracking.
- • OneSignal Push: Native push for real-time reminders.
- • Mailchimp Email: Rich storytelling, post-event reciprocity loops.
Reliability Targets
- • 99.9% successful Lambda executions per day.
- • < 2 minutes median latency from trigger to send confirmation.
- • Automated retries with exponential backoff and dead-letter queues.
- • Channel-specific deliverability > 95% with anomaly alerts.
Observability Checklist
- Instrument CloudWatch metrics for Lambda duration, errors, and cold starts.
- Stream Twilio/OneSignal/Mailchimp webhooks back into MongoDB for dashboard joins.
- Expose Grafana boards for channel latency by hour, aligning with the Control Tower.
- Alert on delivery dips >5% vs. trailing 7-day average; route to Ops Slack.
Runbook Highlights
Twilio SMS
Rotate messaging services IDs quarterly, monitor carrier filtering, and feed click data into the Evidence Pack.
OneSignal Push
Cap silent pushes per user, evaluate open rate vs. send time, and mirror results in the Trigger Matrix.
Mailchimp Email
Sync audience lists nightly, personalize with event context, and track reciprocity follow-ups.
Lambda Orchestration
Use provisioned concurrency for peak windows, version functions per recipe, and log payload decisions for audits.
Operator Moves
- • Review delivery dashboards before shipping new recipes from the Trigger Matrix.
- • Align with lifecycle on channel selection using the Psychology Evidence Pack.
- • Log incident postmortems inside the Four-Pillar Strategy workspace for traceability.