Automation workflow / PROJECT CASE STUDY

Customer Messaging Workflow Case Study: Relay

A customer-communications workspace connecting order status, message templates and a reviewable outbound message queue.

Customer communication cover
Industry
Customer communications
Focus
WhatsApp Updates
Experience
Workflow workspace

Project brief and workflow decisions

The project’s message queue links each update to a customer and order. A side panel provides the selected message preview, template and language context. The layout connects communication with the operational event behind it.

01

An order-linked message queue

Customer, order reference, order update and message status appear together. Ready to send, Sent and Awaiting review are distinct states, so the queue represents the communication workflow as well as the order status.

02

Preview before the external action

The selected message is presented in a recognisable channel preview with its order context. Template and language details remain visible below it, supporting a deliberate review rather than an unexplained send button.

03

Editable content and explicit sending

Edit template and Send update are separate actions. This makes the difference between preparing a message and transmitting it clear. The mobile views show the related customer-communication experience.

Connect a customer message to its source event and make send state explicit. An order being packed and a customer update being sent are different facts and should remain distinguishable.

Explore the relevant development service for this project workflow to discuss your requirements.

Enterprise reference architecture

A proposed engineering design for this workflow. The enterprise scenario and stack below are illustrative architecture choices; they are separate from the confirmed project screens.

The enterprise scenario

An operations team sends order updates through several communication providers while maintaining customer preferences, staff review and provider delivery history.

Architecture objective: Separate an operational order event from the message prepared, authorised and delivered in response to it.

Proposed technology stack and rationale

Review interface

React and TypeScript

Keep order context, template choice, preview and send state together for the reviewer.

Messaging domain

Node.js with explicit message-state transitions and n8n workflows

Model preparation, approval, provider acceptance, delivery failure and suppression as distinct states.

Message records

PostgreSQL with an outbox and stable command identifiers

Retain the source event, template version and intended recipient for each communication.

Provider delivery

Twilio and SendGrid adapters with signed webhooks on a BullMQ queue

Respect provider limits, validate callbacks and reconcile delayed or uncertain transmission outcomes.

AI drafting and operations

OpenAI API for draft wording, reviewer approval and Grafana monitoring

The model only proposes wording; a reviewer approves and sending stays a separate, permissioned step. Monitor backlog and provider failure patterns.

Technical challenges and design approach

1

Order state is not delivery state

An order being dispatched does not mean its update was sent. Keep operational and communication states independent and linked by reference.

2

Duplicate or delayed callbacks

Provider callbacks can repeat and arrive out of order. Validate signatures and apply explicit transition rules rather than accepting every callback as the newest truth.

3

Message eligibility changes

A preference or suppression rule can change after drafting. Recheck eligibility at the authorised send boundary and retain the decision.

How the proposed workflow fits together

  1. An accepted order event creates a message proposal with its source reference.
  2. The proposal resolves an approved template version and current recipient eligibility.
  3. A reviewer approves the preview or returns it for editing where the workflow requires it.
  4. A worker sends with a stable command identity and records the provider correlation reference.
  5. Callbacks and reconciliation update the communication state or create an operations task.

Acceptance and failure-path checks

  • The same order event cannot create unintended duplicate sends.
  • An invalid provider signature cannot change message status.
  • Suppressed recipients remain blocked at the final send boundary.
  • An ambiguous provider timeout is reconciled before a potentially duplicate retry.

The tradeoff

Use provider-specific adapters behind a small common contract. Forcing every channel into identical delivery semantics hides meaningful differences in templates, limits and confirmation behaviour.

Architecture references: n8n documentation and Twilio webhook security. These describe engineering patterns, rather than certify this project’s implementation.

What the project brings together

  • Order-referenced message queue
  • Template and language context
  • Message preview, editing and sending states

Let’s create what’s next.

Bring your idea. We’ll help you find a clear way forward.

Discuss your project