BlueShop Support Recovery — As Built

What this showcase explains: a delivered-order problem becomes one owned case with trusted context, customer-visible conversation, private agent coordination, explicit lifecycle and measurable resolution.

1. What Support Delivers

The customer receives one visible recovery thread while the support team gets the protected context and private workspace needed to resolve it.

Customer continuityThe case begins from a real delivered Order.

Ownership and delivered state are checked before the dispute, order context and initial message become one ticket.

Operational accountabilityOne agent, one lifecycle, one measured outcome.

Assignment names the owner; public replies and private notes remain separate; SLA and events make response and resolution observable.

Practical example: a customer reports a damaged delivered parcel. An agent claims the case, checks the exact order and delivery address, replies publicly, records carrier evidence privately, resolves the issue and closes a read-only record.

2. Delivered-Order Dispute Resolution

The complete film connects customer and agent surfaces with burned-in narration.

CUSTOMER + SUPPORT AGENT / 1M 44.24S

From Delivered Problem to Closed Case

A delivered order becomes a damaged-parcel dispute. The agent claims it, checks customer and order context, exchanges public messages, keeps a carrier note private, then resolves and closes the case.

01
Ground the case in deliveryThe customer reports the problem from the delivered Order, carrying order context and the first message.
02
Make work accountableAn agent claims the unassigned case and inspects protected customer and delivery context.
03
Separate audiencesThe customer sees public replies; carrier evidence remains an internal Support note.
04
Close with a visible outcomeResolution and closure leave the customer with a complete read-only record.

3. One Case, Clear Stages

Delivered Order problem moves through verified case creation, agent claim, public conversation, private coordination, resolution and closure
The customer sees one recovery journey. Support keeps accountability, visibility and lifecycle decisions explicit behind it.

4. Business Result

TECHNICAL DEEP DIVE

How Support recovery works

Support owns the case and its conversation. Orders and Customer remain authoritative for their context; Analytics receives facts rather than commands.

5. Ownership and Access

Support references Order and customer identity without absorbing either domain.

Orders owns order and delivery state, Customer owns account context, Support owns ticket messages SLA and lifecycle, Analytics owns derived read models
Customers read only their own ticket and public thread. Support-agent roles can inspect protected context and internal notes. Analytics cannot mutate the case.
InformationCustomerSupport agentOwner
Own ticket and public threadRead and reply while writableRead and replySupport
Internal noteNever exposedRead and writeSupport
Order status, items and addressOwn Order viewRole-restricted support viewOrders
Customer account contextOwn accountRole-restricted support viewCustomer

6. Trusted Dispute Entry

Support creates a delivered-order dispute only after Orders confirms both authenticated ownership and DELIVERED state.

Customer submits an order problem, Support asks Orders to verify ownership and delivered state, then persists ticket initial message SLA and opened event
The accepted write keeps ticket, initial message, SLA and opened event together in the Support transaction. Orders remains authoritative for the prerequisite evidence.

7. Assignment, Public Replies and Private Notes

New work enters an unassigned queue. Claim records the agent and opens the case; every message is created with a fixed audience.

Unassigned ticket is claimed by an agent, public reply reaches the customer and internal note remains available only inside Support
The first public agent response records first-response time. Internal notes remain absent from the customer thread and protected by row-level security.

8. Ticket Lifecycle and SLA

State controls what action is legal next; persisted SLA policy determines when first response and resolution are due.

Ticket progresses from NEW to OPEN, may wait on customer or internal work or escalate, then resolves and closes with a bounded reopen path
Resolution is legal from OPEN or ESCALATED. Closure follows RESOLVED or customer-wait timeout. Closed cases may reopen within seven days.
Business time: due dates use persisted Europe/Zurich business hours and skip weekends. The current code excludes 1 August explicitly; it does not implement a broader configurable Swiss holiday calendar.

9. Support Events and Operational Analytics

Support saves lifecycle changes with an outbox. Analytics consumes the resulting CloudEvents into independent ClickHouse read models.

Support operation updates Ticket and outbox, Kafka carries support ticket CloudEvents, Analytics creates ClickHouse resolution flow backlog and workload views
Opened, assigned, status-changed, first-responded, resolved, closed and reopened facts rebuild the operational story without querying the transactional Support database.
Read modelQuestion answered
support_resolution_dailyHow quickly did Support first respond and resolve?
ticket_flow_dailyHow did cases move through the operation?
ticket_backlogWhat unresolved work remains by status, priority and category?
agent_workloadHow much active and resolved work belongs to each agent?

Open the live Support Operations & SLA dashboard.

10. Explicit Limits

Assignment is manualNo skill-, language- or load-balanced automatic assignment is delivered.
Recovery is conversationalThe film does not demonstrate automated refund, replacement or carrier settlement.
SLA escalation is incompleteTargets and due dates exist; automatic breach escalation and a full holiday calendar do not.
One case is demonstratedThe film does not prove concurrent assignment arbitration, merge or bulk operations.
Privacy controls existA dedicated identity-verification workflow and retention/anonymization automation remain outside this journey.
Analytics is derivedDashboard read models measure Support operations but never become the ticket source of truth.