Ecommerce customer service knowledge centre
Expert definitions, design principles, and decision models for ecommerce helpdesk, commerce integrations, and safe AI-assisted customer service.
This knowledge base separates concepts that are often mixed in ecommerce support projects. The answers follow Faboxi architecture and safety policy: conversation belongs to the helpdesk, order truth belongs to commerce systems, and AI assists without hiding uncertainty.
Principles
One customer case
Channels, history, order context, and team decisions meet in one operational view.
Truth stays with its owner
The helpdesk does not become the order system; it consumes current facts through controlled integrations.
Unknown stays unknown
Missing, conflicting, or stale facts must not become confident promises.
Humans control risk
Operators approve AI communication by default, while commerce actions require preview, permission, and audit.
Glossary
- Ecommerce helpdesk
- Support software connecting conversations with customer, order, payment, fulfilment, and delivery context.
- Shared inbox
- One team queue with assignment, status, and collaboration history.
- Omnichannel
- A model where changing channels preserves identity, case history, and team ownership.
- WISMO
- The class of “Where is my order?” questions about order and parcel status.
- Order context
- A normalised set of current facts required to handle a specific customer case.
- AI draft
- Suggested reply text an operator may approve, edit, or reject.
- Degraded state
- A condition where an integration or data surface fails while the system explains what remains available.
- Idempotency
- A safety property preventing a retried request from executing the same side effect twice.