Customer Overview
A major European omnichannel retailer with a high volume of online orders across multiple product categories, with a centralised customer operations function responsible for post-purchase enquiries for returns, refunds, delivery issues, and order changes.
Challenge
Customer enquiries were handled primarily through a shared email inbox, which served as the main intake channel for online orders. Requests covered returns, refunds, delivery problems, missing items, and order changes. Under normal conditions, the team maintained responses within hours, but inbound volume spiked sharply during flash sales, seasonal campaigns, and Black Friday.
During these peaks, the inbox quickly became overloaded. Response times stretched from hours into multiple days, which was reflected directly in CSAT scores. Customers followed up repeatedly on refunds and deliveries, creating duplicate contacts and additional pressure, even when work was already in progress internally.
The retailer explored automation as a way to absorb volume. Early efforts relied on out-of-the-box AI agents focused on email categorisation. Precision was inconsistent, below acceptable thresholds, and categorisation alone did not reduce operational effort. Each request still required manual checks across orders, payments, and delivery systems, followed by human follow-up and confirmation.
Noxus Solution
The retailer deployed a Noxus AI Worker as the first line of handling for inbound customer enquiries arriving through the shared inbox. Each AI Worker takes ownership of a case from intake through resolution, following defined standard operating procedures rather than producing standalone replies. For each inbound email, the AI Worker executes a consistent set of steps:
- Reads and categorises the enquiry based on intent and content
- Extracts key details such as order references, customer identifiers, and issue type
- Pulls context from internal sources, including approved FAQs, order and fulfilment systems, payment providers, and customer history
- Applies the relevant SOP to determine eligibility, next actions, and required updates
- Completes in-policy actions in systems and sends a clear confirmation to the customer
A large share of enquiries initially arrived without the information required to proceed, such as missing order numbers, invoices, or customer verification. Based on the enquiry category, the AI Worker identified what was missing and responded to customers with targeted requests for the required information. Once the information was provided, cases progressed without restarting triage or manual intervention.
Non-routine, sensitive, or high-risk enquiries were explicitly identified and routed to human teams. These included high-value orders, complaints, and cases requiring judgement or approval. Each escalation included the full context already gathered, allowing teams to focus on decision-making rather than investigation.
Customer Platform Manager
Following the implementation of Noxus, the retailer formalised a Customer Platform Manager role within Customer Operations. This role owned customer platforms end to end, using Noxus as the core execution layer to translate operational knowledge into scalable automation across multiple operations.
Rather than focusing on agent productivity or templated responses, the Customer Platform Manager was responsible for configuring workflows, routing logic, and integrations, defining standard operating procedures, and governing how AI Workers were deployed. Working closely with Product and Engineering, the role ensured automation aligned with business rules, system constraints, and customer experience requirements.
With clear ownership in place, the team is now positioned to use Noxus as a platform for continued automation, expanding AI driven workflows to cover more customer operations over time. The Customer Platform Manager can progressively reduce manual handling by refining which enquiries are handled autonomously, adjusting human review and approval thresholds, and evolving how data and decisions flow across systems. As this capability grows, the retailer is set up to further lower operational overhead, maintain consistent handling at scale, and gradually reduce reliance on outsourced support without compromising control or service quality.
Results
- 87% of inbound customer enquiries resolved end to end through automation
- 43% lower cost per resolution
- +3.0 percentage point improvement in CSAT
- 24/7 availability for customer enquiries
