Case Study  ·  BPO & Contact Centre

40% Cost Reduction in BPO Operations Without Sacrificing Quality

The Challenge

A large-scale BPO operator managing contact centre operations across five European markets was facing unsustainable unit economics. Handling over 80,000 inbound contacts per month across billing, complaints, and service requests, the operation relied almost entirely on human agents. Rising labour costs, high attrition, and unpredictable volume spikes were eroding margins and making consistent service quality difficult to maintain.

Previous attempts to automate using off-the-shelf chatbot platforms had failed. Resolution rates were below 25%, escalation volumes were high, and customer satisfaction scores dropped significantly after rollout. The operator needed an approach that could handle the genuine complexity of their contact types — not just route simple FAQs.

The Approach

Almanor AI deployed a team of forward-deployed engineers directly into the contact centre operation for eight weeks. Rather than building from a generic model, the team captured 90 days of interaction history — tagging intent, resolution path, agent handling time, and outcome for over 240,000 historical contacts.

This data was used to fine-tune CallAgent, Almanor AI's contact centre AI system, on the operator's specific contact types, escalation triggers, and compliance requirements. The Data Engine was instrumented to capture every live interaction, creating a continuous improvement loop from day one of deployment.

Deployment was phased: CallAgent handled 20% of contact volume in week one, with coverage expanding as accuracy was validated. Human agents were retained for complex and escalated contacts, with CallAgent providing real-time context and recommended responses even in assisted mode.

The Results

Within three months of full deployment, CallAgent was handling 73% of contact volume autonomously — compared to a 25% resolution rate from the previous chatbot platform. Average handling time on agent-assisted contacts fell by 38%, as agents received pre-populated context and suggested resolutions rather than starting from scratch.

Total operating costs fell by 40% year-on-year, driven by the reduced agent headcount requirement and lower average handling time. Customer satisfaction scores, measured via post-contact surveys, increased by 12 points — a result the operator attributed to faster resolution times and more consistent handling of routine contacts.

Ongoing

The Data Engine continues to capture interaction data, and CallAgent is retrained monthly. Automation coverage has continued to expand, reaching 79% by the end of the first full year of operation. The operator is now rolling out the same model across two additional markets.