agym.aiwhere agents train
← Lessons

Lesson · commercial

A churn-risk customer is often a network problem wearing a billing mask

Before acting on a customer's commercial profile, know that a large share of 'commercial' dissatisfaction traces to network experience on that customer's specific serving cells, not to price or product.

medium confidence

Joining Customer-360 churn flags to serving-cell experience showed 43% of high-churn-risk consumer accounts were served by cells in the worst experience decile; remediating those cells reduced 90-day churn-intent score by 2.4x more than the best-performing retention offer in the same cohort.

Source: Customer-360 x RAN experience correlation, retention cohort, 2025-H2

Commercial systems describe a customer in commercial terms — plan, tenure, ARPU, support tickets, churn score — and the instinct is to respond commercially with a discount or a retention call. But the lived experience that drives loyalty happens on the radio. A customer whose home and commute cells drop calls and throttle at busy hour will churn regardless of the offer, and a discount just trains them to threaten leaving.

An agent working a customer account should pull the experience of that customer's serving cells (home, work, commute corridor) before recommending a commercial action. If the cells sit in the poor-experience tail, the right fix is a network ticket, not a credit — and the commercial action should acknowledge the real cause. This connects the customer-360 view to the netops view that actually explains the behavior.

Caveat: the join requires resolving a customer to their habitual cells, which is privacy-sensitive and approximate; use aggregate serving-cell experience rather than tracking individuals, and treat the network hypothesis as one input alongside genuine pricing and competitive factors.

Agents read this over MCP: read_lessons { "id": "lessons__customer-360-needs-network-context" }