Risk score (attrition)

Definition

An attrition risk score is an indicator estimating, for each client, how likely they are to cancel, not renew or let a policy lapse in the coming months. It is built from observable signals: failed debit orders or unpaid premiums, no reply to messages, a recent complaint, a request for a competing quote, falling activity, negative language in a conversation, a renewal date approaching with no contact. Its value is not perfect prediction but ranking: a brokerage that cannot call every client calls the ones the score puts at the top first. A usable score is also explainable, so that the adviser can see why a client is rated as at risk and override the rating when reality differs from the model.

What you need to know

For a brokerage, the risk score is the operational side of retention. In South Africa and Namibia, where a lapsed life or funeral policy triggers commission clawback, the score identifies who to contact on the day a debit order fails. In the UK, where personal lines renewals are contested every year and Consumer Duty expects firms to act on signs of disengagement, it points to the clients to review before a comparison site does. In both markets it complements persistency and renewal rate, which measure the outcome after the fact.

In ORIS, each customer has a risk_score from 0 to 100, shown in the Risk column of the Customers & Segments table and displayed as five levels: Very Low (up to 20), Low (up to 40), Medium (up to 60), High (up to 80) and Critical above that. The score is updated by the analysis of incoming WhatsApp messages: when the classifier detects a high churn risk (HIGH_CHURN_RISK), the score rises, an opportunity of type attrition_risk is created or updated, and a "High Churn Risk Detected" notification alerts the team. The Dashboard shows the At-Risk KPI and the Highest Risk table; the Opportunities & Risks view sorts by Highest Risk and opens the conversation with a Best Next Action suggestion and the client's mood. Risk Score is also a filter for building segments. The score does not replace the adviser's judgement, and ORIS neither collects premiums nor detects failed collections on its own: that information comes from conversations and imported data.

See the lapse-risk guide and the retention and loyalty hub.

Concrete example

A client replies to a premium reminder: "Might cancel, it's getting too expensive." In ORIS the message is classified as high churn risk, her score moves to High, an attrition_risk opportunity appears in Opportunities & Risks with the matching notification, and the adviser opens the conversation that morning to offer a cover review before the next collection date.