MAIF Using Topic Modelling to reduce contact centre bottlenecks.

MAIF is one of France’s largest home and automotive insurance companies, with more than 3 million members.

One of the challenges facing its customer services team was managing the volume of calls coming into its call centre — on average, some 8 million a year.

With no way of vetting calls before they reached an operator, the team was wasting precious time responding to questions customers easily find the answers to on the MAIF website.

To improve efficiencies, we needed to filter out unnecessary calls and free up more time for MAIF’s customer service teams to process more complicated requests.

To understand why customers were calling MAIF’s call centre, we developed Natural Language Processing (NLP) algorithms to analyse transcripts of more than 4 million calls.

We then used topic modelling to categorise every call into one of 35 different request types.

We liaised with MAIF’s business teams to identify which questions could be solved online and which needed a human response or presented a sales opportunity.

Where calls did not represent an opportunity, we advised how to answer these questions online.

Our analysis showed that 32% of inbound calls were ‘low added value requests’ — questions that could easily be answered online.

As a result, we built a roadmap advising MAIF how to solve these questions online and direct people to this content to avoid calling.

Digitising these queries has let MAIF’s customer services team prioritise cases that require a human touch, improving efficiencies and its round-the-clock service.