Industry spotlight: Lee Mostari, Director of Insight and Analytics reveals why speech analytics is critical to better support agents from customer abuse

3rd October 2022

Call agents have reported that not only have they experienced verbal abuse, racism, and discrimination but some have also received threats to life. How does speech analytics support agents handling challenging calls like these?

Abusive conversations can include offensive language, tone or both. Speech analytics helps to detect abusive language or terminology as well as identifying adverse patterns of speech, such as over talk between the customer and the agent, which is also a sign of abusive behaviour. Analytics solutions can also identify highly emotional customers using abusive language. Abusive conversations are likely to occur more often due to the cost of living crisis increasing the amount of financially vulnerable customers. By identifying these types of calls, organisations can then be poised to provide the best post-call support for their agents.

Please can you give us a brief overview on how speech analytics actually works?

Natural language processing and Artificial Intelligence (AI) helps to transcribe call recordings into a text-based conversation. This data is then compared to pre-created language queries to identify keywords and phrases to help organisations better understand components of the call and identify positive and negative customer or agent outcomes. When a language query is detected during a call with an agent, this will create a marker on the call which can be alerted for investigation. With significant technology advances in transcription accuracy too, the resulting queries are highly accurate.

Analytics technology has advanced significantly in recent years. Most solutions are now cloud hosted and can be deployed within hours or days. Previous deployment of such technology would typically take weeks or even months. This has resulted in organisations now being able to introduce analytics into operating models at speed.

At Davies we have over 50 analytics deployments under our belt, so we have many of the query language packs already written, so we can stand up a solution and deliver insights into customer and agent behaviours within days.


From your experience, share with us a project that you’ve worked on where an organisation has deployed speech analytics to support their agents and the impact that this has had?

We recently deployed speech analytics for the UK Police Force that was able to help them identify highly stressful and emotionally distressing calls coming through via 999 and flag these calls to supervisors. This provided supervisors with an opportunity to check-in with their call centre staff to enquire about their wellbeing and mental health shortly after these stressful interactions.


As well as deploying speech analytics, what other CX tech or processes should organisations implement to adhere to the new Service with Respect law (The Police, Crime, Sentencing and Courts Act 2022) and better support the wellbeing of their agents?

We have seen many organisations tackle this issue in many different ways. Some build in scheduled and unscheduled breaks for agents to ensure they manage burn-out during the working day – whereby an agent can put themself into an off-line status if they have had a particular difficult or emotional contact to handle. Others have the ability to flag a call as abusive and a supervisor can join the call to help the agent handle an abusive customer. We have also seen a marked increase in contact centres using online learning and specific coaching models on how to de-escalate abusive or threatening contacts to address this growing issue. The most important thing apart from providing the best CX under challenging circumstances is that no matter what, organisations must take care of the wellbeing of their agents as a priority.

If you would like to continue this conversation, get in contact with Director of Insight and Analytics Lee Mostari on

    Keep up to date with Davies