Reducing fraud in public sector contact centres in recession times

When income levels fall, as they tend to do in recession times, fraud levels invariably go up.

Public sector call centres are a target for fraudsters, as is every organisation that handles money. According to National Crime Agency figures, fraud costs the UK public sector £40 billion a year.

Fraud executed via contact centres has more than tripled in the last five years . Statistics like these have led Gartner to warn organisations not to “let the contact centre be your fraud Achilles’ heel.

The COVID-19 pandemic has also increased fraud risk due to abrupt changes in working practices and a bigger strain on resources . Financial consultancy EY recommends organisations consider new anti-fraud technologies as soon as possible.

None of this is too surprising, but what can public sector contact centres do to reverse the tide?

New platforms, new TCOF savings

Modern speech technologies are providing the foundation for new approaches to fraud prevention.

They’ve evolved beyond classic speech-to-text, providing the ability to identify speakers, analyse the credibility of a speaker, and even recognise conversation topics.

Importantly for contact centre managers who target efficiency, new solutions provide a technological platform for anti-fraud. These are automatic systems that scale with demand – not new staff responsibilities that require costly re-training.

In fact, they may empower call agents to act on fraud with greater insight.

And they can help with reducing the all-important total cost of fraud (TCOF) , a major part of which is fraud losses, anti-fraud tools, and working time spent on fraud.

But how?

Beyond voice biometrics

Using speech technology to combat fraud isn’t a new idea. The public and private sectors have been using voice biometrics, which recognises the voices of known fraudsters, for some time.

But voice biometrics has its limits. It requires enrolment, yet not everyone enrols. It can only detect known fraudsters, not first offenders. It can be tricked by disguised voices.

A more capable, more dynamic new technology is conversational analytics.

Conversational analytics identifies characteristics and patterns of speech that are consistent with deceitful behaviour. These markers include things like hesitations, answering a question with a question, and slow speech.

That means it can…

• Alert call agents when callers are exhibiting dishonest or fraudulent behaviour – eventually it will do this in call, in near real-time
• Spot previously unknown fraudsters, because it doesn’t rely on identifying a particular known voice

It’s accurate too. The product developed by Intelligent Voice, LexiQal, analyses speech at utterance level with 90% accuracy. Ongoing research funded by the UK Research and Innovation (UKRI) is making LexiQal even more efficient, automated, and transparent.

And conversational analytics isn’t just for anti-fraud. It can help you measure customer satisfaction and identify your best-performing call agents too.

Faster anti-fraud action

Under GDPR, anti-fraud decisions and actions need to be defensible and explainable, so the people impacted by them can challenge unfair decisions.

So it’s important that technologies like LexiQal only provide anti-fraud recommendations to call agents. Call centres cannot automate these decisions or allow AI to judge that a person is trying to commit fraud.

But it can help call agents take faster, better-informed action on fraud.

And even though the machine is not making the decision, it is important that these systems have built-in explainability, so the reason behind the recommendation can be easily surfaced and defended.

Another technology that’s useful in this area is smart audio transcription. Smart transcription can transcribe a huge volume of calls at once, identify calls that need review by spotting defined topics, and even provide hyperlinks to interesting bits of the recording.

That can save anti-fraud efforts a huge amount of time once spent reviewing calls. Running on GPU-accelerated HPE servers, the Intelligent Voice solution, SmartTranscript, can process in excess of 1,200 hours of audio per hour.

Learn more about this topic

Economic conditions may increase fraud, but new technologies like conversational analytics and smart transcription offer public sector call centres a new platform on which to build stronger prevention measures.

With input from Intelligent Voice, HPE recently published a white paper that discusses a wide range of speech technologies and their potential uses within public sector call centres.

As well as fraud, it addresses challenges such as improving service quality, managing demand, duty of care, and regulatory compliance.

Click here to download the white paper.

To find out more about Intelligent Voice please contact us at info@intelligent-voice.test or give us a call on 0203 6272670