Monitoring telephony channels: Are insurers missing the signals?

Computer giant IBM runs an experiment for clients. It calls their help desks to see if it can impersonate an employee to reset their password. To date, the deception has been successful every single time.

It is a striking example of the simple power of telephony channels in potential fraud. A threat to which UK insurers with call centres are becoming ever more alive.

“Voice interactions are increasingly important [in detecting fraud] but can still be less visible than digital signals,” said David Phillips, claims validation technical manager at NFU Mutual.

Digital signals of fraud can be immediately obvious to security systems used to detect malicious activity. Device anomalies such as new or hidden IP addresses. Behavioural deviations such as erratic typing speeds or unusual mouse movements. And identity indicators such as recently created email addresses.

But voice-based signifiers can be harder to spot. They often fall into psychological red flags, audio inconsistencies, and unusual requests, which can be trickier for insurers’ call centre staff to determine.

Similarly, where numbers are withheld or spoofed, real-time denylist checks can be difficult, making it hard to judge if a caller can be trusted.

The biggest gap between cross-channel fraud strategy and operational reality is what frontline colleagues can see live.

David Pritchard, NFU Mutual

Fraudsters try to exploit these weaknesses. NFU Mutual – with its local agency network and UK-based call centres that remain the primary point of contact – “regularly” encounters unusual or suspicious behaviour on calls, it said.

Individual caller interactions are not always decisive when it comes to working out a potential fraud, said David Pritchard, NFU Mutual’s claims fraud manager. So escalation routes and experienced judgement remain essential to identifying genuine concern.

In real world, real-time situations, however, this can be hard to pinpoint. “The biggest gap between cross-channel fraud strategy and operational reality is what frontline colleagues can see live,” said Pritchard.

That includes cross‑checking incoming numbers against internal denylists and external feeds in real time, plus spoofing checks, so high‑risk calls can be flagged and routed.

“Fragmented data and limited external visibility can leave genuine risk signals hard to evidence quickly,” added Phillips. Alternatively, if the insurer’s risk appetite is too narrow, it can lead to false positives, with customer journeys being stopped unnecessarily.

Linked activity can be identified, but it is not always straightforward when telephony, policy, claims and fraud data are separate, or identities are disguised across channels, he said.

Joining the dots

Teams would benefit from more joined-up intelligence across prior contacts, linked entities, behavioural indicators and external sector data, helping identify wider fraud patterns earlier and with greater confidence, said Phillips.

“Cross-sector data sharing, including from telecommunications and online social media sectors, would add significant value,” added Pritchard.

Cifas, a fraud prevention service in the UK, reports a notable rise in fraud targeting sectors adjacent to insurance, such as telecommunications and e-commerce. Fraudsters use these platforms to build synthetic identities or test stolen credentials before targeting insurers.

Matthew Crabtree, head of financial crime intelligence and investigations at Allianz, agreed cross-sector data sharing is a missing piece in insurers’ fight against fraud. “We need to get better at working with other financial services sectors or telecom sectors,” he said.

“Fraudulent voice calls are coming through the telecoms industry. So the question is, how can we share what we’re seeing and disrupt that through the telecoms industry? It’s something I’m actively working on.”

A four-pronged approach

Disrupting fraud internally at Allianz takes a four-part strategy, said Crabtree, of which caller interactions are an essential part of the puzzle.

“Without question, we’re seeing an increase in vishing activity, with voice calls coming into us or our customers, as the fraudsters try to seek additional data for their fraudulent attempts,” he said.

For Crabtree, any good fraud strategy needs four pieces of a jigsaw: “Because with the evolution of fraud, you’re not going to get a single solution that is going to be able to tackle it.”

People are an important part. Customers and staff are trained and educated to spot what doesn’t look right, including via calls coming through to the call centre or calls purporting to be the insurer. A number of Allianz staff now have recognised national vocational qualifications (NVQs) in fraud detection.

Technology is the second piece of the puzzle, used to monitor incoming calls and numbers, and look for unusual activity or fraud triggers. Allianz uses US tech to look at voice patterns and issues that could be red-flagged as fraud.

“That helps us identify what issues in voice patterns would then trigger a referral into our claims validation teams,” said Crabtree.

At that point the third part of the jigsaw is activated, skills. “That’s where we get our skilled fraud investigators to then look at that, listen to the calls, and start to investigate the claims properly. We’ll use things like machine learning models, so that’ll be looking for things in data that just doesn’t look right, and having the ability to overlay that,” he said.

Finally, there is intelligence, from internal or external sources. “That’s us asking what intelligence we already have on some of these fraudsters,” said Crabtree.

Part of his role is to look after a financial crime intelligence team. “That intel team will be sitting there sharing the intelligence and making sure that’s loaded into our operating systems,” he said.

If a fraudster comes back – that might be triggered by the same telephone number, an email address, it might be a bank account – “if we’ve got intelligence on that, that again will flag in,” Crabtree said.

Without question, we’re seeing an increase in vishing activity, with voice calls coming into us or our customers, as the fraudsters seek additional data for their fraudulent attempts.

Matthew Crabtree, Allianz

Keeping pace

But even with his comprehensive fraud strategy in place, Crabtree said it’s hard to keep pace with the ever-evolving scammers.

Cifas has revealed that more than 444,000 cases were recorded to the National Fraud Database (NFD) in 2025 – the highest number ever recorded in a single year, and a 6% increase on 2024.

“So you’ve got to keep evolving too, you’ve got to keep pushing yourself and know the gaps because that’s when they’ll start creeping in if you don’t have an agile plan to detect and disrupt,” Crabtree said.

“You’ve always got to learn. I don’t think you could ever sit there and go, yes, we’ve got everything right.”

 

Sponsored content