Key Findings
- Allianz reported identifying more than £93 million in insurance fraud, highlighting the scale of opportunistic and organized fraud across claims
- Allianz chief claims officer Matt Cox described fraud as evolving from small claim inflation to more sophisticated, coordinated activity
- The insurer implemented voice analytics technology from Clearspeed to create a “hostile environment” for fraud while maintaining customer experience
- Clearspeed is embedded directly into the claims process, asking customers yes/no questions and generating a real-time risk signal to guide whether claims should be paid or investigated
- Compared to legacy fraud checks, the approach reduces friction by engaging customers in real time rather than placing them on hold for manual review
- Allianz reported more than 15,800 fraudulent cases identified across personal, commercial, and specialty lines, totaling £92.6 million in the first half of 2025
Why It Matters
- Insurance fraud is shifting toward more sophisticated and scalable activity, requiring new detection approaches beyond traditional rules-based checks
- Embedding risk assessment directly into the claims journey enables earlier identification of fraud at first notice of loss
- Real-time voice-based risk scoring allows insurers to route claims more efficiently, improving both fraud detection and claims handling speed
- Reducing friction in the claims process improves customer experience while still strengthening fraud controls
- Proven deployment by a major global insurer demonstrates how voice-based risk assessment can operate at scale across multiple lines of business
Chief claims officer lifts the lid on how Allianz caught £93m of fraud
By Harry Curtis
Allianz chief claims officer Matt Cox has told attendees at the ITC London conference that the insurer was able to create a “hostile environment” for fraudsters without damaging customer experience through using voice analytics technology.
Speaking at the event in London yesterday (26 January), Cox — who assumed the chief claims officer position in February last year — said Allianz was continually seeing fraud techniques evolve.
“There is always the opportunistic fraud where someone calls up and slightly inflates what they’re claiming for. We will continue to always see that, but it’s the more sophisticated element of it that we have seen growing,” he said.
Among the partners Allianz called upon to get a handle on fraud was voice analytics provider Clearspeed.
“We needed to make our organisation a really hostile place for people to try to commit fraud, either at the organised end of the scale or at the opportunistic end of the scale,” Cox said.
“What we needed was something seamless”
“What we needed was something seamless and embedded within the process. That’s what Clearspeed provided to us.” Matt Cox, Allianz chief claims officer
Cox contrasted the use of the Clearspeed tool with how pre-existing fraud checks figured in customers’ claims experiences.
“Past techniques that have been used have created friction within the claims journey,” he said.
“Many of you, when you would have phoned up to make a claim, might well have been put on hold at various points during that conversation, while the call handler went away and did some checking.”
Waiting around
Explaining how the Clearspeed tool fits into the claims process, Cox said:
“Whether they’ve phoned up or it’s via email or it’s self-serve online, the customer will be directed to phone the Clearspeed tool.
“It will ask them some questions of importance, and it will assess very quickly a risk around potential risk characteristics of the voice of being used.
“That will either inform the handler that this particular case needs to be referred on to the more dedicated team, or it informs that this claim can just be paid.”
While it adds a step to the claims process, Cox said that from the customer’s perspective, using the tool actually reduced friction.
“Rather than being put on hold, sitting there listening to the music, wondering what’s going on, the customer’s being asked questions. They’re engaged, so it doesn’t feel like they’re waiting.”
Staff concerns
Cox also reflected on how Allianz needed to convince call handlers and other claims staff that the tool would work.
“In the past, a lot of those decisions had come down to the individual claims handlers’ experience,” he said. “We were now saying to them, ‘you need to trust this tool — you don’t know how it works, but you need to trust it.’”
He explained how the insurer went about bringing staff up to speed on how it worked, as well as experimenting with it in controlled settings.
“We allowed them to play with it. We allowed them to see it in action, to pilot it in a number of small areas and look at really data-led trials within particular teams to see how it was working.”
He added that Allianz needed to allay staff concerns that the tool would annoy customers, with post-deployment complaints figures putting those fears to rest.
“Over the last year, we’ve had 0.1% of claims that have gone through this technology where our customers said, ‘I didn’t like that being done to me.’ That’s tiny,” he said.
“Customers now expect this sort of thing. It’s just part of doing business now.”
Impact
In September, Allianz reported it had identified more than 15,800 fraudulent cases across personal, commercial and specialty lines, worth a total of £92.6m in the first half of 2025.
Cox said:
“There was obviously the portion of that fraud that has been identified through the Clearspeed tool that we perhaps would have identified through other checks and balances much later down the line, sometimes when it was a bit too late to do anything about it or to manage the cost effectively.
“But it was also identifying a large proportion of fraud that we wouldn’t have found through other techniques.”
He added that one consequence was a much greater volume of work going into fraud teams.
“We needed to make sure they were resourced and skilled up to deal with that, so that didn’t become a bottleneck.
“That was one of our learning points. I don’t think we’d quite appreciated what the impact on that team would be.”
Cox concluded:
“One of the other things we’ve seen coming through from the data and using this technology is a number of people self-select out part way through the claims, because they recognise that we’ve made the environment difficult.”
See article at Insurance Post.
