Fewer fraud cases, bigger losses: why insurers need to look further upstream

Insurance fraud is turning into a smaller number of much bigger problems. That matters more than the headline total, because it changes where insurers should be spending their attention.

The Association of British Insurers published its 2025 fraud figures this week. Insurers detected £1.34 billion of fraudulent claims, up 14% on the £1.18 billion found the year before. Yet the number of cases fell 2.7% to 93,900. That is still more than 250 bogus claims caught every day, but each one is now worth more. The average detected fraudulent claim reached £14,300, the second highest on record and only just below the 2022 peak.

Reading the numbers properly

The pattern repeats across almost every line of business. Motor remains the largest source, accounting for 55% of detected cases, with 51,900 fraudulent claims worth £625 million. The average motor fraud is now around £12,000, the second highest level since 2015. Property looks much the same. Detected cases fell 6.2% to 17,700, while the total value rose 3.4% to £201 million and the average reached a record £11,400.

Travel is the exception on volume. Cases jumped 85% to 4,500, although at £8.6 million the value is modest by comparison.

Exaggerated loss is still the most common tactic, with 26,900 cases. The category that caught my eye is the fastest growing one. Contrived and staged incidents rose 30% to around 5,600 cases.

Strip away the headlines and the picture is simple. The opportunist padding out a genuine claim is still with us, but the money is moving towards people who plan.

Planned fraud leaves a trail

The case studies the ABI published alongside its figures show what planning looks like. One fraudster submitted a run of fabricated home insurance claims for stolen Lego sets and received a 28 month prison sentence. Another ran a crash for cash scheme, staging collisions with women he had met online. A third case involved a £300,000 travel insurance scam built on fake identities, with fabricated medical documents and websites created to support the claims.

What these cases share is preparation. A staged crash needs willing participants and a paper trail that holds up. A fabricated medical claim needs an identity that will pass a check. That work is done weeks or months before anything reaches a claims handler, and it leaves traces. The question for insurers is whether they are looking in the right places early enough.

The front door matters most

For me, the more telling figure in this year’s release is the £1.8 billion of fraudulent applications insurers stopped last year, with detected application fraud cases up 24%.

Every fraudulent claim begins with a policy someone was allowed to buy. In my 35 years in commercial data, the cheapest place to stop a loss has always been the point of acceptance. Once a policy is live, the insurer is reacting to whatever arrives. At application, it still has a choice.

The business behind the policy

The ABI’s headline numbers lean heavily towards personal lines, but the same logic applies with even more force in commercial insurance. A business is an easier disguise than a person. Companies can be formed cheaply, left dormant and revived when they become useful. Directors with a history of failed or dissolved companies can reappear behind a new name.

That is why I would put director checks at the very start of commercial underwriting. A bank would not onboard a new business client without verifying who runs it, and an insurer taking on a commercial risk has every reason to do the same.

Some of this is starting to change. The Economic Crime and Corporate Transparency Act has introduced identity verification for directors and people with significant control at Companies House, and the transition period for existing directors closes in November. Progress has been slow. Companies House management information covering the second quarter showed only 55% of directors and 42% of PSCs had verified by the end of June. A verified register will help, but it only confirms that a person is who they say they are. It tells an underwriter nothing about whether the business they run behaves like a genuine trading company.

So the checks at acceptance have to go further. An underwriter wants to know whether the company is actually trading or is a shell that has recently come back to life, and whether the people behind it have disqualifications, past insolvencies or a cluster of linked companies that failed in quick succession. Fake companies are often easier to spot than people assume, because the signs sit in the public record: accounts that are overdue or far too thin for the business being described, or a complete change of directors a few weeks before cover is taken out.

Those are the patterns our director risk indicators and Vigilance™ signals were designed to pick up. They draw on filed records and court data, which a fraudster cannot edit, so they stand up far better than anything typed into an application form.

From raw filings to real-time risk

Chris Oatts on why the next era of fraud detection starts in the data supply chain, and what ECCTA’s identity checks will and won’t fix.

Read the article

What AI changes

In my view, the biggest risk over the next few years is the cost of fabrication falling sharply. Generative AI makes convincing documents, and even whole websites, cheap and quick to produce. Any approach that depends on spotting a bad document is in an arms race, and the fraudster’s costs are falling faster than the insurer’s.

Independent data shifts that balance. A fraudster can produce a convincing invoice in seconds. Five years of filed accounts or a credible director history are much harder to fake. The more insurers anchor their decisions to evidence the applicant does not control, the less cheap fakery matters.

The same applies after inception. A business that looked sound at application can deteriorate quickly, and financial distress is often where opportunistic fraud starts. Monitoring director changes and financial health across a commercial book in real time, between renewals as well as at them, gives insurers a chance to act before a struggling policyholder becomes a fraudulent claim. In our experience, a falling H-Score® or a run of new county court judgments usually shows up well before a business fails, and that is the window insurers want to be working in.

Raising the bar

The industry deserves credit for what it is already catching. These figures reflect serious work by counter fraud teams and their partners in law enforcement. But fraud is getting larger and better organised, and the tools for creating false identities and documents get cheaper every month.

If I were asked for the most effective ways to combat business fraud in commercial lines, I would start with two. Check the company and its directors properly before the policy goes on risk, and keep watching once it has. Neither idea is new, but with the average detected fraud now above £14,000, the cost of skipping them has gone up.

If you work in underwriting or counter fraud and are thinking about how to do this well, I would welcome a conversation.

Sources: Association of British Insurers, 6 October 2026; Companies House management information, July 2026