The latest State of Insurance Fraud Technology from the Coalition Against Insurance Fraud shows some changing patterns over the past two years. The fifth iteration of the report, its results are based on the results of a 20-question survey answered by 80 members representing a strong majority of insurers in the United States across several product lines
Only four per cent do not use anti-fraud technology to detect claims fraud. Roughly two-thirds use such tools within the new business process and close to 40 per cent now use identity verification solutions.
“Identity verification is a relatively new anti-fraud technique,” the report states. “With the rise of blockchain technology and other digital identity solutions, it will likely be adopted rapidly by a significant portion of insurers over the next three to five years.”
Cyber fraud technology was used 22 per cent during the pandemic as more activity shifted to digital platforms, but the organization expects cyber fraud and identity verification tools to be employed as a primary technique by most companies.
Automated red flags (88 per cent), predictive modelling (80), reporting capability (64), case management (61), exception reporting (51), and data visualization/link analysis (51) are key components in an insurance company’s tool kit. Predictive modelling is seeing a high increase in use since the last study was conducted in 2018, when the rate was 55 per cent. Text mining nearly doubled from 33 to 65 per cent, while 31 per cent deploy photo recognition or analytics. More than half of companies built their system in house.
Image-based fraud prevention techniques are also increasingly popular. More than 35 per cent reported using the tactic in 2021.
“Photo recognition and analysis is becoming extremely important as more insurers look to save costs by not doing in-person inspections of vehicle property damage claims and even on more minor residential and commercial property claims,” the report states. “This technology allows insurers to know whether a photo of claimed damage is real; has been digitally altered; or has been submitted previously on other claims. Photo recognition technology allows for a world-wide search of the image, and even minute alterations or changes in a photo that would not be detected by the human eye through image data analysis.”
Almost 40 per cent of respondents said more than 30 per cent of their referrals came from their automated fraud detection solutions, almost double the 20 per cent that said it in 2018. One in six said their automated fraud detection solution provides 10 per cent or less of referrals, a decrease from 35 per cent in 2018.
Insurers are also changing how they measure the success of their anti-fraud efforts. Forty per cent now measure it against their loss ratio, up from 15 per cent in 2018 and four per cent in 2016. The switch may indicate a shift to considering these efforts as ROI. Tracking fraud cost savings in such a way is a hotly debatable topic, given some states have stronger bad faith laws where such analysis may be subject to discovery in litigation.
Strong data is an important part of a fraud management program, the report states, as it powers technology. Internal data is most relied on (100 per cent) followed by industry fraud-watch lists (88), unstructured data (81), public records (79), third-party data aggregators (55), social-media data (48), and data from connected devices (15). Unstructured data use is surging, as is that generated from social media.
Some of the most common benefits of anti-fraud technology cited are higher-quality referrals (55 per cent), more overall referrals (48) and better mitigation of losses determined to be fraudulent after investigation (33).
The biggest area of investment by far is in claims fraud at 71 per cent, followed by underwriting at 38 per cent. The authors found it surprising that the percentage citing underwriting has barely budged, given the priority of finding fraud as early as possible. They speculate anti-fraud technology, as an underwriting tool is still in earlier stages of development and adoption.
Predictive modelling investments look to be the most popular in the years ahead, with link analysis, AI, and automated red flags and business rules.
There are two main reasons companies invest in anti-fraud analytics. The first is improving the quality of referrals and the second is simple fraud detection. It is important to balance the two, the authors caution.
The anti-fraud technology industry is quickly changing, with AI, geotargeting, and automation among the advancements that will become increasingly popular in the coming years, the authors state. Adopting them comes with challenges, including limited IT resources (cited by 68 per cent), data integration and poor data quality among the biggest obstacles.
Financial considerations can be another hindrance, as two-thirds said their budgets will likely remain flat in the coming years, even though the same percentage said fraud against their companies has risen.
“There was a significant drop from 41 per cent in 2018 to 19 per cent in 2021 of respondents expecting additional funding, reflecting the pressure within insurers to curb costs,” the report states.