Are you missing fraud with outdated tech?

Fraud analysis is a specialized form of data investigation where you are looking for bad actors — for example, people who submit fraudulent insurance claims. The challenges, in terms of data analysis, are the same as with good business analytics in the right context — you want the most up-to-date, relevant information about the problem.

Ten years ago, you couldn’t find this information — or if you could locate it, you couldn’t get to it. That problem has been solved. Today, there is an overwhelming amount of information available to companies. The challenge now is that the traditional IT mechanisms we use to manage that data are not designed for combining data. Rather, they are designed to keep data apart and are used to answer very specific, narrowly focused questions.

Related:  Social media’s role in claims

For example, in insurance, if you want to see all claim amounts and compare them to information in the policies, then perform a statistical analysis to determine which policies pay out certain amounts or within certain ranges and how that impacts policy value, there are great traditional systems for doing that very specific thing, very fast.

But what if you wanted to examine this information within the context of the policy — the combination of policy, vehicle, vehicle owner, owner address, previous address, other vehicles they’ve owned, previous policies with other carriers, previous claims, others living in their household and the vehicles they own? Traditional systems can’t merge that information – they are not designed for that. But without this information you don’t have the whole picture, and are essentially peeking through a keyhole to understand what is on the other side, and whether or not it constitutes fraudulent behavior.

Inertia

One of the biggest challenges companies have with changing how they aggregate internal and external third-party data is inertia. Business analysts find great sources of data and can see its immediate value for fraud analysis. They ask their IT department for data and that’s usually okay because IT just needs to put the data somewhere.

However, when the business analysts say, “I want to use the data in the context of my business,” those requests frequently go nowhere because few businesses have the budget, time or institutional will to break out of this cycle. In addition, data challenges are evolving too fast for IT departments to handle. Data freshness is essential to rigorous counter-fraud techniques.

Experience shows that stakeholder consensus around the data model is fundamental to responding quickly to evolving data challenges like fraud. For the first 50 years of computing, data models were the domain of data architects, data administrators and programmers. But if you can draw a model that is easily understood and validated by executives and subject matter experts, claim modernization can happen on a completely different timescale.

From an analytical standpoint, companies can and should do more to zero in on the relationships that connect customers, claims, policies, property, modes of communication, demographics, transactions and other key entities. These links provide context. Potential fraudsters are very aware of counter-fraud techniques — they study them. They know how difficult it is for most insurance companies to connect all the dots.

Related:  Insurer questions impact of athlete’s injuries

For example, if I were to commit fraud, my insurance company could easily look at my claims history and, based on my behavior, use some pretty sophisticated algorithms to reveal a pattern. The bad guys know this. So what do they do? They create false identities so that it looks like different people who have the insurance, but it’s not — it’s the same person. They use burner cell phones, keep multiple email addresses and create fake social media accounts. They use friends’ mailing addresses or use multiple people living at the same address. They do all of this is because they know those connections live outside the context insurance companies can use to analyze the situation.

Modernize claims infrastructure: 3 major considerations

For companies who want to modernize their claims infrastructure there are three major considerations:

1. Your technology system must handle (model) data in a way that is easily understood and validated by business users.
2. Your data must be connected — it must reside in the same place and in an effective manner to be meaningful in any kind of business context. Your IT department might tell you they can connect all your data from seven or eight different systems. That is not true — you can try, but it’s incredibly expensive and prone to failure, usually after costing a lot of time and money.
3. Lastly, you don’t have a lot of time. If your IT department says they need six months or longer to get this done, you are probably missing the window to illuminate fraud because your adversary is more agile.

Clark Richey (@crichey) is the chief technology officer at FactGem and has over 20 years of experience designing and developing software, primarily for the defense and intelligence sectors. He has investigated non-traditional methods and technologies that use data more efficiently for over 10 years. For more information, contact him at gems@factgem.com.