Allstate detects and prevents insurance fraud with advanced analytics in Tableau


Detect fraud with full view of all claims
Analyze billions of data points to spot trends
Compare regional trends with mapping

The Allstate Corporation is the largest publicly held personal lines property and casualty insurer in the United States, serving more than 16 million households. Allstate Insurance's claims department has a Special Investigation Unit (SIU) that prevents and deters insurance fraud using analytics.

With Tableau, the SIU consolidated many data sources into one view to analyze the full spectrum of claims information. Today, the unit detects out-of-pattern trends within a few clicks, a process that previously involved months of iteration. The ability to spot outliers in a glance speeds the handoff process to investigators and shortens the claims cycle. This early fraud detection increased efficiency across the department and helped position Allstate as an industry leader.

As a disclaimer: This is not an endorsement. Neither Allstate nor its subsidiaries endorse or recommend the use of any software products or data. While Allstate utilizes various innovative technologies and data sources, anyone should review all the products in detail to make sure they meet their needs prior to purchasing them. Further, Allstate is not providing any advice on the use or application of any of the products in conducting the demonstration. Data used to create this demonstration has been adjusted and does not include any propriety Allstate information.

Before it was always numbers in rows and columns. Now I’m eager to see when I do a new analysis, because it's so much easier to actually see the trends, the story, and the power the data brings.

Combating fraud with a universal view of claims

Insurance fraud ultimately impacts all industry insurance customers. To combat this risk, Allstate's Special Investigation Unit (SIU) seamlessly combs millions of records within minutes to find the "needle in the haystack that represents fraud or out-of-pattern behavior."

The SIU team adopted Tableau in 2010 when existing solutions couldn't handle Allstate's 1.5 billion records of unstructured data. Instead of taking months to collect data in Excel and wade through rows and columns of data, analysts can now quickly identify out-of-pattern trends and answer questions.

"Tableau gives us an ability to merge multiple data sources like Hadoop and Oracle, and at the same time, does not require the end user to deeply understand each of the 50 different data sources in order to answer a question," explains Marta Magnuszewska, Senior Analytics Leader for Allstate Insurance Claims.

Instead of reviewing a sample, Allstate can now conduct a full review of all claims to detect fraud, without sacrificing the payment timeframe or holding up the payment of legitimate claims. Today, Tableau is the business intelligence platform of choice at Allstate for analysis and reporting across teams, including risk, operational, and financial teams.

"Tableau has unleashed a creativity in me as an individual, and with that problem solving, it has unleashed the power of data," says Marta. "Tableau has become the tool of choice to do our analysis and reporting."

Data used to create this demonstration has been adjusted and does not include any propriety Allstate information.

Tableau visuals make fraud obvious and they help me tell the stories behind the data.

Investigators identify sources of out-of-pattern behavior with Tableau

Allstate investigators gather initial information from various internal and external sources to form hypotheses. To test these hypotheses, Allstate's SIU team analyzes upwards of 1.5 billion data points in Tableau to understand the claim lifecycle to determine normal versus out-of-pattern trends. Marta calls Tableau a "game changer" for Allstate's SIU Analytics team, decreasing false positives and increasing the team's confidence level in their analyses.

Data used to create this demonstration has been adjusted and does not include any propriety Allstate information.

The unit combines historical and current claims data to develop a fast and efficient "breadcrumbs" approach to analysis. With maps in Tableau, the SIU team can compare trends by state and dig deeper to understand the region driving the behavior along with the individuals behind the claims. In attorney-driven cases, the team can track the claim lifecycle over a period of time within a particular region to identify when an attorney is working with a medical provider to profit from claims.

The unit also analyzes medical claims data, looking into treatment details and current procedural terminology (CPT) codes to ensure that medical providers are charging for the proper treatments. For example, if an optometrist is billing under a physical therapy code, the team can flag it to investigate further.

Tableau has unleashed a creativity in me as an individual, and with that problem solving, it has unleashed the power of data. Tableau has become the tool of choice to do our analysis and reporting.

Data used to create this demonstration has been adjusted and does not include any propriety Allstate information.

When SIU investigators determine that a trend is fraudulent, they can utilize Tableau visuals to showcase the data in court: "We have uncovered extreme information that eventually led to multiple arrests for committing insurance fraud," shared Marta. "When we present fraud cases in court, investigators can pinpoint out-of-pattern trends using those visuals and understand what is going on without trying to explain the whole process of claim and of claim settlement."

SIU data consultants now share findings with investigators over Tableau Server, increasing efficiency across the entire department. Investigators can determine the viability of their hypotheses within a few clicks, narrowing the cycle time and catching fraudulent behavior faster.

"Tableau helped position Allstate SIU as an industry leader. We use advanced data analytics to fight against insurance fraud and protect Allstate customers."

You might also be interested in...