As a software company that has three core products; Pivot Point for Brokers and MGAs; Data Vera for Binders; and Data Vera for Exposure Management, we always ask clients if they have data challenges

Data science sun life

AIG’s CEO Brian Duperreault has previously said “I think data science and algorithmic approaches to decision making, discovery of information, how we go about finding out what’s out there and how do we sift through it and come up with useable information – to me, that’s the real issue in our business.”

We agree. As a software company that has three core products; Pivot Point for Brokers and MGAs; Data Vera for Binders; and Data Vera for Exposure Management, we always ask clients if they have data challenges.

The response is usually a chuckle and then the comment “That’s a loaded question.” They’re right. For example, when completing an insurance application you are sometimes asked the state of your health. Certainly you hope you are in perfect health but most of us, fearing over-confidence, respond admitting only that we are in “good” health and leave unchecked the “perfect health” box.

We suspect that it’s the same when someone is asked if they have challenges with data. Luckily, in most meeting environments, people will eventually start to tick off their data ingress avenues one by one. They will talk about their effort required to clean, validate and format all that data so it can be used for underwriting analysis or for modeling. Surprisingly, people are often quite candid about admitting the scale of work involved in cleaning and validating data.

Occasionally people will tell us immediately if there are certain areas where they are painfully aware of deficiencies in their data collection, cleansing, validation and formatting efforts. This isn’t a surprise. Often, the larger and more complex a company becomes, a corresponding effect is seen in the level of data complexity and number of data sources involved.

That’s where we come in. We designed Data Vera initially for our binder application in response to the multiple types of bordereau received by our clients. Getting that information into an application that tracks premiums, claims and settled bordereau meant all that incoming data represented a gold mine of information –if it could be decoded. Data Vera uses algorithms and avoids the need to map to templates. Cleansing and validation can be done nearly automatically and certain clients rely on the Data Vera to the extent where they are only notified if an exception occurs.

When we first showed Data Vera to our clients we would sometimes be asked whether modelers had seen it. We would then show it to the client’s modelers who, while impressed with the functionality of the application, questioned whether large Statements of Value (SOVs) could be pushed through the system.

Our response was Data Vera for Exposure Management which can validate incoming SOVs of nearly any size; solves the thorny problems of both address verification and COPE analysis; and provides an analytics tool set that permits users to compare SOVs year on year to identify data changes instantly. The data is validated and can be exported to third party modeling services and the modeled result can be returned to Data Vera. Since the client now has the underlying risk information in a validated format the client can now adjust the modeled loss curves within the application to determine pricing and capacity effects.

We’ve seen from our own experience how data drives the transaction. To drive our own transaction applications our clients need to rely on their data. It’s the same dynamic for any transaction system or for any insurer, reinsurer, broker or MGA.

Data cascades through the insurance transaction starting when collected from the insured and runs all the way up the reinsurance chain. How many times that data is re-validated and re-formatted at each stop on that chain is anyone’s guess. Our mission is to ensure clients have the ability to push reliable data through the entire transaction process efficiently and less expensively than having highly qualified personnel occupied with data import cleansing and formatting processes.

The key, as Brian Duperreault said, is to sift through the data and come up with usable information. We can help.