The Everybody Wins Casino promises customers a win rate of 95 percent. Liking those odds, you rush inside and grab a seat at the first open table you can find.

After a few losing hours, and several hundred dollars later, you naturally start wondering: Are you just cosmically unlucky, or did you miss something? As you hand over your last dollar, the dealer tells you that’s just how it goes at the 40 percent table. Maybe, the dealer suggests with a coy grin, you should try your luck at the casino’s 95 percent game just down the hall.

Looks like you missed something.

Knowing what to look for in small commercial underwriting data

What goes for hypothetical casinos applies to small commercial underwriting data. Sometimes, you have to do a bit of digging to understand what you’re really getting.

Take data models.

What if a model offers a very high accuracy rate for returning data on small commercial risks? Sounds promising. But look closer. Is that an accuracy rate for the specific industry classes you write, or want to write, or for ones you’re not interested in? Is it an accuracy rate for a single model or an aggregate percentage from a multitude of models? Is it accurate enough for your real-world needs?

Or take prefill data for insurance applications. By auto-filling firmographic details such as business address, business type (SIC/NAICS codes), payroll, and number of employees, underwriters don’t need to chase down applicants for that information. Prefill data can be judged by its “hit rate” and its “fill rate.” Both are worth unpacking closely.

A hit rate refers to how often a business is successfully located in a database. What it means to locate a business—i.e., what data about the business is actually being verified—may only be a business registration record. A high hit rate sounds ideal but may not actually be telling you much more than this: At one point in time, this business existed.

Then there’s fill rate. This refers to how much of an application can be accurately filled in. Ideally, you want both a high hit rate and a high fill rate, but what constitutes a high fill rate can vary. A “fill rate” of 100 percent sounds great, but it may only be limited to a few application fields, like business name, street address, or an owner’s name. A lower fill rate that nonetheless populates a broader set of fields may give you more useable data.

Understanding these differences can make all the difference to your small commercial underwriting success.

Data quality, unpacked

When looking for sources of high-quality underwriting data on small commercial risks, these nuances are just the beginning of a deeper conversation around data providers and data quality.

Our new whitepaper, Hit the gas: How to find high-quality data providers to accelerate small commercial underwriting, breaks down six essential characteristics of a high-quality small commercial data provider, spanning everything from the sourcing and curation of proprietary and third-party datasets, to the compliance and data governance concerns around the use of potentially sensitive data in modeling and underwriting.

For insurers looking to accelerate and automate their underwriting processes or minimize errors and improve loss ratios, high-quality underwriting data makes it all possible. You just have to know what to look for.

By Grace Jividen and Othman Loudghiri