4 types of data that we wish you had for pricing insights
You can watch our video on this topic on YouTube
At Octagram, we specialize in squeezing out underwriting and pricing insights from data. We've seen data from large carriers to small MGAs, and everything in between. From this experience, here is our wishlist of 4 types of data that are commonly missing, and that we wish you had, to get you competitive insights.
“Here’s our wishlist of 4 types of data that are commonly missing, and that we wish you had.”
Match claims to exposures.
Being able to match a claim to an exposure lets you understand the types of exposures that are driving losses. For example, if you are an auto carrier with 2 drivers and 2 vehicles on one policy, and there is a claim, can you attach a claim to a particular driver and vehicle?
Why it’s a problem if you don’t have this data
If you can’t match a claim to an exposure, then you're stuck with insights at the policy level, which often means taking averages. For example, let's say the two drivers on the policy were 20 and 60. The average age on the policy is 40, and you may lump this policy in with policies where there are two 40-year-old drivers. When you do this, the insights are limited.
This is a common problem for both large established carriers and small MGAs. It's a problem across many lines, not just auto. For example, in commercial property, it's challenging to associate a claim to a particular property on a schedule of locations.
Why you don’t have this data
When I tell some people this, they are surprised. How is this possible? Of course, at an individual claim level, the claims handler can determine which driver and vehicle were involved in the claim. They can look at the list of drivers and vehicles, communicate with those involved, and process the claim. But when you need to match every claim to an exposure for thousands of records, you need a unique exposure identifier attached to the claim, and this often does not exist.
This is because the claims system and underwriting system are typically built and updated independently. In particular, the claims organization alone will not think to prioritize a unique exposure identifier. Their goal is to get good claims outcomes rather than to get insights for underwriting.
How to get this data
It takes a thoughtful data strategy and governance to prioritize and find budget for a unique exposure identifier in the claims system on behalf of the broader organization.
Clean cause of loss.
You may find that your cause of loss is some version of blank, "not available", "other" or "unknown" around 10%, 20% or sometimes 50% of the time. We see many carriers and MGAs struggle with poor quality cause of loss information. However, the best underwriting insights are determined for each cause of loss.
Why it’s a problem if you don’t have this data
For example, a property writer may want to know the impact of implementing a water damage deductible. Without a cause of loss to identify these claims, this is unknowable.
Why are we focusing on this one field out of the many hundreds across underwriting and claims systems? This is the one field that we’ve seen time and time again stop an analysis in its tracks. It’s the one field where we’ve seen companies go back and painstakingly fill in for each claim.
Why you don’t have this data, and how to get it
To have a clean and reliable cause of loss, you should first understand why this is not filled in as well as you’d like. There are many reasons why. For example, one reason I have seen is that the cause of loss is filled in at first-notice-of-loss, where the cause is unclear, so "unknown" or “other” is often chosen. Cause of loss is understood as the claim develops, but no one updates this field.
This is one example of many, and there are different paths to improve the data quality depending on the cause of poor quality.
But no matter the reason, it’s a good idea to track the percentage of blanks, “unknowns,” and “other,” and have someone accountable for this metric to decrease over time.
Capturing prior losses from submissions.
Many carriers will ask for the last three years of loss for each submission. While many carriers keep submission data for pipeline metrics, many fail to keep data on the last three years of loss.
Why you should keep this data
If you do keep this data, then you can significantly increase the data you have for insights. For example, if you are binding 20% of the submissions you see, then you can 5x your data by keeping all the submission data at the same level of detail as you store your bound book of business. You then have a giant data set with both exposures and claims. You can use this in a couple of ways:
To confirm any tentative insights from your bound data. For example, you think restaurants with a Google review of 4.5+ or more perform better than those with a lower Google review. Your submissions have name, address, and the last three years of loss, so you can use this much larger dataset to test your hypothesis.
To provide insights for business areas you don't write but want to expand into. The submission data will highlight the premium volume you could see in this space as well as its profitability.
Your mileage on this data will vary depending on its source. For those writing directly online and also through comparative raters, the data quality may be too poor to use in the ways articulated above. For others, it may provide valuable insights for profitable growth.
Instant customer loss ratio.
You would like to know who your most and least profitable customers are. If your go-to-market strategy considers a customer’s whole account, executing and tracking that strategy's success is challenging without an automated way to roll up to a customer level.
However, it could take a week for many carriers to identify all the policies for a single customer. This is a particular problem in commercial lines where a customer might acquire, divest, and have subsidiaries.
Why you don’t have this data and how to get it
You need a unique customer identifier to see customer-wide loss ratios at the click of a button. If you are in commercial lines, this involves figuring out how you’ll roll up subsidiaries into parent companies and what you'll do as companies merge into one or separate.
As well as these challenges, there's the self-inflicted pain of different spellings of the same customer, like Walmart, Walmert, Walmart Inc. Walmartt.
Rather than assuming you'll come up with a perfect one-time solution, assume you'll need to constantly feed and water your unique customer identifier and continuously measure and improve quality.
Conclusion
At any carrier or MGA, there are three stages of data maturity.
Data for operations: This is the basics. You need data to flow through pipes to issue policies and pay claims. This data usually flows seamlessly through an insurance organization.
Data for financials: You need data for your P&L and performance metrics. For some carriers and MGAs, closing the books every month is seamless. For some, it's a huge task.
Data for insights: This is data to help you quantify the drivers of loss, over and above what you already collect for operations and financials.
For the 4 types of data on our wish list, what they have in common is that you don’t need them to operate your business or report your P&L. You do need them to get good underwriting and pricing insights.
“For the 4 types of data on our wish list, what they have in common is that you don’t need them to operate your business, or report your P&L. You do need them to get good underwriting and pricing insights. ”
Keep in mind that, even with data to operate your company and issue financials, that's not enough to generate good pricing and underwriting insights. And to have good data for insights, you don’t need to boil the ocean. We’ve listed just four important data types to keep in mind.