Part 1 of 2: Why automation across companies is so hard

In this two-part series, we will take a critical look at an aspect of many insurance technology initiatives: automated processing of data from one company to another.

In personal lines products, this is largely a solved problem. Agents today can enter a Motor or HO submission into an aggregator, get multiple quotes, and submit the data in one place.

But commercial lines are a different story. Practically every major commercial lines cross-company data processing initiative has failed: from the insurance exchange concepts of the 1980s to today’s TOM initiatives, the same things keep happening over and over.

Spoiler alert: we believe that most commercial lines cross-company automation is not worth it. If you’ve accomplished straight-through processing within the walls of your own company, you’re probably doing just fine.

This is going to be an unpopular opinion with many of our readers, who pinpoint poor data transfer as a key source of inefficiency that needs to be cleaned up. To them we say: bear with us. Hear us out and see if we can make the case that you should be doing less, not more. We thought a lot about this when we were designing Risk/R, and perhaps our conclusions will resonate.

The first place to start is with forms.

The Problem with Forms

One of the age-old nuisances in insurance is form submissions from agents/brokers to carriers (and vice versa). Forms are the number one method of transferring data across companies; no wonder people want to automate them.

Every policy admin system vendor wrestles with capturing data from business partners, and sending good data back. A number of insurtech startups are out there trying to tackle this space too. They typically take one of the following approaches:

• full-service digital wholesaler
• marketplace
• aggregator
• smart forms

That’s all well and good. But: using a form for a complex insurance submission leads to two types of errors. If you try to automate without an eye towards solving them, bad things will happen. Let’s call them Type I and Type II:

Type I Problems are when a form requires a piece of data which isn’t actually needed for the particular risk.

Type II Problems are when a form does not require a piece of data which is in fact needed for the particular risk.

Type I Problems

Most forms that we encounter in day-to-day life do not suffer from Type I Problems. When you apply for a vehicle registration or a birth certificate, there is no ambiguity in the situation: the state only asks you what they need to know.

But any insurance professional who has tried to streamline an online HO or Auto form knows what Type I is about. Your goal is trim the excess fields, and push the application through with as little inconvenience to the customer as possible.

More complex policies, like CPP, become ever more difficult to streamline. The typical solution for this is to build smart/interactive forms that only bubble up a question after the need for it has been established. Think TurboTax, and you’ve got the right mental model.

Type II Problems

Type II problems are where things get thorny. These are insidious for the insurance industry. There is always some relevant piece of data or narrative which doesn’t make it onto the form.

Here’s an example. We talked to an agent who placed a high net worth policy for a gentleman who lives on Daufuskie Island, South Carolina. There are no roads from Daufuskie to the mainland, only a ferry that leaves every few hours. This gentleman was into woodworking. But he also had quite a nice wine cellar, which he used frequently for entertaining. Wine… woodworking… no route to a hospital… what could possibly go wrong?! The agent wisely picked up the phone to convey the total narrative to her carriers.

This is an underlying reason for why insurance is a relationship business. A good professional communicates the whole story to her business partners. A bad one abdicates her responsibility because the question wasn’t on the form.

Our take on the issue

We have reached a surprising conclusion about forms. We think that the technology actually doesn’t yet exist to handle both Type I and II problems.

The natural way for a person to communicate the details of a risk is to tell the story: pick up the phone, talk to your partners, and talk about the pros and cons. The unnatural way is to fill out a form and hope to god it asks the right questions.

We can’t replicate that story-telling process with machines yet. The technology to make sense of the wine-and-woodworking aficionado on Daufuskie Island just hasn’t arrived.

So what’s left? We believe that for the moment, companies who communicate well and nurture their business relationships carefully are already in the best of all worlds. Don’t try to over-automate form submissions. Instead, emphasize the communication with your business partners that happens alongside those forms. You’ll probably see better business results.

That’s why we at Risk/R felt that honoring and elevating the story-telling process was the thing we needed to do.

In Part 2 of this series, we will look at some other reasons why automated processing across companies is so hard — and in fact, often isn’t worthwhile.