This story begins two hundred years ago.

Two hundred years ago, you could only expect to live to age 29. Now the average worldwide life expectancy is over 70 years, and above 80 years in many developed countries.¹ What changed? And what does Groundspeed – a commercial insurance technology company – have to do with it?

There are many causes of this titanic shift. One of them is that fewer people were needed to perform agricultural work. Fewer workers needed for agriculture meant more people could take part in the wave of industrialization that fueled a long-term rise in quality of life.^{2, 3, 4, 5} 

One major cause of the shift away from agricultural work was improved farm equipment, perhaps best exemplified by the combine harvester, or “combine,” and its predecessors.⁶ Even if you’ve never lived in an agricultural community, you’ve certainly seen videos of machines harvesting grain in fields – those are combines. These machines take in whole plants and process them in the field so grain can be shipped off to make food. This removes the need to spend countless hours cutting down stalks, threshing grain from straw, and winnowing out inedible chaff.

At Groundspeed, we didn’t invent the combine. But for commercial insurance document processing, we invented the next best thing. Our Artificial Intelligence data pipeline does more than harvest your data; it separates the wheat from the chaff and serves it up freshly milled, ready to bake some delicious underwriting success.

How do you achieve underwriting success?

Underwriting success can mean something different at each carrier, though it generally boils down to keeping loss and expense ratios low while growing your book of business.

To achieve that underwriting success, getting information out of documents is critical. As an underwriter, you must decide, most importantly, whether to insure a firm and what premiums to charge. For that, you need to get a lot out of document data. That includes:

1. Risks to a prospective insured from loss history and risk exposure
2. Understanding of how a prospective insured lines up with:

• The lines of business you support

• The profile of firms you’re aiming to insure

• Other prospective insureds you could engage as customers

We’ve covered this in more detail in previous blog posts on Artificial Intelligence in commercial insurance and Groundspeed’s Artificial Intelligence with Human in the Loop system, automations, and loss run processing.

But you need to do more than know what data commercial insurance documents contain. You must do a lot to make that data useful to your underwriting team. Plus, there are things you can do beyond that bare minimum that can slingshot you ahead of your competition.

Groundspeed can help by making it easy for you to get the data you absolutely need, plus information – from both documents you provide and third-party sources – that you either don’t yet use or take a lot of manual time to retrieve. With Groundspeed, data comes back to you quickly with high accuracy and competitive cost.

Retrieving the data

To make data available to underwriters, you need to go through several steps. Here’s what needs to happen in the underwriting process once you’ve gathered your documents:

1. Read documents
2. Capture data, and understand the pairing of field (such as incurred value on a claim) and value (such as $20,000)
3. Calculate summary values (such as net total incurred)
4. Enable comparison across documents, even those from different sources. Field names might differ between documents, so you need to consider that.
5. Research and find data points from other sources, such as motor vehicle records or Google Maps.
6. Do a data quality check.
7. Package up information into a format that the underwriters can use.
8. Get the information to a place where your underwriters can use it.

There are two main areas of document collection that can take you beyond your competition. The first is saving time by automatically calculating summary values and predicting others so your team doesn’t need to. Summary values include the aforementioned net total incurred. However, some values are more difficult to infer from contexts, such as line of business. For those, you need more complex heuristics and machine learning models to fill in.

Joining third-party data sources to the data you collect can also save your team research time, while broadening your access to data you might not yet be accessing because of the effort it takes to retrieve them. Third-party data can give you access to fields such as roofing material and distance to the nearest fire hydrant for commercial property lines of business, seating capacity, and the number of axles for commercial auto lines of business, among others. The more complete your information about a prospective insured, the more competitive edge you’ll have over other carriers.

How you can get that data

There are a few ways you can go from raw documents to packaged data and get summary, predicted, and third-party fields along the way. But, unfortunately, most of them have their downsides.

• In-house manual processing – reading, standardizing in one format, and packaging up data takes a lot of time and is error-prone. Plus, you don’t get the enrichments without a lot of additional labor.
• In-house IT – internal software programs take years to develop and test and come with an opportunity cost; all of that time and money could be spent on other endeavors.
• Business Process Outsourcing (BPO) firms – when you outsource, you either don’t get the summary, predicted, or 3rd party data, or you do, and it takes a long time and/or is low quality and untrustworthy.

To avoid all of those downsides, you can go with Groundspeed.

How Groundspeed can help

The combine harvester was developed over decades and has been iterated for nearly two centuries. Drawing on our deep industry knowledge and ever-growing market experience, we continue to develop, iterate, and deliver an increasing number of enhancements that provide our customers with clean, immediately actionable data and predictive insights that leverage our massive insurance data set. Plus, we get you that data quickly, with high accuracy, and at a competitive price.

Here’s an overview of how we do that right now:

1. We use multiple Optical Character Recognition (OCR) engines and image cleanup techniques to read documents automatically so people don’t need to.
2. We capture a complete set of data fields from each document, as described in our Artificial Intelligence with Human in the Loop blog post here.
3. Calculations are used to fill in missing financial values, calculate others (for example, lag days for a claim), and ultimately, summarize customer data in a unified way across different document presentations. This represents significant time savings and cost savings for clients.
4. Predicted fields come from Groundspeed’s Artificial Intelligence and Machine Learning models. These models look at the information available in your documents and make an informed decision about what the right field value should be, much like a human would. We do this for line of business, sub coverage, cause of injury, body part, and more.
5. Normalizations (for example, carrier name and litigation status) allow us to present raw data in a standardized way that renders it easily legible and comparable. This can save significant time and cost for commercial insurance carriers as well.
6. Third-party Integrations allow us to provide additional features you might not support in-house automatically. Some examples of what we provide include location and company name matches, including latitude and longitude coordinates, NAICS codes, and SIC codes. Plus, VIN number-based lookups, including vehicle make and model, seating capacity, number of axles, and gross weight
7. Quality Assurance (QA) checks ensure the accurate capture of hard-to-read documents. Our pipeline includes a series of QA checks to proactively catch potential issues.
8. Groundspeed packages all data into a format that’s the same every time, so you know the information you’re receiving and where to find it. Whether in JSON or an Excel document, this data can fit into almost every carrier’s underwriting data system.
9. Our API, Email, and SFTP document transfer options mean that you can get your data back from Groundspeed in a way that makes sense for your IT requirements.

Groundspeed continues to add enrichments as our insurance data set grows. Additional enrichments are planned, prioritized, and released based on Groundspeed’s understanding of market needs and customer requests.

Let’s get in touch

With Grounspeed, you can get your underwriting team the high-quality data they need to win more business while keeping loss and expense ratios low. We’ve made the metaphorical combine harvester – all you need to do is farm those fields of prospective insureds.

If you would like to learn more about Groundspeed’s AI solution in general or our data pipeline and enrichments in particular, then schedule a call with our team today. We look forward to partnering with your company and helping you unlock the value in your unstructured data.

Sources

1. Source – life expectancy over time
2. Source – less labor required for agriculture enabled early industrialization
3. Source – industrial workforce in second industrial revolution was in part former agricultural workers
4. Source – industrialization caused quality of life gains in long term, including life expectancy
5. Source – life expectancy and overall quality of life rose in England during industrialization
6. Source – about combine harvesters, and that they were a major cause of less agricultural labor being needed per acre of farmland

This blog was written by:

Bryan Quandt – Bryan is the Product Manager for Groundspeed’s internal software applications. He plans the development of automations and Human in the Loop tools that get Groundspeed’s customers their data as accurately, quickly, and inexpensively as possible.