Enhance internal and external data for better decision-making

There is nothing more frustrating than knowing you have just the piece of evidence or information you need to solve a problem, but have no idea how or where to locate it. This is a daily challenge for underwriters who know they have a mountain of useful information to help them write a policy or process a submission, but are hampered at every turn by complex systems and departmental siloes.

How to break down those siloes is something legacy insurers have been wrestling with for years. It’s an ongoing problem. A data catalogue powered by a knowledge graph will not remove those siloes, but it is capable of traversing them, bringing all the information the underwriter needs right to their virtual door.

In his session at Underwriting Innovation USA on November 10, Brett Hurt, chief executive officer and co-founder of data.world, and insurance industry veteran John Lucker will talk about the data discovery and governance challenges facing the insurance industry today and spotlight some modern technology solutions that can streamline processes and reduce the risk involved in data work.

Hurt spoke to Intelligent Insurer ahead of the event.

What is a knowledge graph?

If you’ve ever searched on Google, connected with a friend on Facebook, or purchased an item on Amazon, you’ve benefited from the power of a knowledge graph—it is a representation of real-world entities and the relationships between them.

Context is key. Take a Google search for ‘Paris’, for example. Using a knowledge graph for the search helps Google infer that you’re searching for Paris, France rather than Paris, Texas, Plaster of Paris, or Paris Hilton. It is the ultimate foundation for machine learning to operate on top of, because you’ll never get to some kind of Star Trek vision of the future where you just ask the computer any question, without it being somehow able to infer context based on lots of well-documented, well-understood data.

Insurers are the ultimate data companies and the easier they make it for their team members to access well documented and governed data, the more competitive they’ll be. A lot of carriers are still operating in very old models to leverage data. Underwriters typically get external modelling inputs via email chains—they email spreadsheets back and forth and search inboxes for quarterly inputs to create and compare and then email on to someone else.

Most don’t have anything like a modern data catalogue that would make that data usable across the company. There’s so much inefficiency there. One major insurer told me that they spend 70 percent of their time just looking for data to help them create products and measure their ultimate profitability.

The power of a data catalogue built on a knowledge graph is that it can take in lots of disparate data, often stored in silos, and connect it in an almost magical way. You can take your models and connect the data together to run things such as federated queries across many data silos.

What impact does a data catalogue powered by a knowledge graph have on an underwriter’s workflows?

Data catalogues help businesses organise, discover, govern, query, and collaborate on data. Not all catalogues are powered by a knowledge graph, but the ones that are tend to organise data in a very people-oriented way. This dramatically shortens the amount of time new team members need to learn the system.

As their colleagues transition out of the business, all their experience and decisions are still documented, and all those queries they ran to make their underwriting decisions are retained. It means that everyone in the business is on the same page when making important company decisions.

The COVID-19 pandemic accelerated the need for knowledge graph-powered data catalogues as it wasn’t just the data that was siloed, but also the people. A knowledge graph allows you to traverse a graph of data in the same way that HTML links to lots of different pages. With a zip code in a knowledge graph, you can traverse all the datasets containing those zip codes.

How do knowledge graphs solve the ongoing issue of siloed information?

Knowledge graphs are inherently connected. Whether information is in a spreadsheet, Snowflake, Oracle, Tableau, Microsoft Power BI, or more, the underlying knowledge graph connects all the internal infrastructure and analytics into a single platform to make it searchable.

It’s also essential that it has an interface that everyone in the company can use, not just the data practitioners. Think of your data catalogue as a corporate brain—all the knowledge about customers, products, sales, and additional external data that helps create more products. With the whole company, including underwriters, connecting to it, you’re able to create competitive offers. It’s all about empowerment.

What do you hope attendees will take away from this session?

I hope attendees come away from the session thinking that they have all this data infrastructure and powerful information they’ve grown or bought. That it’s great they can hold all of this information but that it would be much better if more people in the company could use it, and access that information via an interface that is as easy to use as Google.

Data catalogues don’t replace existing data sources, warehouses, or technologies. They complement the data infrastructure to make it more usable. Everyone has been selling the data-democratisation future but they’ve missed the part that it’s buried in siloes and used only by elite analysts. That’s a huge missing link, and data catalogues and knowledge graphs can fill it.

On November 10 at 3:50 pm, data.world’s chief executive officer and co-founder, Brett Hurt, will describe to Underwriting Innovation USA attendees a new style of data management and enhanced analytics.

This is the biggest online event tailored specifically to underwriting: don’t miss out. Register now!