Revitalising submissions for greater efficiency and profitability

Ahead of the event Tracy Staniland, director of marketing at Chisel AI, outlines why commercial lines insurers should be taking an active interest in integrating automation into their submission intake process.

Today’s commercial insurance submission process is laden with static PDFs and eFaxes. Underwriters are manually reading through these documents, emails and attachments—of which there can be 30 or more. They are currently spending nearly half their day extracting key datapoints and manually assessing how to underwrite a risk or decline it.

These time-consuming administrative tasks are replacing more valuable activities that draw on their wealth of expertise, such as making strategic judgements about overall risk appetite, portfolio strength or nurturing and broadening broker relationships. Automation has the potential to change all this.

By using software powered by AI, natural language processing (NLP) and machine learning (ML), underwriters can extract data automatically from these submissions, taking unstructured information and structuring it in extensible markup language (XML) or JavaScript Object Notation (JSON) files that can be transmitted or auto-populated into downstream systems.

This information can auto-populate customer relationship management programmes, clearance and registration, rating engines and more. As a result, underwriters can focus on more value-added tasks rather than simply rekeying information.

Different solutions explained

Sometimes it can seem as though the terms AI, ML, NLP and others are thrown about as a kind of catch-all to represent automated solutions of some kind. In this context, AI is intelligent document processing, contextually understanding the data.

NLP has a similar role—recognising a noun or an entity such as limits, deductibles, premiums or exclusions within a document.

Then there is supervised ML. Here, you are extracting the data and presenting it in a graphical user interface, which means that when a human looks at it, they can check the confidence score and see if the machine has extracted and interpreted a value correctly. If not, the human will correct it and send it back through the system—known as human-in-the-loop (HITL).

Unlike traditional software, AI is not quite a plug-and-play system. When working with carriers, the first thing we’ll ask them for is data. From that collection of policies or binders, we use the data to train the AI models. The more breadth and depth of data provided to the system, the more accurate it becomes over time.

Automation as driver of future growth

While there are immediate benefits from introducing workflow efficiencies through automation, the longer-term opportunity is to free up underwriter expertise to focus on strategic initiatives such as risk judgement and nurturing broker relationships.

Given the tight profit margins and challenging growth market, anywhere insurers can find gains will be central to their competitiveness.

Automating data flows means carriers can access much more information and perform analytics on it. Not confined to pulling specific pieces of data for a single submission, automation can help insurers ingest a broad spectrum of contextual information that can be funnelled into an enterprise data store for future market analysis—and perform future reviews on submissions to see what types of business are, most commonly, coming in.

In the near future, this holistic view will provide insurers with insights that were previously unavailable.

Building on customer experience

If the submissions process is onerous for the underwriter, it’s an equally drawn-out and sometimes frustrating experience for the potential customer. When brokers submit applications to carriers, it can be at least two or three weeks before any response is returned.

The response might be that the carrier doesn’t want to write that business, or worse, there is no response at all.

With an AI solution, you can auto-route submissions to the correct underwriting teams and even set parameters so the system can auto-decline. A negative answer may not seem as though it’s delivering great customer experience, but if the customer can quickly go on to get coverage elsewhere they will have a positive view of the carrier and potentially become an advocate—if not a paying client.

Everyone is a data scientist

Simplistic user interfaces are key here. So many people within an insurance organisation need to be able to access and understand the unstructured data locked away in submissions, and there is no room to wait days or weeks for underwriting staff to manually extract, analyse or copy-and-paste key data across multiple systems.

Expertise will always be needed, but with a simplistic user interface, more underwriters and staff can access more information and process the more straightforward policies quicker. It does require a small amount of training and education, but it’s nothing that experienced underwriters can’t handle.

Now that COVID-19 has fuelled something of a transition to remote and digital-first behaviours, there is no longer the option to stay the course with a slower, more traditional manual approach. As younger underwriters from digitally native backgrounds enter the industry, an automated intelligence approach will be expected.

The next generation will not expect to spend half the day rekeying data. Adoption of AI-powered underwriting workflows is growing, while market dynamics and customer expectations are rising all the time. It’s up to insurers to embrace innovation to meet them.