But in terms of insurers’ P&L, claims and underwriting are dwarfed by loss. That’s why reducing leakage is an area where even small improvements can make a big difference. And, as it turns out, an area where AI can be instrumental.
REDUCE LEAKAGE
- Detect possible fraud patterns
- Alert claims handlers for early mitigation
- Standardize payouts
Loss is insurers’ number one expense item and premium Leakage is routinely evaluated in the billions (see for example Claims at a Crossroads, Accenture, and The Challenge of Auto Insurance Premium Leakage, Verisk). AI can help chip away at this leakage.
A first example is in the claims workflow, where it can detect possible fraud patterns – for example through inconsistencies in accident descriptions. This helps by highlighting “devil in the detail” cases that would be more difficult to catch by claims handlers themselves, and prioritizing which cases to investigate. In a more general sense, beyond its productivity benefits in claims, AI can also help govern the claims process, for example flagging cases for adjusters’ attention when some aspects suggest a specific approach. A practical example of how this can help in terms of leakage is by supporting more standardized payouts. On the underwriting side, cognitive holds a lot of promise as well. Ongoing developments in the industry suggest that it can be applied at scale to improve customer intelligence and therefore underwriting models.
Getting started with your business case
As these examples illustrate, whether it’s for Claims, Underwriting Expenses or Leakage, AI can be instrumental in helping Insurance. For professionals looking to put this in practice, a key question is how to draft a business case for such applications. A framework that has been applied successfully in the industry is to approach things with a project portfolio approach. As a first step, looking across the business and identifying the specific lines of business or functions that are particularly exposed from the standpoint of expense or loss will suggest candidate application areas.
Drilling down into their economics, two quick metrics help frame the potential value of a solution:
- How many cases/transactions (claims or policies for example) does this area handle
- What target efficiency gain is realistic? (for example saving 3 hours per claim)
The next step is to evaluate the work of customizing and operating a cognitive platform that fits the needs of the application area. Contrasting different candidate application areas based on their respective potentials helps to point out the quick wins and more strategic projects. In light of other corporate initiatives this helps prioritize a cognitive roadmap for your business.
Final Thoughts
Convincing business cases for AI have already emerged across Insurance, whether it’s to control expenses or to reduce leakage. But AI in Insurance is still in its early days, and new applications are emerging every day –I’m pretty curious to hear your thoughts on the matter so please do get in touch and share your thoughts on where this exciting technology may take us! In the meantime if you’d like this post and its predecessor in a more portable format, you can download the full white paper here.
