11.05.21
Bryan Schofield

Introducing the Corvus Cargo Score

Marine Cargo insurance is vaunted as the foundation of modern insurance, with its lineage dating to the insuring of ocean-crossing vessels, and their cargoes, in the fifteenth century. With all that history, change tends to come gradually to the line. (And with generally stable profitability, carriers could be forgiven for not rocking the boat.) More recently, technologies such as connected temperature sensors and GPS trackers have entered the picture for shippers. But the data they generate hasn’t been factored into insurance underwriting at scale. Today, we’re sharing an exciting update that will help to finally bridge Cargo Insurance into the 21st century.

Introducing the Corvus Cargo Score

[DIAGRAM] Corvus Cargo Score

To date, the benefits of Internet of Things (IoT) innovations for shipping have been limited to the minority of businesses who can easily place devices directly inside of their own shipments — and have a significant volume of sensitive goods to track. For a variety of reasons, ranging from cost to lack of control over the end-to-end logistics chain, it’s not always an option. So while Corvus already works with leading sensor providers including Sensitech and eProvenance to provide their customers with opportunities for enhanced coverage based on data these sensors collect, those programs can only reach so many policyholders. 

Now, Corvus is bringing the benefit of IoT insights to everyone. 

With the Corvus Cargo Score, we use historical sensor temperature data to predict the risk of any food shipment with just two pieces of information — no sensors required. With their score in hand, clients will better understand which products or routes may be riskier than others, and even get preferred terms from Corvus underwriters.

The Data Story: How we Built the Corvus Cargo Score 

The Corvus Data Science and Product teams based the score on a trove of data from a leading provider of temperature sensors. All in all, we worked with historical temperature readings for 1.3 million shipments across the U.S. 

With that much data, there’s a considerable scope for analytic findings — but we knew that for the score to be useful, we needed to base it on information that would be readily at hand for brokers and their clients. Thus, our goal was to be able to consistently identify high-risk shipments using a minimal combination of factors.

The team started by designating a group of shipments in the database as “high risk.” That was defined as any shipment whose temperature exceeds its temperature band for two or more consecutive hours. To identify the right temperature bands for each commodity we followed industry standards and added in a small buffer to account differences in temperature between the location of the commodity and the ambient sensor. 

While inheriting vendor data is a great way to gain insights from the real world, we still had to account for various data issues by undergoing a data cleaning process. For instance, for a number of shipments we needed to infer the actual start or finish dates, since it’s common for sensors to be left running long after a shipment concludes — thus registering days’ worth of meaningless temperature readings. Some shipments had obviously bad readings, such as a seafood shipment with an average temperature over 60 degrees Fahrenheit. We weeded out such outliers. 

Once the data cleaning process was done and we had defined and agreed upon our target, the real analysis work could begin. After looking at numerous factors in relation with high-risk shipments, we found the points at which an increase in trip length is most associated with riskier shipments, as well as how those findings interact with the risk of each product category.

Screen Shot 2021-11-05 at 12.59.05 PMFor example, two common sub-sets of shipments we identified were more than 83% less risky than the rest. Findings like these allowed us to develop a set of rules that would allow our underwriting team to cut the rate of high-risk shipments by 50% while only flagging 20% of overall shipments.

But from the very beginning, we’d wanted to expand the product of this study beyond our “internal customers” — our underwriters — to help clients of all kinds understand their risk, and benefit from safer practices. So, through collaboration with the Cargo Underwriting team, we turned the findings of the study into a numerical score from 1 to 100.

Get Your Clients' Corvus Cargo Score - and their Enhanced Coverage 

Getting a score is simple. When requesting a quote for Smart Cargo Insurance, we only need the answers to two simple questions: the duration of the journey (in days), and the type of food being shipped. Corvus underwriters will then generate the Corvus Cargo Score and will provide that information along with the quote or declination notice. 

For favorable Scores, Corvus may offer preferred terms for refrigerated cargo, including a reduced waiting period for refrigeration breakdown.

To learn more about Smart Cargo Insurance, click here. 

 

[RELATED POST] Introducing the Corvus Cargo Score

Introducing the Corvus Cargo Score

Marine Cargo insurance is vaunted as the foundation of modern insurance, with its lineage dating to the insuring of ocean-crossing vessels, and their cargoes, in the fifteenth century. With all that history, change tends to come gradually to the line. (And with generally stable profitability, carriers could be forgiven for not rocking the boat.) More recently, technologies such as connected temperature sensors and GPS trackers have entered the picture for shippers. But the data they generate hasn’t been factored into insurance underwriting at scale. Today, we’re sharing an exciting update that will help to finally bridge Cargo Insurance into the 21st century.

[RELATED POST] The Hard Market for Ocean Cargo Won’t Be Helped by Vaccine Roll-Out

The Hard Market for Ocean Cargo Won’t Be Helped by Vaccine Roll-Out

What are the risks involved in a global vaccine rollout?