Why extraction is not enough to digitise risk submissions

by Juan de Castro, CCO/COO, Cytora

Extraction is not enough to digitise risk submissions

Digitising broker submissions is a crucial step to unlock digitisation for insurance companies. To make risks decision-ready across the customer journey, a digital profile of the risk needs to be created from data extracted from the submission as well as captured from multiple third party data sources.

Only when combining data sources at the point of digitisation, insurers can achieve an uplift in automation

To provide the level of certainty required to automate decision making downstream, submission extraction on its own is not enough. You cannot drive automation on data extracted unless you have a high level of trust in the accuracy of the obtained values.

Third party data provides a way to boost and validate extraction accuracy. Data fields that are extracted from the broker submission can be validated against equivalent data fields from other data sources to ensure the correct entity has been resolved (company and address) and the correct risk characteristics have been captured. 

End-to-end submission digitisation platforms like Cytora cross reference and verify extracted risk data to meet the extraction accuracy levels required to automate decision-making. For example, comparing the turnover value extracted from the submission against the turnover value in an external data source to check if they are consistent. Or for property owners submissions, comparing the building height value extracted from the submission to the building height value from a third party data source. By combining data sources at the point of digitisation, insurers can achieve an uplift in automation by increasing their confidence in the data used for decision making. If the process identifies any discrepancies, the submission is routed to an exception management workflow to be reviewed by a human.

Commercial insurers may also decide to use internal data sources to identify fields that are anomalous relative to other similar risks in the same segment. This can drive ever greater granularity into directing risks to straight-through-processing versus referral workflows. It enables insurers to become more nuanced over time in how they evaluate the accuracy of data fields. 

The point at which data augmentation is performed significantly impacts insurers’ ability to drive higher rates of automation 

Submission data extraction, company and address resolution, and third party data augmentation are three intertwined and inseparable capabilities. 

Often extraction produces a number of candidates that need to be resolved and checked against third party data to ensure the right one has been selected. The extraction confidence score alone is a pure reflection of the technical extraction process, which lacks commercial insurance context. For example, the extraction process might result in several insured address candidates; sufficient extraction confidence on the address field can only be achieved when the insured name is resolved and we confirm that the address extracted from the broker submission matches the one stored in a third party company registry.    

Additionally, using third party data enables insurers to close the gap between data included in the submission and what the insurer requires to make decisions on the risk downstream. For instance, in the new business workflow, the pre-quote decisions of evaluating fit with risk appetite, prioritisation, routing, and technical rating are usually dependent on both submission and third party data. 

These two challenges above (achieving the right level of extraction confidence and closing submission data gaps) represent the two main obstacles to achieving higher rates of automation. These can be overcome by combining extraction, resolution, and augmentation upstream in the new business submission digitisation process. 

If augmentation is done further downstream (e.g. at the underwriters’ workbench), insurers will see a larger percentage of their submissions pushed to exception handling and create unnecessary manual work which absorbs underwriting capacity and drives higher turnaround times.  

This is also the case for optimising the effectiveness of the operating model by augmenting the risk before it reaches downstream systems. Once the submission has been enriched, insurers can make decisions about where the risk should go, that has consequences for which systems receive what subset of data fields. For example, is the risk submission inside risk appetite and, if so, should it go to an underwriter or be straight through processed? The earlier the insurers can make decisions, the more streamlined, synchronised and accurate their downstream multi-system workflows will be.

The key is in the combination 

Ultimately turning analogue risks into decision-ready risks requires a combination of data from the submission, external and internal sources. Combining these data sources at the point of digitisation enables the highest levels of automation and accuracy which continuously increase over time.

Insurers who are able to digitise the risk upstream making it decision-ready for the range of decisions that are made downstream will be able to deploy the most powerful and streamlined workflows and unlock scalable profitable growth.