Today’s Post By: Chris Leach, Director of Perficient
Online forms, paperless processing, mobile applications, digital transformation – all part of the wave of the future. Still, today’s property and casualty insurers remain inundated by paper. One large national insurer recently upgraded their entire ECM infrastructure from FileNet Image Services to FileNet Content Manager, including the replacement of number of ingestion channels with Datacap.
To help quantify “large”, this particular organization ingests between 400,000 and 500,000 documents per day. These documents come from numerous sources. Scanned and faxed documents represent about ½ of the volume, with content from electronic feeds making up the balance.
The nature of the operations is such that there isn’t the need – nor the business value – to try to capture field-by-field data. For the most part, this kind of data (accident date, VIN, policy number, insured’s name, etc.) is already entered online, usually at time of First Notice of Loss. With the subsequent incoming content, all that is generally needed is a claim or policy number to match the inbound document to the correct folder – and to initiate a workflow process.
This matching process, however, can be tedious. Inbound faxes and scanned documents are unstructured, and claim/policy information can appear anywhere on the document. This presents two problems. First, documents need to be manually indexed. And second, because the content on the documents is generally free-form, the needed information – policy or claim number – can appear anywhere on the document. Index users must scan through the content looking for the right information, and then key that information. The need to search through the document and locate the information makes the indexing process significantly slower.
Enter Datacap.
This insurance customer opted to take advantage of Datacap’s full-text OCR capability on each inbound scanned and faxed document. This generates text output for each clearly discernable word on the page. Datacap then searches through those words to locate a claim number.
In this scenario, the claim number will (a) follow a specific numeric pattern; and (b) contain a check digit. By using Datacap to automatically search for the claim or policy number anywhere in the document and then applying specific business rules to validate the claim number, the time required to index this huge volume of documents is significantly reduced.
This insurance customer was able to optimize its Datacap-based OCR solution with two simple steps:
- Upfront analysis: First, by analyzing fax and scanned documents at the outset of the project, it was determined that if a policy or claim number appears on the document, it does so somewhere on the first two pages about 99% of the time. By limiting the full text OCR to the first two pages of each document (rather than searching every page, where the average document size is 8 pages), this increases the OCR throughput by a factor of 4, with very negligible impact on the number of documents that can be automatically indexed.
- Validating and augmenting the data: Second, by leveraging Datacap’s ability to invoke a web service to both validate the claim number and to obtain other necessary index fields, this insurer is able to minimize the amount of manual keying required.
With this combined approach, this insurer is able to automatically index over 75% of inbound documents with no user intervention – and the success rate climbs to over 90% of those documents with a legible, valid claim number! Overall, this allows complete straight-through indexing of a large volume of inbound, totally unstructured documents. This reduces the labor involved in indexing and increases the speed with which these time-sensitive document reach the claim and policy processors.
To find out more about Datacap, please visit: ibm.com/datacap
IBM ECM 