Data mining and analytics speed the client onboarding process.

KPMG team

KPMG team, left to right: Eric Goldberg, director; Justin Giuliano, senior associate; Binoy Palakkal, director; Michael Henry, principal; Taylor Peterson, intern; James Sawyers, intern; Bryan Callahan, director

Credit: KPMG

Large banks, insurance companies and hedge funds are all required to gather reams of data on their clients to determine their risk and to meet regulatory requirements — a process known as client onboarding. The problem was, onboarding typically required humans to read and analyze sometimes hundreds of thousands of pages of documents, with little room for error.

Professional services firm KPMG was looking for a way to automate the costly onboarding process for its clients. Reports needed to capture information from SEC filings, blog entries, social media, text messages and other sources of structured and unstructured data.

KPMG chose MarkLogic's Enterprise NoSQL database to integrate, organize and mine customer data using semantics, text analytics and visualization. "We could take out 70% to 80% of the human process, and because it's done by machine, it's much more consistent," says Michael Henry, general manager of global client onboarding at KPMG.

Today, a commercial bank that once spent 27 hours onboarding a medium-risk client can now automate the process and reduce the time to three and a half hours. Tax classifications that would take a person 90 minutes to read and process can now be completed in about 4 minutes, Henry says.

As human involvement in the process declines, so too will the number of errors in reports, he predicts. "In 10 years, nobody will be doing this manually," he says.

This story, "KPMG" was originally published by Computerworld.