CONTACT CENTER EXPERTISE X ARTIFICIAL INTELLIGENCE

Inefficient hand-offs and occasional miscommunication between the front-office and the back-office can cause inordinate delays in closing customer requests, leaving customers dissatisfied and angry about the delay. This was the challenge faced by EXL’s client, one of the world’s largest life and annuities companies offering its services across 30+ nations.

EXL had been managing back office operations for the firm for several years. On any given day, several policy holders of the client reach out to their contact centers for various kinds of service requests. The front office staff at the contact centers types out these requests in free-form text during the conversation, then passes them on to the back-office for servicing. The back-office was then responsible for pulling up the relevant records, noting down or updating information as needed, and sending out customized correspondence to the customers.

However, at times errors or omissions would be made by the front office staff while noting down the requests. This would then require the back-office to revert for clarifications. The back-and-forth would delay completion of the service request, thereby decreasing customer satisfaction. Further, because the back-office was processing requests manually, the overall servicing timelines were quite stretched.
With customers used to the fast pace of today’s digital world, they no longer accepted old-world inefficiencies in customer service. So, the client reached out to EXL to implement cutting-edge technologies that would speed up back-office timelines, providing more leeway for the occasional hand-off issues between the front- and back-offices.

EXL approached this challenge with Digital Intelligence, its strategy towards digital transformation. First, the context of the problem is understood by drawing on the company’s deep domain expertise in combination with process related data. Then, the right mix of digital technologies and professionals are precisely orchestrated to deliver impactful outcomes.

"Inefficient hand-offs and occasional miscommunication between the front-office and the back-office can cause inordinate delays in closing customer requests, leaving customers dissatisfied and angry about the delay."

Context

With several years of experience in managing the client’s back-office, EXL already had in-depth knowledge of the client’s back-office processes. In these processes, the team would manually examine the free-form text sent by the front-office to determine the type of service the customer was requesting and select the communication template to be used. The templates had predetermined fields to be populated with the relevant details pertaining to the request. The team would fill in these fields by referring to the client’s records database, perform other necessary actions such as updating records or processing payments, and then send out the response letter to the customer. All of this required browsing through hundreds of screens to find the handful of fields required to be used in the communication.

To speed up this manual and time-consuming process, EXL set out to build a natural language processing (NLP) engine that would automatically read and understand free-form text sent by the frontoffice staff. EXL would also deploy robots to automatically carry out the relevant processing commands generated by the NLP engine.

Orchestration

To automatically process free-form service request texts, the NLP engine requires a keyword dictionary of frequently occurring words and phrases. EXL deployed a team of business analysts to spend six weeks with the process associates on the operations floor to observe and document their actions. The outcome of this exercise was a comprehensive keyword dictionary that linked key terms and phrases to the process steps needed.

While the dictionary development was underway, three developers built an NLP engine which could automatically parse free-form text using the keyword dictionary and generate the necessary process-related commands. In parallel, other developers worked on configuring four robots that would be able to automatically perform the necessary steps to service requests basis the commands from the NLP engine.

Further, the solution also incorporated an analytics algorithm that studies historical patterns in customer service requests, and predicts the probabilities of when the next such similar requests are likely to come in. For example, tax gain quote statement requests from some customers of a particular American state are likely to come in at a particular time of year based on that state’s regulations. Thus, the digital bots can proactively send out such statements, even before customers request them means a proactive communication rather than reactive.

The solution also takes care of frequently changing federal regulations which require updating customer correspondence templates. The earlier manual process sometimes mistakenly sent out correspondence based on outdated templates. Customers would then need to get in touch with the contact center asking for the missing information required by new regulations. With automation, the bots are now responsible for version control of the templates. When a federal regulation change notification is sent by the client, the bots mark any related requests as exceptions, which are not serviced until the relevant templates have been updated.

After the solution was rolled out, the EXL developers are working on further refinements. A ‘spy bot’ is being built that will observe the occasional requests that the NLP engine cannot deal with and flags as exceptions. This bot, using assisted machine learning technology, will observe the steps taken by human agents while servicing such exceptions and automatically update the keyword dictionary. Thus, similar requests in future can be serviced automatically without human intervention.

Supporting these teams in the background, the EXL governance team interfaced with the client’s IT department, to overcome policy challenges relating to systems interfacing and deployment of the technology in the back-office.

Outcomes

With the NLP engine - which has a 96% accuracy rate - driving the digital bots, back-office operations are cruising in overdrive. Over 40% of the requests coming in through the contact center and through emails are completely automated. With absolutely no human touch points for these requests, there is a 15-20% reduction in processing errors. This has reduced the need for customers to repeatedly reach out to the contact center and has also significantly improved the average servicing turn-around time.

All of these factors as well as the bots’ proactive prediction and servicing of requests have contributed towards a greatly improved Net Promoter Score, our client’s metric for measuring customer satisfaction. As a bonus, the 30-40% increase in in efficiency delivered by the bots working around-the-clock has contributed $200,000 to the bottom-line through the elimination of 4.5 FTEs. Cutting-edge digital technology can not only delight the customer, but also deliver powerful business outcomes.

Solution Summary

Client Challenge

  • Inefficient hand-offs between the front- and back-office
  • Stretched servicing timelines because of manual processes
  • Customer dissatisfaction due to delays in closing service requests

Context

  • EXL had in-depth knowledge of client processes 
  • EXL solutioning included the components of an NLP engine and a digital workforce

Orchestration

  • Business analysts built a keyword dictionary for the NLP engine through observing manual processes
  • NLP engine built using Microsoft.NET
  • Digital workforce deployed through configuring four bots
  • Governance team overcame challenges related to IT policies and systems

Outcomes

  • Over 40% of back-office processing completely automated
  • 15-20% reduction in processing errors 
  • Greatly improved customer satisfaction scores
  • 30-40% gains in efficiency
  • Savings of $200,000 through elimination of 4.5 FTEs
     

Contact US