While banks have long had the ability to transcribe call center conversations into text, a problem remains: what to do with them? The transcripts amount to terabytes of data in an unstructured format, putting them out of reach for most analytics engines.
That leaves banks using armies of employees to read through the transcripts to find actionable nuggets that can result in business gains, an expensive proposition that’s not scalable.
An intelligent document processing platform offers an alternative. It enables banks to build models that act like an expert listening in on those call center conversations, zeroing in on the pertinent bits that indicate a caller is ripe for additional services like financial planning, or unhappy and in danger of jumping ship.
An AI-based model can read through a call center transcript in a tiny fraction of the time it would take a bank employee to do it, opening up the transcripts to analytics that can uncover valuable business opportunities. Let’s look at a few use cases that demonstrate how intelligent automation can help banks find gold in TBs of transcripts.
Intelligent automation for sentiment analysis
Banks know how important it is to retain existing customers, because it’s so much more expensive to find new ones. And at a time when surveys shows nearly 50% of bank customers would consider switching to a digital bank merely on the strength of positive reviews, it’s imperative that banks do everything they can to keep customers happy.
Those call center transcripts can be a treasure trove when it comes to identifying customers who may be on the brink of making a move. Intelligent document processing models can be trained to look for language that indicates an unhappy customer, whether it’s the straightforward “I’m not happy” or the more subtle “I thought you said.” Depending on how many of the offending words occur in a conversation, an intelligent automation model can score it in terms of how likely it is that it fits the pattern of a customer who has one foot out the door.
The ability to quickly find instances of unhappy customers and have an agent follow up with an offer to repair the damage can mean an uptick in customer retention rates.
Automating call center transcript classification
Banks may also want to classify various call center conversations, to separate the “angry” calls from the more positive ones, such as those seeking information on products and services.
An intelligent document processing model can classify each call, then translate it into a structured format for further analysis. After putting all the angry calls together, for example, a bank can analyze them and determine which products or services are causing the most problems, and why. Here again, that’s valuable data that can then be used to address recurring issues.
Transcript search and comparison automation
Or perhaps a bank already knows it has issues in its mortgage lending business. Banks using intelligent automation can automate the process of searching through call center transcripts to find those that relate to customers who are unhappy about a mortgage issue. Analysts could then compare the calls to one another to further discern what the most common issues are.
Perhaps a common theme is problems related to variable interest rates that go up over time as inflation rises. The bank could then perform some outreach to explain the situation and any options that may be available.
The key is to employ an intelligent data processing platform that’s flexible enough to take on whatever use case you may have in mind. The Indico Unstructured Data Platform makes it a simple matter for your process experts to build models that shave untold hours off the process of reading call center transcripts, opening up those TBs of data to analytics that can uncover valuable business insights.
To see for yourself how the Indico Data platform can help you deal with mountains of call center transcripts, check out this case study from a Top 10 Bank or schedule an in-depth demo. Soon you’ll be putting all that data to good use.
