In this special guest feature, Julie Miller, VP of Product Marketing at Clarabridge, discusses how contact centers leverage AI to remove bias. In her role, Julie leads a team of tech-savvy creatives to clearly articulate the business challenges that Clarabridge solves, explaining how the technology works and showcasing real and measurable value. Prior to Clarabridge, Julie held product marketing roles at companies such as Approva, acquired by Infor, Rosetta Stone and FireEye. With more than 20 years of experience, Julie is passionate about educating buyers to make informed purchase decisions.
Every business must have checks and balances on its contact center performance. After all, how your customers perceive your contact center experience is how they perceive your business. According to PwC, 70% of consumers value speed, convenience, helpful employees and friendly service. Whether it’s a customer seeking product information or resolving an issue, the contact center has the power to make or break the customer experience.
Quality management teams are the checks and balances for contact centers. They improve CX by tracking agent calls, emails and chats against a checklist of scoring criteria in a spreadsheet. However, by harnessing natural language understanding (NLU) to automatically evaluate contact center interactions, teams can better identify issues across large volumes of inquiries and act confidently on them while remaining relatively free of human bias and error.
NLP vs. NLU: What you need to know
A more common method used by businesses for linguistic analysis has been Natural Language Processing. An NLP model applies linguistic and statistical algorithms to text in order to extract meaning in a way similar to how the human brain understands language.
This system can analyze data and reach results at impressive speeds. For example, about 95% of customer data exists in the form of unstructured text – in emails, survey write-in answers, Twitter posts, online reviews, comments in forums and more. Reading through all of this text is next to impossible: assuming the average person can process 50 items of unstructured data in an hour, it would take nearly seven years for one person to read through one million items. An NLP model can parse that much information in minutes.
After reading through heavy datasets, NLP categorizes data into topics tied to a few keywords and phrases. This transformation allows humans to review for patterns in the data. What was once a tedious and subjective human process is now easy to digest and useful for actionable insights.
While NLP offers a powerful resource for businesses, over time, it has been watered down, especially so in the customer experience space. CX teams leverage NLP to build lists of words and topics. But, simply grouping data is no longer enough to provide the useful insights required for CX improvement.
Fortunately, technology has been keeping pace with business needs. Natural Language Understanding (NLU) takes NLP a step further and analyzes what language means, rather than only what individual words say. This area of research and development relies on foundational elements from NLP systems, which map out linguistic elements and structures, but then adds context. Instead of focusing on the words themselves, NLU seeks to intuit the connotations and implications innate in human connection, analyzing the emotion, effort, intent or goal behind a speaker’s statement to uncover their meaning.
Ultimately, to deliver on NLU, a system must have mature NLP capabilities. Developing an NLP engine establishes the linguistic foundation you need. You can then build value-added features to incorporate insights from context and meaning using NLU.
How NLU Can Help Eliminate Bias
Quality management teams need data to analyze how a contact center performs and what processes could be improved. Many teams still rely solely on Net Promoter Scores, which are difficult to scale and often lead to biased results based on subjective responses. Companies should have easy access to insights that help improve quality management, agent response and overall experience without bias. But, with 80% of contact center data being unstructured, quality management teams face a significant challenge to do so.
NLU offers a way for quality management teams to develop useful insights at scale. By connecting NLU to every type of customer interaction in a contact center — calls, chats, messaging and emails — teams can define their own weighted evaluation criteria and automatically score interactions to assess agents based on hard and soft skills. Businesses can identify issues across large volumes of inquiries, determine the best order to address them, and confidently act on objective information. Intelligent scoring driven by NLU provides a consistent and transparent model while remaining relatively free of human bias.
Quality management powered by NLU becomes an automated and efficient process. Freed from performing manual quality assurance tasks, QA managers can dedicate their time to other initiatives better suited to human minds, such as coaching to increase overall efficiency and expertise while providing a balanced, objective measurement for improvement.
Leveraging NLU to Improve CX
To increase efficiency, NLU must overcome the challenges posed by the human language itself. Language is fluid, complex and full of subtleties. For example, two people may read or listen to the same passage and walk away with completely different interpretations. If humans struggle to develop perfectly aligned understanding due to these congenital linguistic challenges, it stands to reason machines will struggle as well.
To help with these challenges, NLU uses rules-based and machine learning techniques to extract, tag and score concepts relevant to customer experience analysis such as emotion, effort, intent, profanity and more. Users can customize many of these elements to reflect their business, use case and industry. When combined with the original text and associated source and customer metadata, analysts and front line teams can uncover what customers mean, not just what they say, empowering truly actionable insights.
Companies can analyze customer experience feedback data using numerous factors, opening the doors not only to improvements in quality management but also to new kinds of business questions and answers. For example, if customers praise or criticize an associate, the system can detect that information and assist CX teams in rewarding or modifying performance internally. Or, by identifying subtle cues like phone numbers or email addresses in text, the system can encourage associates to proactively contact a customer and leave a positive impression. NLU can even capture the names of performers, politicians, business executives and other celebrities who may be external influences on your customers’ perceptions.
NLU can also identify other trends influencing your customers. For example, it can track mentions of events in conjunction with discussions of sales and promotions, such as Independence Day, Black Friday or Cyber Monday, to determine which ones are generating buzz. Mentions of weddings, engagements, baby showers, graduations and so on may help highlight how best to market and price items to target specific buyers celebrating certain milestones. Or, discover how customers use your products by analyzing them in conjunction with mentions of cultural events or other occasions.
Contact centers present a treasure trove of useful information for your business, but you need the right tools to unearth it. From evaluating agents objectively and transparently to discovering the best times to market certain products, NLU has the power to help ensure your quality management teams capture and deploy the best insights to improve your customer experience.
Sign up for the free insideAI News newsletter.
Join us on Twitter: @InsideBigData1 – https://twitter.com/InsideBigData1
Julie, what NLP tools do you use or recommend for finding data in documents? I am looking at several https://www.bisok.com/grooper-data-capture-method-features/natural-language-processing this is one) and am trying to get as much expert feedback as I can. I need to find and extract data from many 80-page contracts every month. Thank you