For the past 20 years, I’ve been a qualitative researcher… perfectly content with interpreting meaning from the dozens of interviews, usability studies, or diary studies that I had just conducted.
Until a year ago, I never would’ve dreamed that I would fully embrace text analytics or that it would become an integral part of the way I work every day.
I love text analytics because it enables me to extract valuable insights from billions of data points, such as survey responses, phone calls, and chat logs. Additionally, I now sit significantly closer to the customer experience—observing first-hand how customers interact with customer service representatives, purchase products, and solve their problems.
If you’re someone who has had a fuzzy understanding of text analytics or has maybe heard conflicting information… this post is just for you!
Text analytics not only identifies sentiment, effort, and emotions—but also employee behaviors such as actively listening, asking open-ended questions, and demonstrating empathy. It helps me immediately uncover tech issues, usability issues, competitive intelligence, etc.
“My qualitative research analysis has transformed from analyzing dozens of conversations to millions—using the exact same level of effort and timeframes.”Kristine Remer (that’s me!)
Here are a few myths I’ve heard about text analytics and my point of view.
Myth #1: Text analytics is just a buzzword or a synonym for NLP or AI
First, what is text analytics?
Text analytics is the best thing that has ever happened to the field of consumer insights. Think about every sentence your customers speak into the phone to your call center agents, everything they write in social media about your company, or every detailed explanation as to why they gave a high (or low) NPS rating.
Text analytics makes it possible to analyze all those billions of sentences—and make sense of them and in real-time by tagging each of them with:
- sentiment (positive, negative, neutral)
- effort (hard, easy, neutral)
- emotion (confused, stressed, angry, delighted, etc.)
- employee behavior(s)
Text analytics, NLP, and AI are closely related, but they’re not interchangeable. Artificial intelligence (AI) is the broadest concept. Natural language processing (NLP) is a sub-set of AI. Text analytics is just one application of NLP. Other NLP uses include speech recognition, translation, and chatbots.
For example, Duolingo uses NLP techniques to teach language to its users in a natural way, including features such as speech recognition and translation. NLP is used to analyze and understand the input from users, such as their pronunciation and grammar, and then provide feedback to help them improve their language skills.
Myth #2: Text analytics is only useful for social media monitoring
Text analytics can 100% be used to montior social media, but it’s just the tip of the iceberg. Here’s a very short list if use cases:
- Voice of customer analysis
- Fraud detection
- Customer segmentation
- Agent QA and coaching
- Customer retention
- Marketing & product development
- Plagiarism detection
- Stock market predictions
Myth #3: Text analytics requires a lot of IT expertise
I’m not an “IT expert,” but I have still been able to gain enormous value from text analytics because of my qualitative research coding and analysis knowledge.
In a little over a year, I’ve learned a lot about text analysis, including learning how to tune transcriptions and sentiment, build comprehensive topic models completely from scratch, and not be overwhelmed by the idea of uncovering insights from literally billions of unstructured data.
Myth #4: Text analytics is too expensive
Text analytics platforms can be extremely expensive, but there are affordable solutions available that companies can use to get their feet wet (and then quickly get excited by the possibilities!). Some cloud-based options do not require significant upfront investment.
Myth #5: Text analytics only provides descriptive analytics
There are three main flavors of text analytics: descriptive, prescriptive, and predictive.
- Descriptive analytics: analyzes historical data to understand the past
- Predictive analytics: uses statistical models or machine learning to predict the future
- Prescriptive analytics: recommends the best course of action based on the outcomes predicted by predictive analytics
Text analytics doesn’t just count how many times a particular phrase or keyword appears in a set of data. It also can be used to conduct sentiment analysis, customer effort analysis, and topic modeling.
In my own CX practice, I use text analytics to identify the root causes of customer issues—without ever talking to them.
If you’d like to hear more about my text analytics learning journey, I’d love to connect!
- Discovering Hidden Insights with Kano Analysis: A Case Study
- ESAT & CSAT: A Survey Alignment Love Story
- UX Research Toolbox: Anatomy of a Compelling Narrative for Your UX Research Reports
- Data Overload? How to Manage the Flow of CX Insights Effectively
- Bias-Free UX Research: Innovative Approaches to Mitigating Bias in the Research Process