Skip to main content

See how XM for Customer Experience works

Watch Demo

Using conversation analytics effectively

13 min read
What is conversation analytics, and how can it help you get a deeper understanding of customer sentiment? Learn how to integrate conversation analytics into your feedback channels and use conversation intelligence for a better customer experience.

Author: Adam Bunker

Subject matter expert: Amy Tinley

What is conversation analytics?

Conversation analytics is the process of evaluating customer conversations with your contact center agents. This includes customer interactions such as telephone or chat conversations with your contact center, social conversations, and any text-based two-way dialogue.

The beauty of conversational analytics is that businesses can analyze both sides of a conversation. So that means tracking what customers say, think, and feel, as well as what the organization’s representatives do in response – with automated scoring for things like agent script compliance.

How does it work?

Conversational intelligence is gathered through the use of sophisticated artificial intelligence (AI), machine learning, natural language processing, and algorithms.

These algorithms turn the often tricky nature of human speech into meaning – parsing the hesitations, half-started sentences and expressions – and delivering transcripts with quantitative and qualitative analysis.

This technology is applied to turn the content of phone calls, chats, and more into customer behavior insights. It analyzes these conversations for call reason(s), customer sentiment, effort, and emotional intensity, and finds patterns that can be useful for a deep understanding of customer behavior.

Additionally, analyzing customer conversations can show how well your customer care agents are responding or guiding people through each interaction.

Free guide: Reimagining omnichannel CX in the age of AI

Customer data and feedback types

It’s important to understand the different types of customer feedback and information that brands can gather, since it can be divided into seven intersecting groups.

Customer-centric data can either be operational (O-Data), behavioral (B-Data), or experiential (X-Data). But the information you collect will also either be structured (facts and numbers), or unstructured (qualitative statements and insights). And it’ll either be solicited (asked for) or unsolicited (generated automatically).

Here’s how these seven different qualities manifest as useful information for businesses:

Data elements categorized as behavioral, operational or experience data and grouped as structured/unstructured and solicited/unsolicited.

When brands think about gathering feedback, surveys are often the first tool to come to mind. While surveys can be incredibly useful, they only represent solicited feedback. Structured, solicited data, meanwhile, covers feedback you gather from CSAT, NPS, CES, and more. This is very useful for developing benchmarks and getting top-level information about how your brand is performing, but it’s only one piece of the puzzle.

Conversation analytics can be used for every data type, but its bread and butter is usually in unstructured, unsolicited data. That means scanning, transcribing, and analyzing some of the following customer interaction types:

  • Customer phone calls: The exact wording of your customers’ discussions with your customer service agents and the responses from the service agents to customer questions and concerns. Speech analytics can even go as deep as non-verbal analysis, examining how fast customers speak, how often they interrupt, the tone and volume of their voice, and more.
  • Chatbot conversations: How your customers interact with your brands’ chatbots – and how well your chatbot responds to customer questions
  • Chat conversations: Your customers’ interactions with your human customer service agents and how your service agents interact with your customers, wherever a chatbot was unable to help.
  • Social media discussions: When your customers mention you on social media but aren’t directly talking to you, what are they saying and how does your company (or social influencers) respond to these posts?
  • Other conversations: Consider other two-way conversations that contain useful customer feedback and or agent responses, such as: email, SMS messages, recorded interviews, focus groups recordings, and even GenAI conversation summaries.

Conversational data is incredibly useful for understanding customers’ true sentiments – and the drivers behind them. It covers the aspects of your customer experience that you don’t often think to ask about, but ones that feature in your customers’ minds.

Additionally, conversational data can assist with quality assurance for your agents by understanding their overall performance with regards to empathy, helpfulness, issue resolution, and other behaviors that are important to your organization.

So unless you gather this unstructured, unsolicited data, you might not be getting the whole picture.

Why is it important to analyze customer conversations?

Understanding what customers are saying – and why – is crucial in helping businesses build customer experiences that meet expectations. Your customers are telling you what they think, the onus is on you to listen.

Here are three key reasons why analyzing customer conversations can pay dividends.

1. Analyzing conversation data helps you to get the nuanced story

With over 50% of customers across all ages using their phones to reach out to a customer service contact center, analyzing speech is vital for getting a comprehensive view.

Your solicited, structured data can only give you snapshots into customer behavior, sentiment, effort, and emotions. With conversation analytics, on the other hand, you can identify patterns forming and take action.

Customer support conversation analytics

With advanced conversation analytics you can identify customer emotion, intensity of emotion, effort and sentiment

2. You reduce customer burden and improve the customer experience

The best part of conversation analytics is that there’s no extra burden on the customer. All the information you glean, you gain through conversations the customer themselves has initiated – and you’re not asking them to interrupt their customer journey to give it to you.

3. It adds balance to your customer data

The content of customer conversations can help bring another side of the story to the sometimes unrepresentative information you gather via solicited feedback. With customers talking to and about you across multiple channels, you’ll have a lot more data to play with when you bring conversation analytics into your dataset.

AI in conversation analytics

The sheer volume of conversations happening with and about any given brand means it’s impossible to analyze things manually. As such, conversation analytics tools lean on artificial intelligence (AI) and natural language processing capabilities to interrogate each interaction for meaning, sentiment, effort, and emotional intensity.

The big benefit of AI in this setting is that it can effectively be everywhere at once. AI can ‘sit in’ on every call, read every email, and monitor every channel – and it can bring information found in these traditionally siloed channels together to surface patterns that point to trends and insights.

Best practices in conversation analytics

Have an omnichannel focus

Conversational analytics provide the most information and the juiciest insights when they’re able to draw from as many sources and touchpoints as possible. Opt for a customer experience or conversation analytics tool that can pull customer feedback from every channel, and you’ll end up with a really robust understanding of people’s opinions and feelings.

Learn more about omnichannel CX

Focus on contact resolution

As well as reporting historically on how customer conversations have gone, some conversational analytic software suites, if imbued with the right kind of artificial intelligence, can help human agents in-the-moment by providing timely, relevant prompts that can steer interactions toward a successful resolution.

Track conversation analytics metrics

The data you gather from your analytics tools is pretty useless unless you act on what you learn and understand the effect of those actions. To that end, you’ll want to track metrics that benchmark your performance – and then monitor how they improve over time. KPIs like Average Handle Time (AHT), First Contact Resolution (FCR), and Customer Satisfaction Score (CSAT) should all improve if you’re using what your tools show you.

Conversation analytics: Examples and use cases

The conversation data you gather with this type of analytics can help you gain insights that are actionable, which can help you improve your metrics and propel your business forward.

Below are some use cases for conversational analytics.

Improving products and services

If you’re analyzing what customers are saying – and doing so across every channel – then you’ll be able to piece together a robust story on the successes and failures of your existing products or services.

If, for example, a large number of separate customers all suggest to support agents that your product would be better if only it did X, you’d only be able to turn that into useful, actionable insight if you’re using conversation analytics to unite those otherwise disparate opinions.

Understanding customer motivations

If a customer complains about not being able to log in to an account, is their true dissatisfaction about the failure to log in, or that they couldn’t achieve what they planned to do afterwards? Conversation analytics would be able to tell that, even though a customer called to have a password reset, their end goal was to accomplish much more than that.

Most CRM systems only allow for a single reason for contacting – often in a very large list of drop-down options. With conversational analytics, you can really drill into why they called and why they wanted to log in, and potentially understand the potential lost revenue to the business for each broken customer journey.

Predicting future behavior

Using metrics is useful for understanding the here and now – how customers feel about specific interactions, for example. However, when it comes to predicting future behavior, you might need conversational analytics to drive business decisions.

For example, say you’re planning on launching a new product – and as far as you’re aware, the last product was well-received and conversion rates have been good. However, a user has spotted an uncommon issue with your existing product that you weren’t aware of – and this issue will be part of a big feature in your new product.

Without social listening, you wouldn’t know that this user’s observation has spread like wildfire across social media, putting off potential buyers for your new product. You wouldn’t be able to tackle the problem, because it hasn’t cropped up often enough in your customer service tickets for you to notice – and as far as you’re aware, your existing customers are satisfied.

Instead of finding out the hard way that your marketing and sales team will have a hard time getting leads, you can keep an eye on feedback that’s not given directly to you and make changes before disasters happen.

Improving customer service

Your customer service agents can always benefit from coaching – and conversational analytics can help you train staff to respond more effectively to phone calls.

For example, say your customer service employees are struggling to de-escalate difficult conversations before they become heated. By using conversational analysis, you can find out if there are certain triggers for customer ire, or if there are signs that a customer might not be ready to cool off.

These insights can then be passed along to your employees through training, improving your customer service, and customer experience.

Conversation analytics case study: JetBlue

Example 1: Airline JetBlue’s promotional marketing declared that it offered free baggage – but after doing research using Qualtrics’ solutions with conversation analytics, it found that 82% of their passengers did not consider this offer as a reason to fly with them. As a result, the company rolled out different pricing options which were much better received.

Example 2: Conversational analysis also flagged to JetBlue that a specific airport gate was the subject of negative commentary alongside a lower NPS score. After investigation, the issue turned out to be a simple speaker fault, meaning passengers couldn’t hear announcements. An automated alert was set up to ensure fixes were made swiftly – and the NPS score improved.

Though the individual changes to how JetBlue operates might seem small and distinct, the overall cumulative customer experience results are significant.

Uncover insights with Qualtrics Omnichannel Experience Management

Your employees can only report so much in real time. With sophisticated conversation analytics technology, Qualtrics’ Omnichannel Customer Experience solution let brands dive into every customer interaction – solicited or unsolicited – at once.

Combine conversational data with other metrics and information, bring every channel and touchpoint together, and create an accurate, live picture of how your customers feel and think.

Our next-generation conversation analytics technology enables brands to deliver best-in-class customer support, spot trends that predict pain points as they emerge, and curate memorable customer experiences that drive satisfaction and loyalty.

Free guide: Reimagining omnichannel CX in the age of AI