Data-driven Voice of Customer analytics is proven to increase lifecycle value and reduce churn by delivering the insights companies need to dramatically improve brand and product experiences. There are six steps involved in building an effective VoC analytics program:
- Identify a question
- Gather and prepare data
- Choose your tools
- Analyze and troubleshoot
- Draw conclusions
- Take action
In order, this article will:
- Outline the business value of good customer experiences
- Demonstrate how to process customer feedback and reviews using text analytics and natural language processing
- Explain how to use these insights to improve experiences and revenue
- Show the value of building a multi-channel Voice of Customer analytics program
- Suggest further reading and resources
What is Voice of Customer?
In simplest terms, Voice of Customer (VoC) is…
“…the customer’s voice, expectations, preferences, comments, of a product or service in discussion. It is the statement made by the customer on a particular product or service.” – SixSigma Institute
Therefore, a Voice of Customer program is a structured system of feedback collection, data analysis, and action planning.
The main goal of any VoC program is to gain better understanding of how people perceive and interact with brands, products, and services.
Good customer experiences create value
In the age of social media and viral reviews, consumers know they have the power. In fact, 89% of businesses will compete mainly on customer experience, according to Gartner research.
The opportunity is huge. For example, research by Bain & Co. has found:
- Companies that excel at customer experience grow revenues 4-8% above their competition
- Superior experiences deliver 6-14x higher customer lifecycle value
- Voice of customer programs result in up to 55% greater client retention
Certainly, there is huge value to be gained from improving customer experiences. And as we’ll see in this article, building a data-driven Voice of Customer analytics program is key to making informed, effective changes.
But building an effective VoC analytics program is not a simple task. To truly listen to your customers, you must be able to understand the text they’re exchanging with you and among themselves.
And before you start analyzing anything, you need to understand the challenge you face.
Understand the challenge of big data
Consumers send 350,000 tweets per minute and post nearly 5 billion Facebook comments every month. Meanwhile, Yelp and TripAdvisor together host more than 750 million consumer reviews. Combined with customer surveys and other sources of feedback, this content forms wealth of information for brands to mine.
But all that glitters is not gold. Consider the volume of data out there. The scale of it goes way beyond any human’s ability to analyze.
Think of it this way: How long would it take you to read just 3,000 reviews? 20,000 tweets? 8,000 Facebook comments?
And reading is just the beginning. Could you draw a connection between the themes expressed in Facebook comment #1,904 and TripAdvisor review #2,361? Could you map sentiment versus net promoter scores by quarter across 150 themes, topics and categories?
This is where text analytics software can expand your abilities and empower you to build an effective VoC analytics program.
Related article: The Crucial Piece that NPS Misses, and How to Fill the Gap
Text analytics tools, such as the Lexalytics Intelligence Platform, are the key to unlocking the value of this data.
These platforms serve as foundations upon which customer experience professionals, data analysts, and software engineers can work together to build an effective VoC analytics program.
How to build an effective VoC analytics program
To generate real value, you must be certain that your Voice of Customer analytics program is producing accurate, reliable insights. And to produce those accurate, reliable insights, you must follow the data analytics process.
1. Start with the question
An effective VoC analytics program is focused on answering questions.
So, before you analyze anything, clearly identify the question (or questions) you want to answer.
For example, a customer experience professional might ask:
- Why were North American sales of our moisturizer down last month?
- What changes should we make to our hotel rooms next quarter?
- How can we improve patient experiences with our drugs and therapies?
The questions you ask will inform:
- The data you gather
- The analytics tools you choose
- The types of analyses you perform
Of course, you may not know the right question to ask. That’s okay! Try starting with something very broad, such as:
“This location seems to be underperforming. Can we see the comments from customers who went there?”
If you’re still not sure, take a step back. Think about your business goals, and what insights would help you make better decisions. Ask yourself: What information would help us improve our products and services?
2. Gather and prepare data
Got your question in mind? Now it’s time to gather data.
Above all, the data you collect should be suited to your question.
For example, a brand-focused question might demand a mountain of tweets that mention your company.
Alternatively, a product-orientated question could lead you to collect a bundle of customer satisfaction survey responses.
Some common text data sources for Voice of Customer analytics programs are:
- Online reviews
- Survey responses
- Support tickets
- Facebook comments
- Chat conversations
- Call transcripts
Still not sure where to get the data you need? We can’t source it for you, but we may be able to point you in the right direction. Drop us a line.
3. Choose the right tools
If you need to compare complex trends over time, a barebones survey analysis tool won’t be able to tell you enough.
Likewise, if your goal is to use insights drawn from social media comments as proxy for customer survey responses, you need a flexible platform that offers rich reporting and customizability.
To get the best return on your investment, choose a solution provider who has deep experience solving problems similar to yours.
Here’s some advice from our own CEO:
“Based on these observations and my own experience, I’ve identified four key questions every enterprise needs to ask before they choose an AI partner:
- What’s the ROI going to be?
- How experienced is this company?
- Can they tell you about their own systems?
- What’s the real cost to me?”
Jeff Catlin – CEO, Lexalytics
4. Analyze your data
In data analytics for Voice of Customer, the reports you generate should be focused on answering your question.
Let’s say you’re asking, “How do travelers feel about Atlanta’s airport?”
To find an answer, you’ve gathered a collection of reviews from Hartsfield–Jackson Atlanta International Airport’s Facebook page. And for the sake of argument, let’s say you’ve determined that Lexalytics’ Semantria Storage and Visualization web platform is the best tool for the job. (This isn’t just self-promotion: we’ve actually run this analysis ourselves).
In technical terms, you’ll be most interested in reporting on the themes, entities and topics being discussed in the reviews, and the sentiment expressed towards each. Other questions might merit different reports, such as categorization or intention extraction.
Here’s one example of an analysis you might perform. This chart shows sentiment scores mapped against volume over time for reviews that mention the topic “Wayfinding”.
So, if you’re struggling to get started, either consult with your solution provider’s support team, or go back to step 3.
5. Draw conclusions
Some insights will be self-evident. For example, in the chart above, it’s clear that “wayfinding” is a significant pain point for many travelers.
Meanwhile, this sentiment cloud, extracted from the same data set, clearly indicates negative opinions around the entity of “TSA”. No surprise there!
Other revelations may surprise you. And sometimes, you’ll find answers to questions you hadn’t thought to ask.
Case in point: a hospitality company, while trying to decide which pieces of furniture to upgrade, uncovered a large number of reviews referring to “trash” and “smell” with strongly negative sentiment.
Digging deeper, they discovered that many guests could smell the dumpsters in the parking lot as they entered the hotel!
Unexpected insights like this, drawn from real customer reviews, are a great demonstration of how a data-driven Voice of Customer analytics program can stand out from the pack.
6. Take action to improve customer experiences
Sometimes, your best course of action will be clear.
In the case of the hospitality company whose guests complained of trash smells, management should just move the dumpsters.
Or, take this Facebook comment from our analysis of Atlanta International Airport reviews:
“It is a bit confusing. The parking is horrific! Not enough signs for direction and the plane train was definitely anxiety because I didn’t know where the heck it was taking me.”
In this case, it’s evident that just adding more signs and directions would go a long way towards encouraging travelers to return to ATL.
Of course, some conclusions will be more complicated. You may need to drill down into specific data points or run more analyses.
Consider this sample collection of text analytics reports, using traveler reviews from the San Francisco Airport.
In this example, we’ve written up our interpretation of the results in the Summary section. One conclusion is that there is a problem with flight scheduling, and this problem has been increasing over time.
Now, what’s the best course of action here? It’s hard to say. Flight delays may be caused by hundreds of factors.
However, by collecting, combining, and comparing reports into dashboards like this one, you can uncover the insights you need to start moving down the path to better customer experiences.
Achieve more success with multi-channel Voice of Customer analytics
Analyzing customer reviews and social posts to address individual pain points at speed is a great way to create immediate, tangible value from text analytics (see this KPMG success story).
Ultimately, however, your VoC analytics program should be used to drive decisions across departments. Gartner puts it best:
“VoC data and insights inform many diverse marketing aspects across customer experience, brand, competitive analysis and product development.”
But don’t just take their word for it. According to research by Aberdeen Group, the key to building a best-in-class VoC program is operationalizing feedback and sentiment data across multiple channels.
McorpCX, a customer experience services company, argues that data analytics software is best used as a foundation to build larger processes.
In fact, they describe VoC analytics as “a great tool to underpin the data gathering and distribution process – especially if you’re collecting and analyzing data from multiple sources and in various formats.”
For example, scroll back up to the dashboard of traveler reviews from the San Francisco Airport. Remember how we noticed a problem with flight scheduling, and that this problem has been increasing over time?
Related article: The Crucial Piece that NPS Misses, and How to Fill the Gap
This can’t be solved by one department alone. But the information contained in these repots can easily be exported and presented to department heads across the company. Together, using these insights, SFO can start working together to solve the underlying issues.
In short, VoC data analytics should be used to drive decisions in every department: customer success, product management, operations, marketing, and sales. Ultimately, your voice of customer program will become successful by fitting itself into a larger context.
Troubleshooting your Voice of Customer analytics
One last note before we wrap up. If you’re struggling to get any useful information from your reports, take a step back. A messy, unintelligible output may be caused by a badly-configured analysis, but could also indicate a corrupted data set or system failure.
- Starting with the wrong question.
- Using messy, unprocessed data.
- Not configuring the analysis properly.
- Gathering the wrong data for the question.
- Drawing the wrong conclusions.
So, go back to your original question and trace your steps.
- Are you sure you gathered the right data to answer your question?
- Has the data been properly prepared for analysis?
- Have you consulted all available documentation, tutorials, and guides?
Remember, no one knows your data analytics tool better than the people who built it. Therefore, don’t be afraid to ask your solution provider’s support team for guidance or advice. Requesting help early and often will save you time, money, and headaches.
Success stories from our partners
As we’ve shown, text analytics is a key component of a successful Voice of Customer analytics program. By analyzing surveys, comments, and other text documents, customer experience professionals can gain valuable insights into how customers feel, and why they feel that way.
Here are a few real-world examples of how our partners use text analytics and natural language processing to help their clients build data-driven Voice of Customer programs, or to do more with their own data.
- VOZIQ helps contact and call centers reduce customer churn 2x faster and increase retention by 10%. – Read more
- AlternativesPharma delivers insights to pharmaceutical marketing teams to help them grow market share and demonstrate value. – Read more
- Microsoft tracks 1,000+ products and brands while generating actionable insights 3 weeks ahead of traditional survey responses. – Read more
- KPMG Nunwood identifies and understands customer pain points while powering their rapid-response voice of customer solution. – Read more
Further reading and resources
Here are some great resources for anyone looking to build or improve a data-driven Voice of Customer program:
- Gartner, “Use Voice of Customer Data to Improve Customer Experience Analytics”
- Six Sigma, “Capture Voice of Customer”
- CMO.com, “The Real Value In Voice Of The Customer: The Customer Experience”
- Aberdeen Group, “The Business Value of Building a Best-In-Class VoC Program”
- McorpCX, “Designing a Voice-of-the-Customer (VoC) Program: Beyond Customer Listening to Customer Understanding”
Or, to learn more about text analytics and natural language processing:
- Lexalytics resources library – Natural language processing and text analytics white papers, case studies, and solution profiles
- Lexablog (you’re already here) – This is where we post technology insights, results from our own text analysis projects, and more
Lastly, if you’d like to see our own natural language processing and text analytics solutions in action: