Are you a Zendesk user looking to improve your support center? In this series, I’m sharing some of the insights that we’ve heard from our conversations with CX professionals using Zendesk.

Recently, we’ve seen a lot of interest in our integration. Keatext is a great add-on to what is already a great support center. Zendesk is one of our most successful integrations, and that’s because we’ve been able to directly address some of the pain points that people have come to us about.

The first one I want to cover in this series is the need to categorize tickets based on the content of what the customer wrote.

---

We’ve heard that Zendesk users may be struggling to understand issue categories and detect new issues in tickets. If this is you, a big part of your challenge is getting to know the context and the root causes behind specific categories.

It often takes manual work. You can see all the categories, but you don’t have an easy way to know why a ticket was listed under a specific category. You have to spend time sifting through the support tickets grouped there to find out.

So the people on Zendesk we’ve heard from are trying to get around this limitation, but their solutions are reading tickets individually or exporting the tickets to try to understand issues using another platform. These are just creating new pain points in terms of time, and manual effort and frustration.

---

The solution has to be a way to identify issue categories without all of this work. That is what has always guided our integration with Zendesk at Keatext.

What we’ve built is an integration that connects right to Zendesk and pulls all your data in from the start. It categorizes all the tickets and builds your list of categories, even for new issues you haven’t seen before, and you have visibility into the customer comments that the system considered for each category. You have this really powerful flexibility with Keatext.

What I want to show in this series is not only Keatext’s ability to address pain points in Zendesk, but that we’re paying attention and listening to people like you.

That’s how we’ve been able to build a platform that addresses the real, day-to-day issues that matter to our customers.

For Zendesk users, categorizing tickets is only one piece of the puzzle. Coming up next in this series on Zendesk is how to cut through noise in the support center, a data quality problem of sorting through tickets.

---

Before you go, check out Keatext on the Zendesk Marketplace. We just gave our page a fresh look for the new year and updated it with our latest innovations. Thanks for reading!

The post Zendesk spotlight: Categorizing tickets based on content appeared first on Keatext.

For our 2024 year in review, we’re doing something a little different. Before we look forward to the future, we wanted to first look back at the conversations we had with CX professionals this year.

Those who shared their experience with us have primarily used genAI through the chat interfaces of models like ChatGPT and Gemini. We wanted to share some of their insights here as we felt that this could benefit you. Let’s dive in!

Many businesses have recognized the potential of genAI and decided to experiment with it for their customer experience tasks. To facilitate this exploration and drive innovation, these businesses offered CX experts access to powerful chatbots like ChatGPT and Gemini.

These chatbots allowed CX experts to tap into the capabilities of genAI, enabling them to gain insights from customer feedback, identify recurring issues, and enhance their reporting.

However, this journey towards genAI adoption in customer experience hasn’t been without its hurdles. CX professionals have encountered several challenges, particularly in areas like prompt engineering, report generation, and data preparation.

Perhaps the feedback we heard the most from CX professionals was how difficult it is to write prompts that consistently generate the same expected output structure.

Even among those who were happy with the quality of the insights from genAI chatbots, creating reports became a challenge that required a lot of manual work to combine charts and statistics and add their own comments and recommendations.

Another significant issue we heard was about data preparation, often in the case of support tickets where there was a need to detect the most recurring issues among thousands of pieces of feedback.

In this case, chatbots like ChatGPT and Gemini were able to successfully provide a sorted list of issues. What was more challenging was getting a sense of how many times those issues were reported. This kind of reporting is important for people like contact center managers, who measure the extent of issues in order to organize priorities among their teams.

These experiences suggest that despite the power of genAI chatbots, CX professionals need to invest substantial time and effort to recreate the reporting they used to have in other feedback analytics platforms.

Overall, the feeling that we’ve heard is that using genAI on its own – through its respective chat interfaces – is not enough to provide the level of reporting needed to monitor and understand trends over time.

At Keatext, we felt these pain points too. That’s why we continued to integrate genAI into our platform to give you the best of both worlds: powerful insights combined with outstanding automation and reporting.

Since March 2023, we’ve been leveraging genAI in Keatext to deliver the most advanced insights and strengthen your reporting. This year, we worked to:

• Enrich the insights Keatext generates to extend the types of insight you can analyze, going beyond sentiment analysis to enable you to reorganize insights for your reporting
• Expand conversational analytics into both customer and agent interactions and bring in more data sources like post-call surveys to understand the entire customer-agent relationship
• Add more connectivity to help you seamlessly integrate with your platform of choice including Alchemer and our new partner QuestionPro

Last December, we also ran a survey about what you would like to see next in the platform. Over half of respondents reported that you wanted to see a chatbot for insight queries. You asked, and we delivered!

Data enrichment in Keatext allows you to address specific insight queries in your reporting:

• Report beyond sentiment by reorganizing insights according to dimensions such as customer journey stages
• Never miss new issues so you can proactively resolve them as soon as they come up
• Identify categories automatically without the time-consuming work of reading customer inquiries

This capability enables you to unify insights across channels. A great example is bringing call transcripts to explain post-call NPS survey results, altogether providing a more comprehensive analysis of your customer journey.

In other exciting news this year, Keatext has partnered with two companies leading innovation in the CX space, QuestionPro and CallMiner, to deliver conversational analytics.

QuestionPro chose Keatext to unify insights across call transcripts and post-call surveys. The partnership will enable QuestionPro customers to unlock deeper insights into the customer journey.

With CallMiner, Keatext joins the new CallMiner App Marketplace as a solution to leverage customers’ existing CallMiner data and generate impactful customer experience recommendations.

These companies join our existing partnership with CustomerCount, a feedback management platform to which Keatext adds a layer of advanced text analytics to monitor and improve brand reputation.

Keatext’s partnerships represent the evolving impact of conversational analytics and unified CX insights. As more businesses look to the possibilities of genAI, Keatext plans to stay ahead of the curve, delivering advanced analytics that contribute to real business outcomes.

The post 2024 Year in review: Perspectives on genAI from CX professionals appeared first on Keatext.

Keatext and QuestionPro, two companies leading innovation in the CX space, have partnered to provide conversational analytics to our joint customers. What does that mean for you? Read the press release, or stay here for the details that matter most to you as a CX professional working with multiple feedback channels.

---

The landscape today

Call transcripts and post-call surveys are often treated as distinct channels. Analyzing them both together requires a lot of manual work and presents a significant barrier to insights.

In order to understand why customers gave a certain NPS, for instance, CX professionals need to find and read the transcript associated with each call. At scale, this process is inefficient and businesses are losing insight into customer satisfaction as a result.

By working to unify these channels, businesses are instead able to gain across-the-board visibility into the conversations associated with post-call surveys. This new dimension of analysis illuminates trends and issues that may not be clear from the NPS alone.

CX professionals need a strong analytics solution that brings together these feedback channels for more advanced insights. Today’s partnership between Keatext and QuestionPro aims to address this need in the market.

---

Making an impact with conversational analytics

Applied to the contact center, conversational analytics uncovers the “why” and the root causes of low NPS given in post-call surveys. It can even pinpoint the exact moment in a conversation that affected the NPS. This level of granularity is important for a complete understanding of issues in the contact center.

Even more interesting is that conversational analytics provides the ability to separate a conversation into the interactions from the customer and the agent. This enables CX professionals to monitor agent performance and address issues that may be impacting the NPS.

Before, CX professionals would have had to manually find and read the call transcript from a post-call survey with a low NPS. With Keatext and QuestionPro’s conversational analytics, unified insights are automatically developed. These insights can be actionized to prioritize efforts to address issues in your contact center.

---

At a glance

Keatext and QuestionPro’s partnership supports CX professionals and contact center operations. Here’s a quick look at the benefits that Keatext provides for QuestionPro customers:

• Unified insights: QuestionPro customers can now centralize all their customer feedback sources, including surveys, reviews, support tickets, and call transcripts or chats from the contact center into a single platform for unified analysis.
• Conversational analytics: By analyzing call transcripts alongside post-call NPS surveys, QuestionPro customers can pinpoint the exact moments in conversations that impact customer satisfaction. This granular level of analysis will empower businesses to take immediate action to improve customer experiences.
• Efficient reporting: Keatext’s cutting-edge text analytics technology enables QuestionPro customers to analyze unstructured text data with greater precision and speed in order to report efficiently on customer satisfaction.

---

About Keatext

Keatext enables businesses to unify all their customer feedback channels and understand satisfaction at every stage of the journey. With all customer insights in one platform, CX professionals can easily pinpoint trends and issues, report more efficiently, and make informed decisions that consider all aspects of the business.

The post Keatext and QuestionPro Introduce Conversational Analytics appeared first on Keatext.

The post How Product Operations is Evolving to Drive Business Decisions appeared first on Keatext.

The post Understanding the Impact of Member Experience on Credit Union Growth appeared first on Keatext.

The post Transforming Customer Insights with Alchemer Integration appeared first on Keatext.

The post How to Create an Effective Customer Feedback Loop appeared first on Keatext.

The post Voice of Employee for the Modern HR Leader appeared first on Keatext.

If you already started searching for solutions, you probably recognize that there is an inherent problem with your current approach to survey analysis. Manually reading responses is time-consuming and labor-intensive; sampling your responses makes it all too easy to miss critical insight from your audience; and relying solely on quantitative survey metrics leaves you with huge blind spots. What gives?

Automated survey analysis enables you to easily process the unstructured, open-ended text data you get from survey responses, shedding light on your blind spots and identifying patterns and trends that matter for your business. In this post, we’re focusing on the business applications of survey analysis for the voice of customer and voice of employee. Keep reading for our complete breakdown of survey response analysis and report building.

1. Types of survey questions

When we talk about survey questions, there are two major types that you should consider in your survey design: close-ended and open-ended.

• Close-ended questions: These types of questions have a predefined set of responses that your audience can choose from. Think of true or false, yes or no, or multiple choice questions. They are “closed” in the sense that it is not possible to input an unexpected response outside of what the survey provides. From a survey design perspective, there is little to “read between the lines” with these questions. However, they are great for survey respondents as they are generally low-effort and quick to complete.
• Open-ended questions: On the other hand, open-ended questions have no predefined responses and allow respondents to input whatever they want. Questions you’ve probably seen before like “How was your experience shopping with us?” or “Tell us if there’s anything we can do to improve” are bread-and-butter open-ended questions. The idea is to open the floor to your audience to get responses that you might not be able to predict. For example, a respondent might say “Your product arrived with missing parts” and you had no idea that this was happening. (And you’re probably not including a survey question that asks, “Did your product arrive with missing parts?”, ya know?) A methodology that includes open-ended questions gives you a level of organic insight you otherwise wouldn’t be able to get.

Similarly, another important distinction is between structured and unstructured data formats. Thinking ahead to the analysis of your survey responses, it comes in handy to know and recognize these data types.

• Structured data: This type of data is easily categorized because it has predefined responses. So as you can imagine, close-ended responses are a form of structured data because there is no possibility for variation in the data cells. Analysis of structured data is generally straightforward. Responses can be categorized with simple functions, cells can be counted, and so forth. Numerical data like NPS scores falls into this category.
• Unstructured data: In comparison to structured data, this type of data does not fit a predefined model of organization, like a table with rows and columns that has an expected set of values. Open-ended survey responses are unstructured text data. It’s important to consider how to analyze this kind of data because unlike structured data, you can’t just sort an Excel table. We’ll cover this in the coming sections. Anyway, here is a more in-depth post about unstructured data.

All this being said, what kind of questions and data formats should you use in your survey? Actually, rather than choose one over the other, you should incorporate a mix of both. With the right survey analysis methodology, you can bring together the relationship between quantitative data like NPS or CSAT scores with the qualitative insights you get from asking open-ended questions. Pair up an NPS question like “How likely are you to recommend the product you bought?” with an open-ended one like “What did you like or dislike about the product?” Responses to the second question will provide critical context as to why people gave a certain NPS score, so you can draw correlations between low or high scores with actual business elements, product features, and so forth.

Analyzing the responses to your survey questions will also be related to the different measurement scales. Close-ended questions are generally going to be nominal or ordinal scales (like NPS rankings), and open-ended questions will first need to be analyzed, and quantitative insights drawn out of them, before being measured properly. This two-step process is what makes qualitative analysis a bit challenging, but don’t worry, we’re going to get into it in a second!

2. Survey design considerations

Okay, so we’ve been over the general types of questions you can ask and data formats you collect in a survey. Which maybe you already knew, but it helps to cover it anyway. Now, of course, you have to actually plan and send your survey. What we want to get across in this section is that you should think ahead to how you’re going to analyze the survey first, as this will help you design a really great survey.

What’s great about having some automation in your survey analysis is that it actually enables you to think bigger in your survey design. Knowing that you have to analyze responses manually can limit what kinds of questions you might ask and the types of responses you opt to collect. It can easily lead to a quantitative-focused survey that misses insights coming directly from your audience. I mean, come on, we’d all shy away from asking open-ended questions if it meant we’d have to comb through all 1000 responses on our own.

As you design your survey, anyway, you should think about your overall objectives and your use case. Customer satisfaction? Employee engagement? A good exercise to begin with is to write down questions you have and want to answer. Keep an open mind because the responses you get could challenge your assumptions, and you have to remember to listen to the data. Here are just a few points to go over as you get into the nitty gritty of developing and wording your survey questions:

• Interrogate your questions: For each question, think about what kind of data format it returns and if it’s aligned with your objectives. Let’s say you want to learn about customer experiences at your store. You could ask, “Did you have a good experience at our store? Yes/No” but a more insightful question might be “Tell us about your experience at our store” as an open-ended text field. Whereas one question returns a nominal, structured value, the other returns unstructured data that contains richer insights. However, also make sure that you’re using the right type of question. If you use “Did you have a good experience at our store?” as an open-ended question, the responses are probably not going to be as great, because people might just write “Yes” or “No” without much explanation.
• Avoid biasing respondents: For an open-ended question, you should consider how a respondent might answer (or be biased to answer in a certain way) based on how it’s worded. The idea is to “stress test” your survey questions to avoid biasing the answers you get. Examples of bad survey questions are leading questions like “Did our food give you a stomachache?” Sure, you might get some answers, but if your objective is to get unexpected feedback from customers, it’s too narrow and not giving space for people to highlight what they feel is important, by adding details and context. How can you know about an issue without asking about it? You have to let your audience mention what is important to them.
• Consider sample size: Sample size is the number of respondents in your survey. You want to send your survey to enough people, and get enough responses back, to have a good amount of data that represents a broad spectrum of your audience. General advice is that you should have at least 100 respondents in order for the results to be statistically significant. When you have an automated survey analysis tool, too, you can increase your sample size without worrying about what to do about the analysis of responses.
• Cross tabulating segments: When you design a survey, you should consider the segments of your audience you might want to look at in more depth like age, gender, and location. For instance, you might find an important pattern between age and location in your data. Then, you can also pull in the open-ended responses from these segments and better understand what’s going on in your audience. All of this information is called metadata, and it plays a big role in analysis and reporting, which we’ll cover in later sections.
• Plan your analysis methodology: This one is kind of a no-brainer given the subject of this post, but hey, it bears mentioning regardless. Thinking ahead is the core concept that we’re trying to convey here. Before you build and send a survey (which takes time and effort!) and get responses from your audience (time and effort that they’re giving to you), you want to be sure that it’s going to provide the right insights.

The TLDR of good survey design is that it always depends. It’s a delicate balance of understanding your business, knowing what you want out of your survey, and crafting the right questions to give you actionable insights about your audience. As long as you’re informed and thinking ahead, you’re setting yourself up for success in the next step of the survey process: analysis!

3. How to analyze a survey

The first question is, what do you want to analyze? There’s a lot to consider. Since we’re focusing on open-ended surveys in this post, we won’t dive into quantitative analysis methods. You’re here to learn about what to do about all that text data!

When it comes to analyzing open-ended questionnaires, the first step is coding. Coding a survey is just the term we use for organizing text responses into categories for ease of analysis. You see, the two-step process of qualitative data analysis that we mentioned earlier ultimately results in being able to conduct quantitative analysis on that data. Coding translates unstructured data into structured data. By sorting similar responses into buckets, you can more accurately detect patterns and see trends. Now, there are two ways you can go about coding:

• Manual coding: As the name implies, manual coding is done by a person who is responsible for reading responses and categorizing them correctly. There is variation even within this, for example oftentimes the person relies on a codebook of predefined categories that the business uses for analysis. Manual coding is precise but time-consuming. Also, whenever a person is involved, there is a chance for bias to impact the results of the analysis. For small-scale surveys, manual coding is not a bad idea. However, it’s inflexible if you ever want to scale the size of your survey.
• Automated analysis: Automating survey analysis is usually achieved with text analysis solutions that use machine learning and natural language processing techniques to process large volumes of unstructured text data. The long-term advantage of automated coding is that it’s scalable. Because AI is in the mix, these solutions are strong enough to handle a lot of data, and new data at that. Many of the disadvantages of manual coding are solved by automated analysis. For instance, let’s say you send a weekly survey with the same questions every time. With manual coding, each person doing the analysis will have a different bias, and the organization of data won’t be consistent. Automated analysing in this sense enables you to keep analyzing new data with the same, consistent framework that returns accurate results and is reliable over long periods of time.

The added benefits of automated survey analysis is the visualization of data. Good survey analysis platforms allow you to view insights and build dashboards in a seamless fashion. With manual analysis, you’ll spend a great deal of time fussing with the data, exporting it, and building visualizations that you just might be happier to automate.

Case study: BRP turns 10 years of surveys into insights with Keatext

Bombardier Recreational Products (BRP) is a global leader in powersports vehicles that has built a name for itself crafting beloved products like the Ski-Doo and Sea-Doo. With a wealth of meaningful open-ended survey responses but no clear way of turning it into insight, they turned to Keatext to automate their survey analysis.

• The situation: The team at BRP was building their voice of customer program from the ground up – that meant 10 years worth of surveys largely unanalyzed
• The problem: With survey feedback mostly siloed at the organization, the team needed a way to not only consolidate insights but implement a solution seamlessly
• The solution: Keatext’s advanced text analysis was able to provide rich insights from BRP’s qualitative survey responses and create a consistent framework for analysis
• The impact: Decision makers came together to lead organizational changes at BRP that put the customer at the center of their work, which was now possible thanks to Keatext

BRP’s global customer advocate, Myshka Sansoin, says: “The biggest improvement Keatext brought to us is facilitating or even participating in our culture change. BRP wants the customer to be at the center of everything it does. Having an easy way to understand and share what the customer wants helps us accomplish this.”

Read the full case study here.

4. Visualizing survey results

Visualizing data and insights from your survey is a key element of your own exploration of insights as well as, ultimately, what you present in your survey report. Let’s take a look at some of the common visualization strategies that you can use or might come across.

• Word clouds: These visualizations are generally used to capture your interest and bring forward the biggest topics identified in your data. Word clouds rank word frequency by size to create an easily understandable image. However, word clouds lack most of the context, sentiment, or “why” behind the word frequencies that you see. They simply can’t present any meaningful relationship between topics and opinions, let alone any metadata in your survey. So while word clouds are easy (and admittedly, a bit fun) to create, they come with a lot of blind spots, especially for those of us looking for a deeper level of insights.
• Excel: Using a spreadsheet tool is great for structured data with predefined responses. It’s possible to fanagle around with Excel to infuse some element of automation and code your data, and build visualizations from this, but ultimately this is not a sustainable approach. Again, for small-scale surveys you could get away with it, but the limitations are clear, and in the business context, you won’t get too far with this method.
• Tableau: Dashboarding tools like Tableau allow you to build visualizations from your data. The level of customization and accuracy here is much greater than word clouds or Excel charts. Think of it as a more powerful Excel visualization. However, it relies on the analysis already being done and identifying topics and opinions in your survey responses – something that is more efficiently achieved with a text analytics solution.
• Text analytics dashboards: Most text analytics platforms today offer dashboarding functionality to visualize results of the survey analysis like sentiment categorization, topic detection, category grouping, and so forth. The advantage of a text analysis dashboard as opposed to one through Tableau, say, is that it is fully integrated with the data analysis. Whereas in the latter you have to export insights and reupload into a system unfamiliar with your data categories, with an integrate dashboards you would have everything at your fingertips that you need to build strong visualizations.

The strength of your survey result visualizations comes down to the analysis of open-ended responses: topic detection, keyword extraction, sentiment analysis, and all that, and the metadata you have in your survey responses: demographic information like age, gender, location, and so on. This information acts as a layer on top of the core text analytics engine in order to enrich visualizations.

Dashboards are probably your best bet for presenting your survey results in the most comprehensive way. You can add multiple charts, apply metadata filters, segment groups, and pull in as much context as you need to get to the bottom of what’s happening in your survey data. Let’s take a look at some visualizations you will see on most text analysis dashboards. Images are from Keatext dashboards!

Line graph: A line graph is best used to track changes over periods of time. In the example below, you can see the relation between two lines: number of records (i.e. comments from customers) and the overall sentiment score (i.e. how positive or negative those comments are). You can see that in November there was an increase in positive sentiment.

Bar graph: A bar graph is best used to compare things between different groups of metadata. A best practice is to use a bar graph with more than 3 or 4, but less than 12 data groups. In the example below we can see the most frequent topics identified in the survey broken down by the sentiment of the associated opinions.

Pie chart: A pie chart is best used when you’re trying to compare parts of a whole and see the relative percentages. You should use a pie chart with up to 5 groups but beyond this we recommend using a bar graph. For surveys, this visualization is an excellent way to show a breakdown of NPS or sentiment scores. The summary value in the center is the calculated average.

Table: You should use a table when you want to plot a lot of data, generally if you have more than 12 groups you want to visualize together. A simple table plots rows and columns while a pivot table allows you to “pivot” over time or any other variable. Tables are good for showing the whole picture, and the advantage is that they are very sortable.

Heatmap: A heatmap shows the intersection of different data groups. In our example, it is an excellent way to understand long-form text questions, by showing the interaction between topics and opinions. We can easily identify the areas that have the most frequency and determine if there are any problems – such as “shocking customer service”.

5. Preparing a survey findings report

Analyzing a survey for your own internal purposes is one thing, but especially in the business context you will eventually want to show what you have discovered with other members of your team or upper management. This section walks you through what makes a good survey results report and how automated survey analysis solutions can support your workflow in this area.

What goes into a survey report? In general, you’re going to present a series of insights that you identified as valuable for your business along with some written explanation, context, and recommendations. Depending on who you’re presenting to, you could include the raw numbers from your survey like how many respondents there were, the results of close-ended responses, average NPS or CSAT score, and so forth. It’s really up to you, but these are the essential components.

The survey results report is essentially the outcome of all your work thus far to analyze the survey data and explore insights. In many cases, this report is how you communicate directly with decision makers about how to improve customer or employee experiences. Here’s what can go into your report:

• Visualizations: Based on the objective of your report, and the audience, you should build out the most appropriate visualizations that dive into the big insights and support your main talking points. As we saw in the previous section, you can choose the right visualization for what you are trying to convey. The more you do this, the more intuitively you will understand how each visualization can contribute to your survey results report.
• Interpretation and context: You have to do a little “fine tuning” to get your report just right. Good reports have some level of interpretation and context rather than only a collection of visualizations. Every visual should have some text that explains what we are seeing and what to make of it. Your interpretation of the results is likely what your audience is looking to hear!
• Recommendations: Depending on your role, you might be responsible for providing some guidance and recommendations as to what to prioritize in the business. Strong text analytics platforms like Keatext are now able to provide recommendations for elements or topics that have the greatest impact on metrics like your NPS score, for instance. Just like your interpretation of survey results, your recommendations can hold a lot of weight in your survey analysis report.

The advantage of using a text analytics dashboard to visualize your insights for your own purposes is that you can use those same visualizations in your survey analysis report, seamlessly. Platforms like Keatext allow you to export dashboards or share a public link so other stakeholders can see it. Keatext even automates reporting for you, providing an executive summary you can download that highlights your top opportunities to improve, along with actionable recommendations on how to start making changes, generated in natural language.

Choose Keatext as your survey analysis and reporting solution

At Keatext, we’ve built a strong platform for the analysis of open-ended text data like survey responses, and we’re constantly iterating on it to improve its usefulness in all elements of your everyday work. Here’s what makes our platform ideal for survey analysis:

• AI-driven text analytics: A survey analysis platform is only as good as the quality of insights it can uncover for your business. Our core text analytics engine is strong enough to learn and adapt to the unique context of your data.
• Automated, ready-to-share reports: Keatext consolidates your top opportunities for improvement identified from your survey responses in one easy PDF. You can instantly export a summary of these insights to share directly with the right decision makers.
• Recommendations: Thanks to one intelligent integration, we built a way to use Keatext as a knowledge base for OpenAI’s GPT to generate natural language recommendations on how to improve areas of your business that have the most impact on your NPS, CSAT, or CES scores.
• Dashboard visualizations: Customize your dashboards with flexible widgets like a pie chart, sentiment score, heatmap, time series, and more. (Remember the visualizations we went over earlier? We’ve got them all!)
• Integrations: Keatext integrates with survey platforms like Surveymonkey for a fully centralized analysis experience. Data can be pulled in real-time so you never miss critical insights from your survey responses.

As we have seen together, there is a lot that goes into survey design, analysis, visualization, and reporting. We hope that this post has shed some light on these important elements and made you more informed and confident about managing surveys at your business. Thanks for considering Keatext in your search for a survey analysis solution. Good luck with your next survey analysis project!

The post How to Conduct Open-Ended Survey Analysis appeared first on Keatext.

On top of that, maybe you have some procedure for how to manage and automate those tasks – or maybe you don’t. Either way, this work is complex and can be strengthened with the right technology and software solutions.

In this post, we’ve brought together the most valuable insights from other posts we’ve written, in order to give you a complete view of the activities involved in the voice of the customer. As a text analytics company, we’re deeply involved in the customer experience space and we know how the pieces of the process fit together. We’ve tried to keep this post conversational and less “SEO”. If that’s your style, read on!

First, let’s define a few things

Before we jump into this post it will be useful to set a few definitions at the start. So let’s get to it:

• Customer experience (CX): Refers to how customers feel at every interaction they have with a business, whether it’s online, on the phone, or in-store
• Customer feedback loop: Describes the activities involved in collecting feedback, analyzing data, finding insights, and taking actions to ensure customer satisfaction – altogether these activities help you “close the loop” with the customer
• Voice of customer (VOC): In business, VOC refers to what customers are saying about a company; a business can collect voice of customer data from online feedback channels
• VOC program: Refers to what a company is actually doing about the feedback they are getting from customers, using voice of customer software to close the customer feedback loop

1. Why you need a voice of customer program

We have a complete post on this topic that you can check out! The TLDR: businesses that invest in the voice of the customer see a measurable return on investment. Studies have shown that these companies:

• Have much higher client retention and employee engagement
• Spend less on customer service
• Generate a year-over-year increase in annual revenue that’s 10 times greater than other companies

Especially since the huge shift to digital experiences in 2020, customer experience has emerged as the new battleground for brands. Having a voice of customer program is mission-critical to stay competitive. The benefit of a voice of customer solution is clear: you simply have more data to make decisions. Wouldn’t you want to know exactly what this data can tell you, if it would help inform the direction of your business?

Business aside, there’s a philosophy to the voice of the customer, too. It’s not only about data – it’s about people. When people take the time to write comments and reviews, answer surveys, or chat with your customer support team, it’s because they have something to say. If you don’t take the time to use that data, it’s a breach of trust. And trust is what customers value above all else today: as much as 83% won’t buy from a brand they don’t trust.

In reality, though, there are lots of organizational challenges to fully embrace a data-driven, customer-centric mindset and put a VOC solution at the heart of your business. Not only that, but you would also want to do it well. The bigger a company is, the more work involved to centralize omnichannel customer feedback for easy access to insights and clear paths to action. That’s what the rest of the customer feedback loop is for – and where technology comes into play.

2. What to consider in collecting customer feedback

The quality of everything that follows in the customer feedback loop depends on the quality of feedback that you collect and how you organize it. A solid voice of customer methodology that standardizes your VOC practices will help guide your feedback collection process. The main challenge is when you have several distinct channels where feedback might be coming in: reviews, surveys, support tickets, chat conversations, social media, even contact center phone calls.

Building trust with customers means opening channels for communication. Instead of shying away from the voice of the customer – especially when it’s negative feedback – giving customers a platform to voice their opinions is actually a good thing for your business. That’s because the customer can speak their mind and it allows you, as a brand, to respond thoughtfully and mitigate the potential loss of a customer. Most customers who complain just want to be heard; in fact, up to 70 percent will do business with a company again if the complaint is resolved, and if the complaint is resolved quickly, that number jumps to 96 percent. Essentially, what matters is that you use negative feedback in a positive way.

Negative feedback is generally what we talk about with “closing the loop”, as the actions you can take to respond to these customers can have a huge impact on your business. For instance, you can follow up directly with customers who left survey responses associated with a low satisfaction score (detractors, if you’re using NPS terminology) in order to try to save them from completely churning. Voice of customer tools enable you to manage these follow-ups.

As you can see, the name of the game today is actionable feedback. If you’re collecting feedback and listening to the voice of the customer, it shouldn’t be empty data; it should lead to a proactive next step. Let’s consider how you can make different feedback channels actionable:

• Surveys: Crafting the right open-ended questions, paired with an NPS or CSAT score, to understand exactly which elements of the customer experience are creating friction
• Reviews and social media: Responding quickly and thoughtfully to negative comments to mitigate the customer’s bad experience and their impact on brand reputation
• Chat conversations: Using conversational data to map out and automate answers to common questions
• Support tickets: Measuring customer sentiment associated with a particular agent and using these insights to inform agent training and help them better answer customers’ concerns

3. Setting up your feedback channels for success

Each feedback channel can be managed in different ways to enhance the quality of VOC data collected. You can imagine that certain channels like surveys can be curated more than others. For instance, CX teams have control over the questions presented in a survey, whereas they do not have the same control over product reviews. So, considering how to best collect feedback from each channel is important for the success of your VOC analysis.

• Voice of customer surveys: Any kind of customer survey allows you to directly hear open-ended feedback from customers. Of course, this is where survey design comes into play. Especially if you have the right VOC tools in place to analyze the responses, you can transition from closed-ended questions to open-ended questions. Imagine the different levels of insight you can get from asking “Were you satisfied with your experience with us” versus “How do you feel about your experience with us?”.
• Comments on social media and in reviews: In this scenario, feedback is simply aggregated. Interestingly, though, this channel is highly important because of the fact that you have no way to guide customers’ comments. We will see later in the post that “learning what you don’t know” is a huge part of successful voice of customer analysis. Customers on social channels are telling you exactly what they care about, regardless of whether you asked or not. It’s your job to listen!
• Chat and contact center interactions: Any time a customer speaks with an agent through a chat, ticket, or even phone call, you may consider that there are two parties involved in the customer’s experience: your general brand and the person representing it. Customer satisfaction can be measured in relation to their experience with a particular agent at your company. As these people are working often with at-risk customers who are experiencing friction with your products or services, their ability to answer questions, respond empathically to concerns, and make the customer feel supported contributes greatly to your customer experience. Don’t let it go unanalyzed!

The voice of the customer goes beyond the process of understanding what the customer says. Ultimately, you have to understand the root causes of customer issues to truly fulfill the feedback loop. It’s not only about responding to customers, necessarily, but fixing product issues, improving customer service experiences, and making the customer feel like you have taken steps to address their concerns. In order to get to this level of actionable insights, look no further than a text analytics solution.

Case study: Lenovo analyzes global licensee feedback for 250+ products

As we saw in the previous section, collecting feedback is the important first step to accurate text analysis and actionable VOC insights. Keatext helps Lenovo aggregate feedback from all their global licensing channels and manage a unified feedback analysis system.

Quick view:

• The situation: Licensees in 180+ markets with 250+ products
• The problem: Lenovo faced difficulty accurately collecting customer feedback to measure licensee performance
• The solution: Keatext’s scalable voice of the customer analytics have tackled 120,000 comments and counting for Lenovo
• The impact: Digital transformation of Lenovo’s feedback analysis system to improve licensee performance and onboarding

Implementing Keatext has resulted in a simpler, more automated way for the team at Lenovo to manage quality across a diverse portfolio of licensees and marketplaces. Keatext adds another layer of depth and analysis, improving visibility into licensee performance while providing Lenovo with the actionable insights they need to make ongoing improvements.

Read the full case study here.

4. Making your VOC program actionable with text analytics technology

Collecting feedback, of course, is only one piece of the puzzle. Analyzing the feedback you’ve collected is a challenge on its own, and there comes a point where you will need a technology solution to manage the volume of feedback you’re getting and dig into more granular insights. Text analytics technology is the bread and butter of the analysis involved in a VOC software. It automates the important work of simply reading through all the comments left by customers.

As a technology applied to the voice of the customer, text analytics can:

• Identify topics mentioned by customers and the associated sentiments
• Detect trends in the data and track them over time
• Pinpoint the topics that have the greatest impact on a customer satisfaction rating like NPS or CSAT
• Generate recommendations on how to improve business areas that are reflecting a poor customer experience
• Produce quantitative voice of customer metrics from qualitative text data

This level of automation leverages AI, and when compared in this lens, not all text analytics solutions are the same. Keatext’s CEO Narjès Boufaden tells us that “with old-school text analytics, companies had to come up with 20 or 30 keywords to search. That isn’t good enough anymore. AI now identifies the important words on its own, learning in context without the need for anyone to customize a vocabulary of relevant words or phrases for each industry. Then it points to issues you don’t even know exist. It uncovers new insights and new client needs.”

That right there is an essential component of an effective VOC strategy. Text analytics removes bias from the analysis of feedback and ensures the accuracy of insights. Without it, you could think you know where the problem lies, end up spending money on the wrong things (even with the best intentions), and ultimately jeopardize your VOC program’s success and executive buy-in. The ability to uncover insights that surprise you – the idea of “learning what you don’t know” – is a crucial element of truly listening to the voice of the customer.

When we talk about making insights actionable, getting them to the right business units is the necessary second step after finding the insights themselves. Text analytics technology is the engine that enables you to pinpoint voices of the customer (see what we did there?) that have a big impact on overall satisfaction. Then, you have to report and share these insights. How do you do that?

5. How to report and share insights to close the feedback loop

The basic idea of this stage of closing the feedback loop can be summed up as “insights are no good unless you can get them to the right people”. It’s easy to overlook this stage: out of the 95% of companies that collect customer feedback, 10% actually apply it to product developments, and only 5% follow up directly with customers about the feedback they left. That’s a pretty elite top 5% to be a part of!

Generally speaking, VOC best practices rely on feedback management software like Keatext to assist with the reporting process. This kind of software has a core text analytics engine to analyze data, and built on top of it are functionalities that make the solution practical and capable of addressing the changing needs of business users. Features like data visualization, dashboarding, report creation, and predictive analytics help the people on the ground prepare and share insights with the right stakeholders, who are then responsible for taking the appropriate actions.

A strong VOC program ultimately supports the business, but in order to do this it has to first support the end users (that’s you!). Managers and analysts within CX teams are responsible for preparing reports, developing recommendations for action, and sharing these assets with the correct business units. Any automation that can be applied to these tasks reduces the time it takes to close the feedback loop, whether it’s replying directly to a customer or planning important product changes to solve customer pain points. And as we saw earlier, time is of the essence when it comes to responding to negative comments and detractors.

Case study: Intelcom builds a data-driven VOC program with Keatext

Keatext gave Intelcom’s team the scalable analytics they needed to evolve from doing manual VOC reports to championing a robust CX strategy that brings together multiple teams at the organization around a common methodology: using customer insights to drive decision-making.

Quick view:

• The situation: 5000 daily responses from post-delivery and post-return CSAT surveys
• The problem: No data-driven, scalable VOC solution to gain insights from open-ended responses
• The solution: Keatext’s strong text analytics and SWOT recommendations formed the foundation for Intelcom’s VOC methodology
• The impact: Digital transformation of Intelcom’s CX practices, led by the team at Intelcom championing Keatext

Keatext introduced a level of automation and data analytics that enabled Intelcom’s team to identify the most common concerns, questions, and suggestions from customers. Using Keatext for this purpose enables Intelcom to automatically categorize cases and route them to the most appropriate business unit. This has a high ROI for the team, directly impacting responsiveness and helping them to better structure their service offerings.

Read the full case study here.

6. Getting executive buy-in for your voice of the customer program

For those of you who are reading this article in the hopes of launching your own voice of the customer strategy, this section is for you. We’ve all been there: you totally know that a certain strategy is the right move for your business, and no matter how well you champion it to upper management, they just don’t see the value. Thankfully, that won’t be you – because you’re going to have a plan.

Executives considering a new initiative will mostly be concerned with the return on investment (ROI). As a business, if you’re going to spend money on something, you hope to see that investment reflected in greater performance. In the realm of customer experience, this could look like:

• Greater customer retention and less churn
• Increase in customer satisfaction scores like NPS
• Improvement in brand reputation online
• Customer loyalty reflected in repeated purchases, upsells, or resells
• Understanding of how CX metrics translate to larger business objectives, especially revenue

Advocating for the tangible results you hope to see from your VOC program will go a long way in demonstrating ROI at the executive levels. Other considerations beyond this are related to any specific software solutions you hope to include in your program. Especially for text analytics software, you should carefully consider how you will handle elements like adoption, organizational change, scalability, and integration with your existing tech stack. Never fear, though: we have a full article on these considerations!

Securing the executive buy-in involves some level of predicting performance based on research, case studies in your industry, and benchmarks. Once your VOC program is approved and launched, you can begin to measure performance and track the impact on revenue much more accurately. This will enable you to continue to get that executive buy-in and support your future initiatives! Planning ahead and preparing what you need to pitch your idea will not let you down.

Choose Keatext for your voice of customer program

If you found this post useful, you might find our platform even more so! Here are a few points about how Keatext is made to support you and your VOC program:

• AI-driven text analytics: A feedback management software is only as good as the quality of insights it can uncover. Our core text analytics engine is strong enough to learn and adapt to the unique context of your data.
• Automated voice of customer reports: Keatext aggregates the top opportunities identified from your VOC data to improve customer satisfaction, all in one easy PDF. You can instantly export a summary of these insights to close the customer feedback loop.
• Recommendations: Thanks to an intelligent integration with OpenAI’s GPT, we built a way to use Keatext as a knowledge base to generate natural language recommendations based on the voice of the customer.
• Dashboard visualizations: Build and customize VOC dashboards with flexible widgets like a pie chart, sentiment score, heatmap, time series, and more.

A lot of the work we’re doing is to automate the activities involved in closing the loop with customers beyond the analysis of data, like visualizing insights, preparing recommendations, and sharing voice of customer reports. At Keatext, supporting people like you in your everyday work is always on our minds as we continue to build our platform. If you believe in your VOC program, we believe in you too!

The post Considerations for Your Voice of Customer Program appeared first on Keatext.