Why you need to analyze your customer feedback data

How to analyze your customer feedback data

Very few management executives need to be convinced of the value of customer feedback. At a time when customers have a multitude of choices available when selecting a product or service and brand loyalty is highly prized, there’s little margin for error.

Customers are quick to switch brands when they’re dissatisfied, with as many as 33% reporting leaving a brand after a single negative experience.

The good news? Customers who are deeply engaged are eager to leave feedback, and they show their appreciation with their wallets, buying 90% more frequently and spending 60% more when they purchase, while reporting brand satisfaction 5x higher than other customers. Listening to feedback from these customers is an excellent way to reward your greatest brand advocates with more of the products and features they love.

However, while some customers are eager to share critical feedback before taking their business elsewhere, many do not - 91% of customers who leave do so without leaving any feedback. Successful companies are diligent in gathering feedback, reviewing for insights, and creating action plans to deliver enhancements that keep customers coming back for more.

Turn customer feedback into a winning business strategy

Just gathering feedback isn’t enough to spot gaps in your service or catch potential deal-breakers in product offerings. Instead, a serious and structured analysis program is necessary to actually dig in and mine feedback for the most valuable and actionable data points.

Understand your customer’s “why” with an analysis program

The truth is, your customers are probably already leaving you feedback, even if you don’t have a collection program in place. They’re just relying on a variety of third-party review sites - Google Reviews, Yelp, social media platforms - to share their thoughts about your company. While it might be a morale boost to hear what customers enjoy about your product, and a few nuggets of useful information may surface from time-to-time, relying on third-party sites for customer insights isn’t a winning business strategy.

A true feedback analysis strategy starts with a goal - an impactful metric that your company wants to move the needle on, or measuring progress towards a milestone on a strategic plan.

The next step in a concrete strategy is to deploy a simple-but-effective method of feedback measurement, such as CSAT or NPS. For example, maybe you want to measure the effect of a recent feature update on your retention rate and overall customer loyalty trend. This is an excellent opportunity to target high-use customers with an NPS survey that can show whether the new feature is converting Passives into Promoters, while also gathering open-ended feedback about specific reasons why the update is a winner.

Use text analysis to unlock deeper data points

No matter how feedback is collected, it’s an unfortunate truth that far too many companies simply rely on the top-level results and never look deeper. In other words, they simply look at the percentage of satisfied customers appearing on a CSAT (customer satisfaction) survey or track the trend in self-reported Promoters and Detractors from an NPS (net promoter score) report.

Both CSAT and NPS survey methods have come under scrutiny lately, with many critics dismissing them as too simplistic or calling them inaccurate predictors of actual customer happiness. The truth? The critics really are right, honestly - but only to the extent that the companies they point to are either using these surveys incorrectly (poorly designed and ill-targeted surveys deliver dubious results) or because they never dig below the easy-to-spot quantitative numbers to get to the rich quantitative data points in verbatim comments.

Determining that survey scores are “largely positive, so we should keep doing what we’re doing,” or that “scores are trending down, we need to find out why and fix it” give companies starting points for further analysis. However, reaching actionable data points to make meaningful change requires digging deeper, looking into data sets for correlations between feedback and actions that can be leveraged to drive real business growth. Use high-impact text analysis tools and methodologies to mine verbatim feedback for deep-down insights about how customers truly feel about your company, and what you can do to keep them engaged. With this added context, you’ll gain a clearer understanding of the relationship between data points and how tweaks to your product can shift customer feedback trends in a positive direction.

Identify actionable data with categorized analysis

Consider a set of NPS data where you’ve identified a particular group of Detractors that you want to target for retention. Their raw score has already segmented them for you, telling you they’re dissatisfied and likely to churn - but now you want to know why they’re unhappy, and what you can do to re-engage with them.

The best way to accurately gauge what is motivating a particular group is to read the details in their open-ended feedback, then categorize them for accurate data mapping and clearer analysis. This way, you can quickly translate comments such as “the new pricing structure doesn’t fit my team’s budget” or “the reporting interface makes it difficult to generate reports I can use in planning sessions” into quantitative data points. Information like this might be filed under one or more categories, such as “pricing structure” and “budget,” or “U/X” and “reporting tools.” From there, it’s a much simpler step to begin mapping trends and correlations between customer feedback comments.

Now you’re on your way to building a data set that accurately conveys the “why” behind this group’s inclination to leave your platform.

This will give you true VoC feedback that can be used to make intelligent strategic decisions.

Executive leaders and project managers need smart analysis like this to make critical decisions, and hearing that a significant percentage of your targeted Detractors are frustrated with a specific issue simplifies leadership decision-making when allocating resources and planning project roadmaps.

Supercharge the power of text analysis with machine learning

The process of categorizing verbatim feedback may seem off-putting, and understandably so; many companies still approach this as a manual process, developing a coding/framework system that employees apply to review and tag each feedback comment individually. As you can imagine, this process gets very tedious very quickly when you have hundreds or thousands of customer feedback responses to analyze! With a small data set, there may not be much time involved overall, but with larger data sets or ongoing survey campaigns, the coding process can be exhausting and time-consuming - and not to mention, inaccurate. However, there’s a more efficient way to approach this - with the help of modern text analytics technology.

With advances in machine learning, it’s increasingly simple to automate much of the process of analyzing and coding text responses, delivering deep, actionable insights into critical feedback data at an impactful pace. Machine learning enables text analysis at scale, identifying trends in feedback by tracking specific topics, terms with strong statistical relationships, sentiment of customer comments and how these comments have changed over time.

Machine learning models break down into two categories, broadly speaking.

Unsupervised models are highly autonomous, and can quickly identify trending topics with no manual filtering. This means that analysts can see everything customers are talking about, not just the things they know to look for.

This is a powerful way to spot trends that could slip by manual reviewers without notice, surfacing actionable insights very quickly.

Supervised models are also powerful, but require significant time investment to set up and on an ongoing basis, as team members monitor results and rearrange topics to help train the program. The downside is that even with all of this ongoing training, a supervised system still only looks for what an analyst tells it to look for. The fall out from this is that as new topics emerge, they are not automatically discovered and organizations are none the wiser and on the back foot as new and emergent customer issues arise and become systemic in the business.

While both models can identify and track the way critical topics trend across datasets (different customer demographics or data collected over specific intervals of time, for example), unsupervised models offer greater scalability and efficiency, allowing for quicker reaction and the ability to pivot strategies more efficiently.


Customer feedback is a gift. Think about it – your customers are taking the time to respond to a request for information about their experience with your company. They’re just as busy as everyone else, so the time they take to give you feedback is incredibly valuable. They expect that in return, you’ll be reviewing their feedback and considering how to adjust your services to make their next experience even better. Your customers know this too - with a majority of customers stating that they believe companies need to listen to their feedback to make improvements, it’s clear that they have something important to tell you. Companies that only look at basic quantitative data and pass over or ignore the deeper insights available through textual analysis are not only missing out on valuable opportunities to deepen customer engagement but are ultimately leaving themselves open to potential risks.

Everyone wants to do customer insights right but they're setting themselves up to fail if they rely on tools that automate what humans already do manually. That's just automation and isn't true innovation. Customer insights teams today are allowing technologies to reveal the areas that need the most improvement.

Share to: