People data doesn’t always fit neatly in a spreadsheet. Much of what we want to know can’t be quantified numerically. Often we want to learn about the qualities and nuanced aspects of work or work processes. Or, we want to understand the feelings, sentiments, and experiences of people. That’s where qualitative data comes in. Qualitative data is any information that cannot be measured numerically. It is often unstructured (meaning that it doesn’t have any formal pattern or way in which it is organized) and can come in various formats like text, audio, images, observations, and more. For example, qualitative data in people analytics might include verbal responses to interview questions, transcripts of what was said during a focus group session, open-ended survey responses, feedback comments from performance reviews, observations of what occurs during team meetings, social media posts related to the company, and even employee-generated content like blogs or presentations.

When you go into qualitative analysis you leave behind the comfort and simplicity of numeric responses and easy-to-read charts and graphs. Qualitative analysis is messy. There are no hard metrics and there are no “right” or “wrong” answers which can be daunting, especially for those of us who like being right or getting “A’s” on our homework assignments. In qualitative analysis, we work with interpretations. This can be tricky as not even the best qualitative researchers will always interpret the same data in the same ways, but these interpretations are the strength of the qualitative analysis. When you don’t force things into preset options, you can be rewarded with new and interesting insights you could have never imagined. There is no “fix” for the ambiguity inherent in qualitative analysis because the ambiguity is not necessarily a problem to solve; embrace the messiness. That’s where the value lies! (Tip: For those of you who can’t handle complete ambiguity, don’t fret. There are data management, research methods, and validation techniques to help give order and create confidence in your data, analyses, and findings.)

While some organizations are doing awesome qualitative people analytics today, most qualitative data goes unused. How many surveys, interviews, comments, workshops, or sessions have you seen where information was captured but nothing ever happened with it? It’s not that people don’t care or don’t want to use the information, it’s that many people haven’t invested the time in learning or applying the analytical approaches available for analyzing and making sense of it. If you were to ask companies today why they don’t do more with the qualitative input they receive from their employees, they would say it takes too much time or they don’t have the resources. But all analysis can be time-consuming depending on how you do it. And, while it is true that analyzing lots of unstructured data can be more time-consuming, especially if you are using manual analysis processes, there are methods to simplify the process and recent technology has decreased the difficulty level and time needed for undertaking the effort. Interviews can now happen over video conferencing, observation sessions can easily be recorded, text can be automatically transcribed, and there are even automated tools to help structure, code, and theme the data. With the technology available today there are fewer excuses not to pay attention to what people are experiencing or to listen to what they have to say. Companies already invest in spreadsheet, dashboard, and database hosting software to understand numerical data, it is just as easy to invest in tools that allow us to catalog, map, and contextualize qualitative data. Qualitative analysis allows us to reach beyond the artificial limitations of trying to force our understanding of people into only numbers, categories, and check-boxes and instead seek to understand the experiences and emotions of the full human.

There is also a terrible but common misconception that qualitative analysis is ‘fluffy’ or not ‘real’ analysis. I will admit that I have seen some people give qualitative analysis a bad name by picking and choosing which subjective information they share, by doing a biased personal summarization of information, or by making a word cloud and calling it a “qualitative analysis.” Those things are not qualitative analyses. Qualitative analysis is a systematic, rigorous approach to finding insights by analyzing information that just so happens not to be numbers. It is very much ‘real’ analysis.

Qualitative data can be used for descriptive, diagnostic, predictive, and prescriptive analytics. It can answer questions like what happened, why did it happen, what might happen, and how can we make it happen, just like quantitative data. But it is particularly well suited to answering an additional question, “In what way?” When we understand the qualities and details of a thing, our understanding of what, why, and how becomes deeper.

There are many skills necessary to be successful in qualitative analytics, including many non-technical skills like active listening, which we discussed earlier. Since we are in “Step 3. Build Your Technical Skills” we’ll focus on the technical analytics skills for qualitative analytics in this section. If you want to turn unstructured qualitative information into insight, here are some commonly used skills and techniques you can work on building:

• Qualitative Data Coding: Coding is the process of assigning a code (you can also think of it as a label, category, or even a hashtag) to segments of your qualitative data to help you organize and analyze the data. There are different coding approaches you might use depending on your specific goals:

  • Object Coding: Assigns codes to identify specific things, events, or people mentioned in the data. You might use object coding to identify process improvements for employees at work (e.g., tagging which systems were mentioned by employees), to determine which learning resources to offer (e.g., a topic that receives multiple asks for help in an IT help forum could be identified as one worth developing a training course for), or to identify individuals who might benefit from leadership development (e.g., specific managers mentioned in exit interviews).
  • Structural Coding: Identifies the structure or organization of the data itself. You might use structural coding to reveal how decisions are made within an organization by coding who is involved, what information is considered, and the criteria used for making choices.
  • Emotional Coding: Identifies and categorizes the emotions and feelings expressed in the data. You might use emotional coding to identify things like frustration with the workload, feeling undervalued, or excitement about new learning and development initiatives. This approach is well suited for open-ended responses to questions concerning experiences or feelings like engagement and well-being surveys since it allows organizations to identify items that are linked to emotions.
• Sentiment Analysis: Assigns a sentiment score to the data, usually rated on a scale of negative-neutral-positive. However, variations on different emotional scales are possible. You might use sentiment analysis to quickly get a sense if employees felt positive or negative after a big announcement, new initiative, training feedback, etc.
  • Note. I do not consider sentiment analysis a true ‘qualitative’ analysis since it seeks to quantify the sentiment and turns the data into numerical values. I’ve included it here because even though I personally don’t view it as a qualitative analysis, it is an analysis of qualitative data and you will usually hear others talk about this analytical technique whenever they talk about options for assessing qualitative data.
• Thematic Analysis: One of the most common approaches to qualitative people analytics is thematic analysis. This involves coding data and then identifying, analyzing, and interpreting recurring patterns and themes within the data. It requires a thoughtful and systematic examination of your coded data – it’s not just a summary or your own ‘gut feeling’ about what you see in the data! If done well, you can uncover underlying meanings, perspectives, and experiences that aren’t visible at first glance.
  • One of the more powerful applications of thematic analysis in people analytics can come from analyzing what employees say or do regarding topics they may be hesitant to respond to with complete honesty or where employees are cautious in what they share. For example, in exit interviews employees often give reasons for leaving that are most ‘socially acceptable’ rather than most honest (e.g., they may say they are leaving for “family reasons” when really they are miserable about something else). It’s normal for people to do this to be polite, to avoid having a difficult conversation, or to maintain a positive relationship with the company. Many managers and experienced HR professionals know when they hear a response like this and it’s not the ‘real’ reason but it can be hard to pin that information down. Thematic analysis allows you to analyze all themes and separate common ‘socially accepted’ themes from others. It can also help to identify patterns where reasons can be aligned with emotional responses, context, and descriptions associated with those reasons.

Text Analytics

Analytics can be done on any form of qualitative data, but text analytics is by far the most popular in people analytics today. You may have heard this term and wondered if it was different from the list of analytics techniques above. It is not. If you are doing any of the above techniques on text data, you are doing text analytics. The term text analytics simply refers to the analysis of text. So, if you are doing thematic analysis or sentiment analysis and it is of words (as opposed to images, video, sounds, etc.) you are doing text analytics. With the growing use of communication platforms and surveys within organizations, text analysis techniques are becoming increasingly important. The growing use of technology has also meant that sometimes people may be using the term “text analytics” to imply that they are referring specifically to the analysis of text completed by technology tools. Which is an approach to qualitative analytics that has been growing rapidly recently as well.

This introduces a big question: should you analyze your data manually or through the use of automated tools? Everyone is likely to feel a little differently on this topic. But most who have compared the outputs of human qualitative researchers to the outputs of AI-generated and automated qualitative analysis tools have found the automated tools to be great at simple and more quantitatively focused approaches to understanding qualitative data and far inferior to human researchers for more nuanced understanding. If you just want to understand what topics people are talking about most often (which is actually a quantitative analysis since it would be a count of things occurring in the qualitative data), they do very well. This is what all those cutsey wordcloud* generators are doing. You can ask automated tools to do coding and thematic analysis and they will produce very plausible seeming results. But, they tend to be pretty terrible at capturing true human experience, emotion, and the nuance of what is in qualitative data. Today they are still very weak at thematic analysis – despite what the people who work at the companies who build those tools tell you. The better ones have gotten good at tagging certain words and giving general summarization of sentiment (making these tools pretty good at sentiment analysis) and at tagging certain words that tend to occur together, but they are still pretty terrible at understanding how emotions fit together in the broader context of what the person is experiencing or discussing.

Personally, I think the automated tools are fantastic, but as a supplement, not a replacement, for the people analyst. I think there needs to be a human involved who knows how to manually conduct qualitative analyses and therefore can identify when the tools have captured insights accurately and when there are aspects of the data that have not been captured appropriately. These tools are game changers for addressing the issues of time and resource constraints in qualitative analysis because they provide a starting point that minimizes data prep. They can be an amazing efficiency tool for drafting initial frameworks or for assisting in quality checks to help the human analyst identify additional things they may need to look into and analyze. But because the purpose of most qualitative analysis is to get at the heart of the human experience, I believe it still does require a human to understand and interpret the data fully.

So, for those looking to build their people analytics qualitative analysis skills, I recommend building skills to analyze your qualitative data manually and also learning to leverage available technology to help you do so faster and with higher quality (we’ll talk more about leveraging tools and technology later in this chapter). There are a lot of additional methodologies and techniques to do qualitative analytics but if you are just starting on your journey, I recommend starting with coding, thematic analysis, and sentiment analysis. Then consider going broader by building your skills in qualitative research methods. We will talk more about building your research skills, but while we are on the topic of qualitative analytics it’s worth covering a few powerful research methods in this section. 

Qualitative Research Methodologies

Once you’ve built your qualitative analytics skills, you may want to consider learning new ways of analyzing the data by leveraging techniques from specific types of qualitative research methodologies. The following methods take different views on how to think about qualitative data, by looking at the data differently, the results of your analysis will also be unique and may provide additional insights. There are many, but some of the more powerful for people analytics include, Narrative Analysis (examining the structure of the stories that people tell) and Phenomenology (examining the lived experiences of people from their perspectives) both of which are great for understanding employee experience or perceptions of workplace culture from employee experiences. There is also Discourse Analysis (where you identify bigger concepts embedded within the data) that could help you analyze company communications to assess concepts like power dynamics, and alignment to core values, assumptions, or beliefs. Or you could try Grounded Theory where you let theories and information emerge from the data itself, rather than prescribing what you hope to find first. There’s even Conversation Analysis (where you examine the actual spoken interactions), allowing you to see how conversations are structured, how turns are taken, and how meaning is constructed through language and nonverbal cues. This provides complicated but fascinating insights into how organizations operate, how teams work together, how communication happens, and signals of where cohesion is/isn’t present. If you manage to work through all of these, congratulations you are quite the expert and likely have also come across some of the other possible approaches in your travels. 

Qualitative Data Creation

If your people analytics career has you headed toward qualitative analytics, you should keep in mind that while there are often heaps of unused qualitative people analytics data going sadly unused, the available data may not always be useful for a specific question you are trying to answer. If the data creation process was not designed effectively, that information may not be well suited to your purposes and you may still have to make your own. Most people analytics professionals who focus on qualitative analysis, are usually also responsible for the design and implementation of the qualitative data creation process. This makes the skills for qualitative data gathering and creation, especially important. Data creation is important for all people analytics professionals and it is worth calling attention to some of the more qualitative data-specific creation skills you will need:

• Designing effective interviews, open-ended surveys, and focus group questions: This includes ensuring questions are appropriate for the context, will produce valid and reliable responses, are ordered appropriately, do not lead, bias, or confuse respondents, and much more. Don’t make the mistake of thinking that designing effective data collection methods is as simple as coming up with questions that you would like answers to. There are entire fields of study and many, many resources available on how to design surveys, interview scripts, and focus group facilitation methods.

  • Interviewing: Master the art of asking insightful, people-centric questions that go beyond simple “yes or no” answers. This includes active listening, following up, probing deeper for details, and understanding motivations behind employee behaviors, attitudes, and experiences.
  • Open-ended survey question design: Craft survey questions that encourage detailed and thoughtful responses about people’s experiences, perceptions, and suggestions. Avoid leading questions and answer choices that limit the range of perspectives you might capture.
  • Focus Group facilitation: Guide group discussions to ensure all voices are heard and the conversation stays focused on the topic at hand. Your role is to ask prompting questions that encourage open dialogue, manage group dynamics to ensure a respectful exchange of ideas and synthesize key themes that emerge.

• Observation techniques: Skilled observation allows you to gather data by watching and recording employee behaviors, interactions, and processes. This can be particularly useful in understanding workplace dynamics, work processes, communication styles, and company culture.

  • Train yourself to be a meticulous observer, paying close attention to details like body language, facial expressions, and the overall atmosphere during meetings, team interactions, or even informal chats.
  • Develop a system for recording your observations, whether through detailed notes, audio recordings (with proper consent), or even video recordings of specific work processes (with proper consent and anonymization).
  • Consider both direct observation techniques – ones in which you are the person doing the observing – and indirect techniques. For example, instead of direct observation, you might choose to use an observation method known as a diary study where participants record their experiences for you to later analyze.

• Building Rapport: Establishing trust and connection with the employees you will gather information from is critical. People are more likely to share honest and detailed information if they feel comfortable, respected, and heard.

  • Present yourself with a genuine and approachable demeanor.
  • Actively listen and show interest in what people have to say about their experiences.
  • Maintain confidentiality and assure participants that their responses will be used ethically to improve the workplace experience for everyone.

_{* [Rant. While commonly used to give a visual anchor for qualitative storytelling, a wordcoud is in my opinion the worst possible visual choice you could make to represent your qualitative analysis. First, it’s actually quantitative since it is just a visual display of word frequencies. Second, and more importantly, it is antithetical to the purpose of qualitative analysis. Qualitative is meant to show the quality, experience, emotion, and nuanced detail. The only time I support a wordcloud as a visual for qualitative analysis is when it is manually designed to emphasize words based on the conceptual strength of themes, not from a quantitative measure. If you feel you must use one because you want the attention-drawing ability of that kind of visual, manipulate it until the end visual represents the qualitative data appropriately.]}