24 Data Visualization

One of the most sought-after skills in any career that uses information is visualization. Data visualization helps us simplify complex data sets, tell a story, and can even be used as a tool for discovery and analysis. Data visualization is, as the name would suggest, the visual display of data. It’s a critical skill for working with information because we humans are specially designed to take in and process information visually. Our brains have evolved to be particularly awesome at seeing patterns in visual things. We learned that certain patterns on animals equal danger, we learned that colors can let us know when food is good to eat, and we got really good at judging heights and distances visually to manage navigating the world. You show us something visually and our brains instinctively look for meaning in it. Because of this, when you want to share data findings with others or find something in the data yourself, data visualization is going to be one of the more powerful skill sets you can build.

Data Visualization for Storytelling

We already talked about the importance of data-storytelling in a people analytics career in the non-technical skills section and data visualization is one of the best tools we have for doing data-driven storytelling. Think of being a little kid, wouldn’t you want a storybook that comes with pictures? Our people data story is no different. The visuals we provide can help stories be stronger, clearer, and just more exciting and engaging. A well-designed chart or graph can instantly communicate complex workforce dynamics in a way that text often can’t. Visuals can help capture leadership’s or employee attention, they can highlight key findings, and they can make your story more memorable and impactful.

In addition to all these great things, what I really like about data visualizations is how they allow you to share information faster than using words alone. For example, it can take quite a few words to explain how trends in employee satisfaction scores have changed over time but a line chart tells that story in a matter of seconds. This leaves more time to discuss what happened and what to do about it. I don’t know about you, but I would much rather spend time designing ways to improve the employee experience than trying to explain the numbers.

I also like that data visualizations have an ability to spark curiosity and encourage deeper exploration of people data. Back in my day (yes, I am “old” enough in the people analytics field to say that), there was limited interest in people analytics. The term ‘people analytics’ didn’t even exist yet, so few people were aware of it and even fewer were asking for it. But, a data visualization could draw interest and almost always drew questions and requests to know more. For instance, employee attrition is often a relatively well known topic and calculated statistic in an organization. So simply stating that a department had an increase in attrition didn’t really capture much attention. The response might be, “yeah, we know, Jeremiah left last week and Preeti left last month.” But, when that department’s spike in attrition is shown on a heatmap compared to other departments, it might prompt leaders to ask questions about the cause. The comments then change to questions like, “is our attrition really that much higher than everyone else? what is going on in our department?” which leads to further investigation and potential solutions. By incorporating compelling visuals into your stories, you can engage others in conversations about people analytics insights that are more meaningful and action oriented.

Data visualization is not just for quantitative data. Because qualitative data is still less frequently used, visuals can be even more important to telling your qualitative data story. They can help your audience understand the key findings and even help give some credibility and support to your analysis process. In the world of people analytics, you might use visuals like mind maps to show the structure or relationship between analyzed codes and themes (e.g., different types of employee concerns). You might visually display areas of contrasting findings. For example, let’s say you assessed the effectiveness of a training program by interviewing employees about their ability to apply the trained skills into their work. A heatmap could show each skill colored-coded by a subjective assessment, highlighting which skills the training taught effectively and which need to be improved in future offerings of the training. Other times, you don’t just find themes, groupings or differences. Sometimes you find flows, structure, and ordering. Luckily, visualizations are fantastic at representing processes and structure. You can use visuals to show the order of more intangible events and things that tend to follow a pattern in your organization. For example, you could display commonly occurring career pathways using process or flow diagrams even using the visually to help showcase things harder to quantitatively measure such as transferable skills that lead to career flexibility. Remember that qualitative data isn’t limited to things like employee surveys or interviews and the visuals don’t need to be limited either. A fantastic qualitative analysis and corresponding visual might come from observation or diary studies that are analyzed and then put into a visual to show the process and flow for how employees experience a day of interacting with work systems, or the employee lifecycle at a company. 

Sometimes the qualitative data itself can serve as great material for adding a human element to your work. For example, if you display a quantitative chart to highlight the high employee engagement rate on a graph, you might also choose to overlay a textbox on the chart with an employee quote saying what they love about the company. It’s a powerful way to emphasize the human experience and add the qualitative story visually along with the quantitative data story.

Finally, when telling your story, keep in mind that it is a story and is therefore meant to be an experience. Don’t limit yourself. You can get as creative as you like and use any format of information. You aren’t limited to only numerical, text or process based data, you can make maps for non-text data (e.g., grouping images together, or playing video clips together in a media compilation). If you work in a physical space, you could even use physical objects. Who knows, maybe when you need to do that presentation including a statistic about the troubling 10% rate of safety incidents on the job sites, you’ll skip the powerpoint chart and instead choose to line up 9 perfectly fine safety helmets followed by one horribly damaged helmet on the table in front of you as you stress the need for better safety training and compliance. 

Data Visualization for Reporting & Sharing

Data storytelling is powerful, but you aren’t going to need to do it every time you have analytics to share. Not every piece of information needs a presentation or story. Sometimes the information just needs to get distilled to insights and then get to others as effectively, clearly, and quickly as possible. Visualization is still key for this. 

Did you know that reports, dashboards, scorecards, and any other mechanism you use to share data are all examples of data visualizations? For example, let’s say you are asked to provide a regular report on absenteeism. First, you have to decide what you will provide: will absenteeism be shown as a count, a ratio, a percentage? Will you give just an overall number, or will you display the values for each department or location separately? Will it be for one period of time, or over multiple periods of time? These questions form the basis for the report, but there are still questions about how you will provide it. Will the values all be provided together in a single table? Will they be shown in a chart occurring over time? Will it need formatting like bolding of key metrics, or color to highlight positive or negative trends? Will it need clear and concise labels to help people more quickly scan and understand the report? Will the report be static or does it need to be hosted on a website with animation or an ability to switch between different subsets of data? Is there an absenteeism level the company deems ‘acceptable’ that will need to be incorporated as a reference so leaders can interpret the current rate in relation to the acceptable rate?

Effective sharing of data requires thoughtful visual design principles if it is going to provide businesses the information needed to navigate complex information and make decisions. This applies whether we are talking about static reporting in the form of manually created scorecards, reports, or presentations. And it applies even to the reports that we “automatically” generate out of information systems (e.g., the ones produced by an HR Information System). If you are going into a data system and selecting the values, calculations, and layout of the report to be generated, you are making visualization choices. And, it also applies to even the most complex business intelligence system interfaces. If the goal is to present information, data visualization is necessary to achieve that goal.

Every decision made about what to show and how to show it is a data visualization decision. Even whether to create a visual or not is a data visualization decision. Which brings up an important point, not all data needs to be visualized. What if it turned out that the person asking for that absenteeism report really only needed to know the current rate to answer a simple question and they were unlikely to be asking for that information again in the foreseeable future? Would it be necessary to make a report at all? Maybe it would be best to just tell them the rate, saving you the time of making a report and helping them answer their question quicker. It’s also true that there are just some things that are not going to be best represented visually. Sometimes it is better to use words or numbers to describe what you need to share and sometimes  the context or situation isn’t appropriate for including visualizations. Use visualizations, but use them when they support and enhance your goals.

Speaking of things not everyone realizes about data visualization, not all data visualizations are charts, graphs or diagrams. Sometimes the best data visualizations come from applying visual design principles to things like words, images, or a even just a single numerical value. A text-based report with a single metric in a font 4 times larger than the rest of the text can have tremendous impact. A key theme from a qualitative analysis bolded for emphasis can draw attention where needed. Even the placement of items on a screen or page has an impact on how effectively others understand your findings. Data visualization is about helping your audience see what is important to be seen. 

Data Visualization for Analysis & Discovery

Visualization isn’t just for sharing information with other people. In fact, data visualization is one of my favorite techniques for doing analysis. It’s one of the first things I do when starting to analyze a new data set. Some things are too difficult or impossible to see in raw data or through statistical approaches, yet they seem to magically ‘pop out’ with the right visual. And, some things are just easier to see visually. Visuals can be a quick way to spot outliers, anomalies, and data issues. Often discrepancies, patterns, or groupings visible in an image informs the entire approach of how I clean, analyze, or interpret data.

You can use visualization to conduct exploratory analyses or directly answer some of the same questions you might also approach using descriptive, diagnostic, predictive and prescriptive analytics. Take for example, a manager wanting a detailed understanding of performance ratings for the employees that they supervise. Statistics like average, minimum, or maximum may give a less intuitively helpful description than the manager would get by looking at the distribution or a frequency chart of employees allowing them to quickly see how many employees are rated high or low performers; bar charts, histograms, and pie charts are helpful for doing this quickly. Creating scatterplots, histograms, or boxplots could also give a quick visualization of the distribution, spread, and relationship of ratings among the employees. Visuals could show comparisons of performance distributions across job types, experience levels, or demographics to identify variations and address any group-specific concerns. A line chart could show performance changes over time. Scatterplots are an effective way to identify possible relationships between factors. For example, the manager may wish to know whether performance ratings connect with other factors like tenure, training completion, or specific skills. A scatterplot showing what appears to be a straight line of data points with performance on one axis and training completion on the other, may highlight a possible correlation in which more training appears to be related with higher performance. Boxplots can make spotting outliers a breeze – something extremely difficult to see when only looking at raw data. And, scatterplots can show those outliers in relation to other variables, providing an opportunity to assess whether the outlier is problematic or  an important unique piece of data. For example, if the manager spotted an outlier they could look at it to see if it was a data entry error that needed to be fixed or if there is an exceptionally well performing employee that deserves to be recognized and celebrated. With a mix of visuals the manager could get a holistic sense of how performance is occurring, how it is related to various aspects, and where they might want to explore the data further. Data visualization tools that allow for quick creation of visuals (e.g., Tableau, PowerBI, Excel) make the creation of these types of visuals sometimes as fast as or faster than calculating statistics making data visualization a great tool for analysis not just for creating visuals to share.

Data Visualization for Explaining your Qualitative People Analysis Process

Visualizations are powerful because of how well they help us explain complex things to others who may not be as familiar with the information we wish to share. This makes visualization a handy skill that you can also leverage when you need to explain something complex or uncommon. This can be useful for displaying employee workflows, talent management processes, workforce planning scenarios and so much more. But I would like to call special attention to your opportunity as a people analytics professional to use visualization to help you explain, share, and add credibility to your qualitative analysis process. In my experience, most people don’t typically want to hear the full process of how you got to your people analytics quantitative insights – in fact, I would strongly urge you not to share your quantitative process in detail and to stick to the findings (more on that in Step 4 where we talk about applying your skills). However, I encourage the opposite when it comes to qualitative analysis. Because most people have  never tried to do qualitative analysis and may have never been presented with qualitative analysis findings, they don’t really know how to understand them. When you pair that with the erroneous misconception that qualitative data is not rigorous, there tends to be an issue with your audiences ability to understand or to believe in the strength of your findings. To fix that problem, your audience needs to be shown that the process you took in your analyses are rigorous enough so they can trust the results. Unfortunately, explaining qualitative analytics techniques is extraordinarily difficult. Instead of trying to force your audience to sit through a lecture about research methodology, I encourage you to “show your work” for qualitative analysis by using visuals showing the steps of the process. 

Show them how you took a massive and messy unstructured data set and transformed it into the your key takeaways and insights. For example, you might provide an image showing an anonymized subset of interview transcript text that has different selections of the text color-coded to indicate your data coding work, and you may include noticeable lines or symbols connecting the text to themes. In general, data visualizations should be as simple and clean as possible, but I personally think this is another area where it might be to your advantage to break the rules a little bit. I think you should feel free to overwhelm your audience a little bit here, help them understand just how much information you have sifted through, show them the effort and rigor, give them a sense of the systematic and intentional way that results were formed. Just don’t make them look at all of your data, give an example (e.g., a screen shot, or anonymized sample subset) and make sure that you are not sharing anything private or confidential. Any visual that can help give your audience a sense of the process that displays the rigor of analysis will be powerful evidence that your results are true analytics not just an opinion.