Organizations often ask me how to hire people who are able to create truly useful charts, and this post summarizes my (current) answer to that question. While this post is written as a guide for employers, it can, of course, also act as a guide for those who want to become data visualization professionals themselves, and work for the organizations that so urgently need those skills.
I also feel that a post like this is needed since there’s a lot of confusion around exactly what skills a candidate needs to possess in order to create useful charts, and this confusion can be seen in the job postings that organizations post for data visualization roles. Most such postings read like job descriptions for expert software users (Excel, Tableau, Qlik, JMP, etc.), data analysts, or graphic designers but, as I suggested in this recent post, those skills aren’t sufficient to allow someone to reliably create useful charts, and organizations that place people with only those skills into data visualization roles are usually disappointed with the results.
As the list below suggests, a surprisingly broad variety of skills are required to create truly useful charts, including elements of graphic design, data analysis and software product expertise, but also many other skills and types of knowledge. If you can think of other things that should be added to this list, please let me know in the comments and I’ll update this post as needed.
Essential skills
Data communication skills
This is the core skillset that you’re looking for, however, it’s also the hardest to measure, or even describe. “Data communication skills” is an umbrella term for a set of loosely related skills that include knowing how to select the right chart type and arrangement for a given situation and audience, how to figure out what data needs to be included in a chart in the first place, and many others (the Practical Charts course description contains a more detailed list). Unfortunately, I’m not aware of any widely-recognized standardized tests to evaluate these skills, so the best indicators of competence are typically whether or not a candidate has taken courses or read books from respected experts like Stephen Few, Alberto Cairo, Colin Ware, and Edward Tufte.
Domain knowledge
If you’re an insurance company, the person who’s designing your charts better have a fairly good understanding of insurance. This might sound a little surprising to some readers since data is data and charts are charts, right? What difference does it make to the person who’s creating charts whether the data is about insurance or cricket?
As anyone who’s taken my Practical Charts course knows, creating useful charts requires a surprisingly deep understanding of why the audience needed to see that data in the first place, and exactly what the audience will be using the charts for. Obviously, that’s not possible if the chart designer doesn’t really understand what the audience does and the environment in which they operate. Key data viz best practices such as adding informative callouts, visually highlighting the most important part of a chart, or adding meaningful comparison values for context can’t be followed if the chart designer has only a superficial understanding of the audience’s work and the environment in which they do that work. This also means that organizations should be wary of candidates who claim to be able to create useful, effective charts for any kind of data from any domain. IMHO, this suggests an incomplete understanding of what it takes to create truly useful charts.
Candidates don’t have to be domain experts, but your new hire should have a pretty good understanding of how organizations like yours work, much in the same way that a copy writer who’s writing a report or marketing copy for you has to know quite a bit about your organization and its environment in order to produce useful copy.
After your new hire starts, this domain knowledge will help them to get up to speed quickly on the specific, current challenges of your organization, an understanding of which is essential for them to be able to create charts that will be useful to decision-makers. To help them get up to speed (and be more useful to you), invite them to listen in on planning and strategy meetings early on so that they know what information decision-makers really need; don’t wait to involve them until all of the big decisions have been made.
Data visualization software expertise
Obviously, your new hire has to be reasonably competent at Excel, Tableau, JMP, Qlik, or whatever other data visualization tool you want them to use. This is generally pretty easy to assess by verifying that any training certifications that they provide are valid and credible, or by having them present a portfolio of visualizations that required more than just basic product knowledge to create.
Unfortunately, many job postings for data visualization roles focus almost entirely on these skills but, as this post suggests, software product knowledge is a relatively small component of what’s needed to create truly useful charts.
If your organization is creating custom, complex, or highly specialized charts, your new hire might have to have very deep technical knowledge of these tools, or even knowledge of custom data visualization code libraries. Most organizations, however, don’t need nearly that level of technical depth, and passing some intermediate product certifications will be enough. At that point, the other skills in this post start to eclipse technical product knowledge.
Knowledge of basic statistical concepts
This is an area where people who have primarily graphic design or software product backgrounds often fall short. Your new hire should understand the difference between means, medians and modes, how to correctly calculate percentage change between time periods, and be able to perform other basic statistical calculations. Ideally, they should be familiar with concepts such as quartiles, percentiles, and sampling. Without these skills, there’s a high risk that your new hire will accidentally create charts that misrepresent reality.
Almost essential skills
While not absolutely required, candidates who lack these skills should be given a lower priority in hiring decisions.
Statistical reasoning skills
These should really be under “Essential skills” but it’s so hard to find people who can reliably avoid conflating causation with correlation, notice survivor bias in data, and not step onto other common statistical reasoning landmines that I demoted these to “Almost essential skills”.
Notice that I didn’t call this category “Advanced knowledge of statistical techniques” since your new data visualization pro probably doesn’t need to know how to perform complex regression or distribution analyses. They should, however, at least know about the landmines so that they don’t inadvertently plant them in your charts. Stephen Few’s The Data Loom and Charles Wheelan’s Naked Statistics are excellent places to start developing solid statistical reasoning skills.
Data cleansing experience
In almost all cases, your organization’s data will need to be cleaned up before it can be visualized and it’s not unusual for data visualization professionals to spend the majority of their time just doing that. This means that your new hire is going to be a lot more productive if they can do that work themselves instead of relying on others, and knowing a bit of SQL or other querying/scripting languages could be really valuable. There’s also a major side benefit of this skill which is that, if the person who will be creating the charts is the same person who cleaned up the data, that person will have a deeper understanding of that data, in particular, any problems or limitation inherent within it. They can then ensure that these are noted in the charts that decision-makers see.
Ability to write clearly
It might be surprising to see this in a list of data visualization skills since charts usually contain relatively little actual prose. I’ve included it here, though, because being a poor writer suggests that a candidate lacks fundamental skills that are necessary to communicate information clearly in any medium, including charts. It's not a coincidence that virtually all of the competent data visualization pros that I know are also very good writers, and often great presenters, as well. If a CV isn’t well organized, doesn’t flow, or contains awkward or disjointed sentences, I’d probably set it aside even if it checks all of the other boxes in this list.
Note that I’m talking about the ability to write clearly here, not the ability to write correctly. While most people who can write clearly have reasonable spelling and grammar, some don’t. That shouldn’t be held against them in a hiring process, though, especially for candidates for whom English (or whatever language in which they’ll need to work) isn’t their first language.
Indirect indicators of data visualization competence
While not absolutely required, candidates who have the following skills or experience should be given “bonus points” in a candidate selection process.
Product design/product management experience
It’s only in retrospect that I’ve realized just how beneficial the years that I spent in product design roles have been to my data visualization work. Understanding that most (though not all) charts are products, and that what ultimately matters is how well they serve the audience’s/customer’s needs (as opposed to “how well they show the data”) requires a shift in thinking that experienced product people have already made. Having deeply assimilated concepts like user intent, the curse of knowledge, and user-centric design are huge advantages when creating charts.
(Before anyone panics, I’m not suggesting that charts should just show the audience what they want to see, or that deceptive charts are OK. What I am saying is that many charts that “show the data well” are also useless in the context of the target audience’s current needs, and someone with product design experience would be able to recognize that.)
User research experience
Candidates who’ve conducted focus groups, usability testing, or other forms of user/customer research have a big advantage when creating charts. Having this experience increases the chances that your new hire will be a good listener, be able to set aside their own perceptions and expectations, and “get inside the heads” of decision-makers to figure out what they really need. A background in anthropology can even be helpful.
Participation in the data visualization community
Following the work of people such as Andy Cotgreave, Alberto Cairo or other data visualization thought leaders is obviously a good sign, as is participation in data visualization discussion forums, conferences, competitions, and events, such as Makeover Mondays.
Graphic design training
Organizations often focus a lot on graphic design skills in job postings and interviews, but those skills are only a high priority if your new hire will be creating charts that are designed primarily to get attention (e.g., infographics), as opposed to being designed to communicate information as clearly and quickly as possible. Expert graphic design skills are a plus, but an intermediate grasp of color selection, font selection, decluttering, etc. is sufficient to design most charts that will be used inside of organizations on a day-to-day basis.
An interest in cognitive biases, decision-making, and the psychology of visual perception
Most (though not all) charts are created to support better decision-making, so an interest in the work of Daniel Kahneman, Dan Ariely, Shane Parrish, the Heath brothers or others who write about decision-making is a good sign. An interest in findings from research in the fields of visual perception and human memory is also a plus, though not absolutely necessary since these findings are (or, at least, should be) incorporated in the “Data Communication skills” mentioned earlier.
Final thoughts
Another common misconception that organizations have is that data visualization roles can be performed successfully by relatively junior candidates. As the list above suggests, though, it takes time to build up the surprisingly large repertoire of skills needed to create truly useful charts, so candidates with relatively little experience are unlikely to be successful. A junior person could be brought on board, but only if someone with the above skills is available to mentor them for some time.
What do you think of this list? Did I miss anything? If so, please pipe up in the comments!
By the way…
If you’re interested in attending my Practical Charts or Practical Dashboards course, here’s a list of my upcoming open-registration workshops.