Excel Charts: If It Isn’t Broken…

I humbly accept our business visualization reality: 90% of all business charts are created in Excel and most business users are unwilling to learn yet another application and go through all sorts of hassles (data management, compatibility issues, file sharing) just because a different product offers a bigger chart gallery and some obscure defaults no one cares about (except the experts).

You don’t take driving lessons in a Ferrari, specially if you don’t know the difference between a Ferrari and a Fiat. Why would you want Tableau or Spotfire if all you want and need is a pie chart?

If we want to make users more curious about the way they use their data, if we want to persuade them that 3D charts are useless, if we want them to adopt what we believe to be information visualization best practices, we must start with the application the users are familiar with.

It would be great to have a better tool in every computer, but we don’t have it. What we can do is to undermine it from inside, by educating the users and showing them how mediocre the tool is.

If done right, this should be visible in market surveys, sooner or later. And Microsoft will react to that, hopefully. There is no other way.

Leaving the Excel Flatland

That said, we can agree that Excel’s chart gallery is notoriously poor, defaults and options a nightmare, and the implementation of advanced visualization techniques a pain in the ass. And while you can argue that most new chart formats are outperformed by traditional ones (and a large majority is) you cannot assume that the plain ol’ bar chart will be here forever. We are using XVIII century visualization tools to display XXI century data. Things change very slowly indeed.

We cannot understand information visualization if our worldview is framed by the Excel chart gallery. Yes, we can recreate in Excel some formats that we see in other applications, but something is always lost in translation. Like in this song:

Brilha, brilha lá no céu
a estrelinha que nasceu.

You may know it as:

Twinkle, twinkle little star
how I wonder what you are.

The music is the same, but the user experience is much different (specially if the user doesn’t know Portuguese…).


Kelly O’Day over at ProcessTrends if offering the Learn R Toolkit, target at Excel users. I’ve just bought it. Why?

  • Because I like the idea;
  • Because I’m curious about creating charts using a programming language;
  • Because it’s not easy to create this in Excel, no matter how hard Jon tries :).

Now, let me be frank. I don’t believe that regular Excel users would want to use R. If recording simple macros are outside their comfort zone, an application that requires programming even to create a basic chart is clearly out of limits.

R seems to require a lot of programming to create more advanced charts. Surely you can create a bar chart with three lines of code, but that’s the equivalent of

print "Hello World!"

Not very useful, but I’m sure Kelly’s Toolkit will encourage me to explore and break some handcuffs.

I don’t want to write a blog about Excel charts. I just want to write about making sense of business data. The tool shouldn’t matter (but it does).


10 thoughts on “Excel Charts: If It Isn’t Broken…”

  1. I think you’ll be pleasantly surprised at how little code it does take to do charting in R. Certainly there is more substantial code when you want to generalize from one data set to a dashboard like solution that will intelligently produce ranges etc.

    That said I’ve produced graphics for publication in 5-10 lines.

  2. While I agree with your conclusion that most business users will only ever use Excel for charting (visualization), I disagree with the general sentiment and tone of your post. But perhaps we have a different definition for the term “business user.”

    From my perspective, the vast majority of “business users” are folks who are trying to do just that: run a business. They have a hard enough time just getting the data they need to make a decision let alone acquire and learn other tools to make a chart that “looks better.” In my experience, those same business users have relatively little control over the tools at their disposal to capture the data and produce charts. For this group, Excel is the de facto tool sitting on their desktop and likely one they have been using for years. Heck, in a progressive company, they might even get training on how to use it! 🙂 The challenge with this audience: how do we best educate and enable them to make business decisions more quickly and effectively with tools readily at their disposal?

    On the other hand, if by “business user” you mean folks like data/report analysts, then maybe moving to a tool like R Toolkit makes a lot more sense. More than likely, people in this category are more inclined to be able to learn a programming language regardless of its simplicity. Since the audience is much more limited, the ability to acquire and learn an additional tool is much more likely. The challenge with this audience: helping them understand how their lives will be easier utilizing the new toolset…will they be able to produce better results with less effort? If so, the learning curve (and battles with the local IT folks that want to lock down the environment) will be worth the energy they need to spend.

    Once upon a time, I was a developer and now find myself more on the business end of the “business user” world. I’ve seen some of the R Toolkit scripts and I start to glaze over. Although I could likely learn it relatively quickly, I’m not convinced it’s worth the additional energy yet. In the interim, I’m learning how to do a better job with the journeyman tool of Excel and doing everything I can to push the business forward.

    Thanks for the post. It was really good to have something for my brain to chew on at the end of the day. Thought provoking.

  3. Michael: My underlying conclusions are these:
    – for most business visualization needs, Excel is good enough;
    – for beginners, the problem is the low literacy rates, not the tool;
    – training is focus on the tool, not on the task, and that’s wrong;
    – the only way to force vendors to create better tools is to improve the literacy rates of current and new users;
    – managers should be aware that a single tool will harm data analysis and visualization (and the business) and they should give more advanced users access to more advanced tools.

  4. “for beginners, the problem is the low literacy rates, not the tool”

    I wouldn’t say low literacy rates, as such. I’d attribute the difficulties to lack of documentation, and lack of self-documenting tools. This self documentation is where the BonaVista Chart Tamer shames Excel’s chart creation interface.

  5. Jorge – I think we agree! Honestly, I didn’t quite get those thoughts from the original post. Then again, I’m reading at the end of long days…maybe I’m tired.

  6. If we’re trying to coax business users to try software having more readily accessible graphical and statistical firepower than Excel (sorry Jon), I would suggest Minitab as a more generally viable choice than R. Admittedly, R is a free package with tremendous capability, but the activation energy of learning its programming language when someone just wants to make good quantitative — maybe even statistically enlightened — graphs, will convince a lot of people to stick with Excel. By contrast, if one can figure out the Excel menu, one can probably figure out the Minitab menu.

    BTW, would starting a war over statistics packages be grounds for banning someone from an Excel forum? ☺

  7. @Dale: Most people shouldn’t leave Excel (or their spreadsheet of choice, not necessarily Excel) to create a chart, because Excel is capable enough. They don’t need Minitab, or SAS or R, not even Tableau. They do need to know something more about data visualization and how better practices impact their current data analysis. As I say above, power users should have access to more powerful tools, if needed. That would encourage diversity of thinking about the data. If your competitor is using Excel+Spotfire it may have a small competitive advantage [note to self: write a post about this].

    Wars are welcome here, as long as they remain… civil 🙂

  8. In a previous life, I was involved in corporate R&D. Being in R&D we had need for extensive data analysis; being in Corporate we had access to corporate tools, i.e., Excel. We made do.

    For a while we used a third party package called DOE KISS, which ran in Excel, and basically gave some enhanced calculational power and a better interface to graphical tools. When we upgraded to Excel 5, that software didn’t work, and the upgrade was not approved.

    Then we got Minitab, which I’d used in a line command version a generation earlier in grad school. It made quick work of the statistical analysis, but the outputs were tricky to make use of, and I thought their graphics engine was horrid. I had much better control over Excel’s charting output than over Minitab’s. I would use Minitab for part of my analysis, then copy-paste into Excel for my reporting and charting.

    Eventually I got tired of switching back and forth, so I wrote myself routines and built templates so I could do the same analyses in Excel.

    Having two packages is not convenient. It’s a pain to switch, assuming you have access to both and know how to use both. And everyone had Excel, while you had to get about three levels of management to approve your use of Minitab. The Minitab interface was not bad, but it was different enough that some people never got over the learning barrier.

    As Jorge points out, Excel is “good enough” for 99.9% of its users. What businessman needs multiple ANOVA? or DOE or SPC? And if you do need these capabilities, there is a thriving market of third party add-ins for Excel, which provide this capability in the same application. Fifteen years ago, I was saying that Minitab should create an Excel add-in version. They never did, but if you want a recommendation for a decent add-in package, contact me.

  9. @Jo* Excel is wonderful: a veritable Swiss army knife for dealing with numbers. I sincerely appreciate those who share their expertise at how to use Excel better.

    Still, the position that “Excel is ‘good enough’ for 99.9% of its users” or “having two packages is not convenient – it’s a pain to switch” strikes me as the viewpoint of someone who has been wielding a hammer so long and so well that now everything in the world resembles a nail.

    For many jobs where numbers are important — for example, serious statistics, large database management, solving systems of stiff differential equations, many specialized tasks in engineering & science — there are better if usually more expensive tools available than Excel. Third-party add-ins do help close the gap, but in my experience they are less capable that the dedicated best-in-class stand-alone software tools. [Jon, I provided you an example by e-mail of that add-in gap last week in the SPC arena. And, in my present life, I can’t tell you how many R&D colleagues I’ve seen set up basic SPC charts incorrectly because they used native Excel rather than an SPC add-in or a real statistics app that could help teach them about SPC or cover for what they thought they knew but didn’t.]

    Are these business tools? Well, yes, in the broad sense: [some!] businesses buy licenses to such tools because they believe this makes their employees [far more that 0.1% of them!] more productive at doing certain tasks for the business. As such a business-user-by-day in this inclusive sense — and computing hobbyist-after-hours who can’t afford Minitab but is listening to figure out if R would be worth its learning curve — I honestly believe Excel can’t compete with Minitab (and others) for making powerful statistics available to data analysts, or with MathCad (and others) for solving engineering equations, or with Aspen for modeling a chemical process. I say that as someone who usually figures out ways to solve the easy-to-moderate problems using Excel. When my ability to wield Excel and/or my knowledge of the subject domain isn’t advanced enough to solve the tougher problems at work, I eventually turn to such “power apps” that owe their commercial existance to being able to do certain things better than spreadsheets, and I sometimes figure out how to bring the needed functionality back to Excel later (if it’s a problem that needs to be solved repeatedly).

    Spreadsheets are wonderful — but they aren’t the only tool I want in my toolkit for visualizing and understanding data. Let’s keep Excel as a key tool for dealing with numbers, but not let it become a pair of handcuffs.

Comments are closed.