why don't you capitalize your graph titles?

Through virtual and in-person workshops around the globe, we have taught tens of thousands of people how to communicate effectively with data. This series captures some of the noteworthy questions we hear during those sessions—and our answers.

In school, I was taught that you should center-align and capitalize the first letters of words in titles. I’ve noticed, though, that storytelling with data charts only capitalize the first word in the chart title, use ALL CAPS for the axis titles, and don’t center-align anything. Why?

This is a great question! There are good reasons why many tools use title-case, center-aligned text as the defaults for headers and titles—these formatting choices help that text stand out.

• Most written text is left-aligned, because it’s easiest to read. Center-aligned text tends to grab a reader’s attention due to its contrast with other text on the page. 

• Capitalizing every word in a phrase helps set it off from surrounding text, and makes it seem more formal and important.

In this bar chart, the header and the axis titles are following common default settings. The text is centered with each word’s first letter capitalized (“title case”), and the axis titles are in bold.

In graphs for business communications, though, the title shouldn’t be the element that stands out the most. It will certainly provide important context for the viewer, but data and insights should be the stars of the show. Some slight tweaks to these default settings can make the title a stronger supporting player, without stealing focus from the key elements of a communication.

Left-align graph titles for visual framing

Center-aligned text helps a reader to see that a new chapter, section, or page of text is about to begin…but a graph is already easily distinguishable from any nearby text, and doesn’t need center-aligned titles to emphasize that difference. 

Left-aligning chart titles creates a visual frame around the graph, without requiring us to draw any additional lines. Putting the title in the top left also ensures that a viewer sees that important context first.

Left-aligning the title creates a visual frame around your chart, without the need to draw in additional borders.

Use sentence case for clarity and speed of comprehension

Title case does a great job of setting text apart from any surrounding words…but in our graphs, the title sits on its own, and doesn’t need that kind of distinction. Sentence case—in which one capitalizes only the first word of our titles—offers several advantages:

• It is easier and faster to read, especially in a longer block of text. It Should Be Clear In an Instant That It Is Not Easy or Pleasant to Read the Text in This Particular Sentence When Every Word Is Capitalized.

• Sentence case feels less formal and more approachable than title case, which can be beneficial in those situations where your audience might be unfamiliar with the data you are presenting (or with data in general).

• Proper nouns are hard to pick out when we use title case; sentence case makes them easier to see.

Use sentence case in your chart titles to make them easier to read, and to make the visual feel more inviting for an audience.

Make axis titles easy to find and easy to ignore

Axis titles provide important information to a viewer of the graph, but they shouldn’t be distracting. Instead, they should be easily discoverable, while at the same time can fade into the background. They’re part of the skeleton of the graph—providing structure, but letting the data itself take center stage.

Axis titles provide important context, but they shouldn’t stand out or draw attention away from the data.

There are several ways to achieve this “easy to find, easy to ignore” balance in axis titles:

• Set the text to a neutral, but readable, gray color;

• Use ALL CAPS, so the title’s outline is a regular rectangle, rather than a jagged mix of “ascenders and descenders” (the parts of lowercase letters that stick up above the average height or drop below the baseline of the font); and

• Align them to the top of the y-axis or the left of the x-axis so that they can always be found easily, and so that they also create implicit (rather than actual) borders around our graph.

Some techniques for making axis titles useful without being overwhelming include using a neutral text color, writing the titles in all caps, and aligning the titles to the top of the y-axis and the left of the x-axis.

Keep in mind, of course, that these choices are not right or wrong; they merely reflect our data storytellers’ styles, preferences, and priorities. In business, it’s important to maximize understanding and clarity in charts, and these choices are all about accomplishing that most effectively. If your organization has an established style guide or follows other existing guidelines (e.g., AP, Chicago, APA), then stick with those conventions. Otherwise, consider the impression you want to make on your audience, and make your own capitalization and alignment choices accordingly.

A few small changes to default settings for chart titles and axis titles can make visualizations cleaner and more approachable, while at the same time making the data and the insights easier to understand.

how to do it in Excel: adjusting bar width

Today’s post is a tactical one: how to adjust the widths of bar charts in Excel (and why you should). 

Before we get into the step-by-step, I should mention that there aren’t any strict rules for optimal spacing between bars. Rather, it’s personal preference similar to wearing white after Labor Day (in the U.S., that’s the first weekend in September). As a resident of the muggy Southeast, I’ll be rocking white until fall temperatures arrive in mid-October. However, if you live in cooler climes and consider Labor Day the symbolic end of summer, your preference might be to say sayonara to white until Memorial Day. 

The same gray area goes for optimal spacing between bars. The actual width is not set in stone. Our goal is to enable our audiences to compare the lengths of the bars (instead of the area between them), so general guidance is to thicken the bars to minimize the surrounding white space.

Let’s turn now to how to accomplish this in Excel. In the spirit of Labor Day, I’ll use some data from the Bureau of Labor Statistics (BLS) showing the top ten occupations in the U.S. as of May 2020. 

Compare the bar spacing in the two visuals shown below:

On the left, the gaps are attention-grabbing and create an unnecessary shimmer to the visual. The adjusted version puts emphasis on the length of the bars.  Download the Excel file to following along with these steps to manually adjust:

1. Highlight all the bars, right-click and choose Format Data Series:

2. In the Format Data Series menu, under Series Options, adjust the Gap Width dialog box:

The result is this:

Another benefit of doing this is that now there’s enough space to pull the long data labels into the ends of the bars. This is just one of the decluttering steps we can take to reduce perceived cognitive burden. Here’s how to achieve this:

3. Click on any data label to highlight them all, then right-click and choose Format Data Labels:

4. In the Format Data Labels menu, select Label Options, and in the Label Positions section, choose Inside End. (While you’re at it, in the Label Contains section, uncheck “Show Leader Lines.” These are almost never necessary.)

My graph renders like this, due to my color scheme. I’ll adjust the font color so that the labels have sufficient contrast against the dark blue bars. (TIP: you can use the online WebAIM contrast checker to see if your text is sufficiently readable against your background color.)

5. To adjust the font color, click to select all the labels, choose the Font options dropdown arrow, and then select a different hue (you can also do this in the Format Data Labels menu if you still have it open):

The final visual looks like this:

I might even choose to further format the numbers by displaying the units in millions.

Just as the modern-day guidelines for wearing white after Labor Day are subjective, so too are the rules for the exact spacing between bars. As the designer of your own graphs, experience and personal preference will help you find your own “Goldilocks” of spacing: too thin, too thick or just right.

More Excel how-to’s:

• To depict a range of values, add a shaded band

• To create a frame of reference, embed a vertical line 

• To show a distribution of data, create a dotplot

• For cleaner alignment, put graph elements directly in cells

• To have more control over data label formatting, embed labels into your graphs 

• To make a stellar line graph, add thoughtful design elements (available to premium subscribers)

unsolicited feedback

Your graph is garbage.

No productive conversation has started with words like these. However, in the past week, I’ve read them (directed at someone else’s data visualization) and had a similar comment posted to me about a blog article I published in 2012 (in which my words were referred to as “rubbish;” despite the similar trashy language, stated by a different critic). This is the dark side of the internet: where everyone has access to everything and feels they have free rein to say whatever they want. I think one of the things that irks me most in these scenarios is the clear imbalance. On the one hand, someone takes time and energy to create something and put it out into the world, perhaps with the hopes that someone will learn or be inspired, or simply to share something of which they are proud. Yet it takes very little effort to criticize.

This type of exchange is, unfortunately, all too common online. Anyone who shares their work regularly in a public setting, or follows others who do, has likely experienced or witnessed it. Such comments can send content creators in all sorts of unproductive directions: mentally (or actually!) drafting and redrafting killer comebacks, obsessing over the motivations of the commenter, or perhaps even questioning their own abilities.

While some fault-finders may criticize as a way to make themselves feel better, I try to avoid assuming malevolent intent. Instead, I imagine a scenario where the commenter is actually trying to help, but is simply going about things in a poorly-considered manner.

With that in mind, I thought I’d share some practical tips for providing unsolicited feedback. While I’m writing these from the perspective of giving unsolicited feedback online, these strategies could be applied equally well in a work setting to give input to others.

Assess: should you share your opinion? Just because a thought crosses your mind when you see something someone else shares, it does not mean you need to say it. Consider: do you have something of value to contribute? Are you interested in a productive exchange? How would you feel if someone said the words you are about to post to you—in particular if you didn’t ask for feedback? If you find yourself thinking of adding a reply or comment because you think others will like it, it may make you look good, or it might embarrass someone else, please refrain. 

Connect directly. Rather than commenting in public, consider sending a direct message. This can reinforce pure intent, because it makes it clear that you aren’t simply commenting to score internet points. This is not to say that all feedback should take place in private: there can be value in having conversations in a public setting for shared learning. But it’s best if both parties agree to that game.

Ask permission. Asking someone if they would like your input is a simple gesture that puts the power of choice in their hands. “I have some ideas for how you might improve your graph—would you like to hear them?” It starts the exchange off in a courteous manner, which makes it easier to continue in that way. 

Don’t assume—ask questions. The creator (of the graph, slide, data visualization, article, or whatever it is) invariably faced or imposed constraints to which you have no visibility. Additionally, people often place their own priorities, preferences, beliefs, and goals on other people’s work. For instance, when I’m visualizing data in a business setting, I tend to optimize for efficiency of information transfer when creating graphs: I keep things simple and take steps to make the main message clear. But this is not the only reason to visualize data, so it does not necessarily make sense for me to put this lens on someone else’s work. To deduce that, I need more information. Asking questions can help build context and having context means you can provide better input. Consider questions like, “Why are you visualizing this data?”, “Who is your audience?”, and “What are your goals?”

Practice the Socratic approach: ask more questions. In addition to understanding the situation, asking questions can also be a terrific way to offer feedback. For example, if I encounter a graph that could be described as Rainbow Brite’s nightmare, rather than lead with that, I might ask, “I’m curious: what led to your color choices in this visual?” Direct feedback can be reframed into a question to soften it. For example, “You should have used a bar chart here,” could be rephrased to “Did you consider using a bar chart for this data?”. 

Be respectful. While these seem like words that shouldn’t have to be said, it’s clear from the awful exchanges I see take place that a simple reminder to be nice is warranted. When you provide input, make it about the work, not about the person. One filter I sometimes find useful: don’t say anything online that you wouldn’t say to the person if they were sitting with you in your living room, or to a stranger you pass in a cafe. There is no good reason not to be nice to others. 

Share this the next time you see someone post something snarky about someone else’s work. Read it again on the occasion you’re tempted to do so. We each have a finite amount of time and energy and get to choose how we spend those things. Consider how you’d like to spend yours, and the impact that choice has on others. 

When you do decide to offer unsolicited feedback, please do so thoughtfully.

Related resources:
For more on feedback, have a listen to our very first episode of the SWD podcast, the art of feedback.
To practice giving input to someone who asked for it, check out the feedback section in SWD community.

how to do it in Excel: adding data labels

Today’s post is a tactical one for folks creating visuals in Excel: how to embed labels for your data series in your graphs, instead of relying on default Excel legends.

To illustrate, let’s look at an example from storytelling with data: Let’s Practice!. The graph below shows demand and capacity (in project hours) over time.

There are a few different techniques we could use to create labels that look like this.

Option 1: The “brute force” technique

The data labels for the two lines are not, technically, “data labels” at all. A text box was added to this graph, and then the numbers and category labels were simply typed in manually. This is what we affectionately refer to as “brute-forcing” your tool to make it look the way you want it to, regardless of its defaults. Remember: your audience only sees the end result of your work, even if the behind-the-scenes steps aren’t exactly elegant. 

One benefit of this approach is that I have greater control over the formatting: size, position, and color of the labels. I can easily make them appear how I want them to appear by simply adjusting the formatting, which is much easier to do with a text box than with a genuine data label. The downside is that this method may not scale easily with many graphs, or those that will be frequently updated with new data—as the data changes, the text labels won’t move with them.

Option 2: Embedding labels directly

Let’s look now at an alternative approach: embedding the labels directly. You can download the corresponding Excel file to follow along with these steps: 

Right-click on a point and choose Add Data Label. You can choose any point to add a label—I’m strategically choosing the endpoint because that’s where a label would best align with my design. 

Excel defaults to labeling the numeric value, as shown below.  

Now let’s adjust the formatting. Click the label (not the data point, but the label itself) twice, so that these white boxes appear around it:

Right-click and choose Format Data Label:

In the Label Options menu that appears, you can choose to add or remove fields by checking (or unchecking) the corresponding box under Label Contains. To add the word “Demand”, I’ll check the Series Name box.

The result is this:

The label appeared in all-caps (“DEMAND”) because it’s referencing the underlying data—I could adjust the header in Column M to “Demand” if I didn’t want the entire word capitalized (this is a stylistic choice).

To adjust the number formatting, navigate back to the Format Data Label menu and scroll to the Number section at the bottom. I’ll choose Number in the Category drop-down and change Decimal places to 0 (side note: checking the Linked to source box is a good option if you want the labels to reformat when the formatting of the underlying source data changes).

My resulting visual looks like this:  

From here, I can manually adjust the label alignment by highlighting the graph and making the Plot area smaller so that the label doesn’t overlap the line:

I’ll repeat the same steps to add the Capacity label:

The final thing I’ll do is clean up the formatting of those labels—move the numbers in front of the words, change the number format to be rounded to the thousands place, switch the colors of the labels to match the lines they refer to, and make the font for “24K Capacity” bold.

The benefit of this embedded approach is that as my underlying source data changes, the labels will update accordingly. For graphs that will be refreshed frequently, setting up these steps once will save you the headache of searching for and manually manipulating text boxes.

More Excel how-to’s 

For more tactical instructions, check out these previous posts (and let us know in the comments if there’s something you’d like to see): 

• To depict a range of values: add a shaded band

• To create a frame of reference: embed a vertical line 

• For better virtual presentations, animate text boxes to control what your audience sees 

• For cleaner visual alignment: put elements directly in cells 

• No more excuses for messy graphs: here’s a clean slate template and from data table to story

---

This post was inspired by a recent conversation during our bi-weekly office hour sessions. Do you ever need quick input on a graph or slide, or wish you could pick the SWD team’s brain on a project? Subscribe to premium membership for personalized support and get your questions answered. Our team has enjoyed getting to know many of you during these fun and interactive sessions!

tactical tip for virtual engagement

As virtual meetings become the new work norm, a question we’ve been hearing frequently is “How do I share data effectively over video?” Most of us have all experienced a massive transition from in-person meetings—where dimensions such as the physical space and how we move around within it are in our control—to being reduced to a tiny box in the corner of someone’s screen. This means we have to do a few different things in our delivery to keep people’s attention. 

Today’s post is a tactical tip for keeping your audience engaged when presenting data virtually. I’ll illustrate how using white text boxes can keep your audience engaged—by only showing one or a few elements at a time. This effect can keep your audience tuned in so they don’t miss what comes next. 

To illustrate, let’s build on the diabetes trend example from my previous post. Imagine that you work as an analyst for a large healthcare system. Your analysis of the patient base has uncovered an alarming trend: diabetes rates are on the rise and the latest forecast suggests additional resources may be needed to provide an appropriate standard of care. You’ve summarized your main findings in the single slide shown below. 

As discussed previously, while the number of words on my summary slide might be fine for an audience to consume on their own, this is entirely too much text for sharing data live. When we attempt to use visuals like this in our virtual presentations, many people will quickly scan it and then tune out as they start to multi-task or attend to other competing demands in home offices. 

To reduce the likelihood of this scenario, here’s a step-by-step overview of how to leverage white text boxes to show only a few elements at a time. You can download the PowerPoint file to follow along. 

I might start my presentation by setting expectations verbally that I’ll talk through the data piece by piece. Then, I’ll orient my audience to what metric we’ll be discussing by showing an empty graph with just the chart title and axes. This allows me briefly to explain these pieces before moving into the data.

To achieve this effect, I added two text boxes over the data and the annotations. Then I formatted the text boxes to white fill by right-clicking, choosing Format Shape, Shape Options, expanding Fill, and selecting the Solid fill and choosing White in the color options.

Finally, I removed the default outline by selecting Shape Format on the ribbon, choosing Outline then selecting No Outline in the dropdown.

You can carry through this formatting (saving you a few clicks the next time you insert a text box) by right-clicking on the shape, then choosing Set as Default Text Box.  

You can see where I’ve positioned the two text boxes here:

Next, I’ll talk the audience through the overall rate by only showing that piece of the graph. This would be important if I anticipated my audience might have questions about the methodology behind the forecast: would I need to convince them that my methodology was sound? If so, having fewer elements on the screen would more likely allow for my audience’s full attention as I answer questions related to this. 

Due to the nature where the data series are within the graph, I accomplished this effect by inserting new white text boxes, then rotating them (click the shape, then use the curved arrow at the top to manually rotate until they hide the parts of the graph). You can see how I positioned them below:

Next, I might choose to emphasize that we’re seeing an increase in all medical centers that are above the overall rate.

In this step, I removed the first rotated white box so that only one remained—still covering the centers below average.

Finally, I can speak to the centers below the overall rate by removing the remaining rotated white box:

Now that we’ve gone through things a few elements at a time, I can end with my fully annotated slide. This summary can serve as the leave behind for those who might have missed the meeting or as a reminder to those who attended.

If you’re following along in my PowerPoint file, you’ll notice that I elected to have separate slides for each piece in the progression shown above. You may instead choose to reduce the number of slides and instead leverage animation to make the white boxes appear and disappear on a single slide. If that’s the case, be sure to prompt your audience to open the presentation in presenter mode (or save it as a .ppsx type, which will open automatically in slideshow mode) so they can experience the progression.

While this is not the only way to achieve this piece-by-piece build effect, this approach can work if you don’t have a ton of time and need a quick hack to improve engagement in your virtual presentations. 

For more on presenting data virtually, check out these resources:

• Listen to Cole’s recent conversation with Tableau evangelist Andy Cotgreave in the SWD podcast episode presenting virtually.

• Register for our next virtual public workshop on September 15 and experience how the SWD team tailors their presentation style to virtual delivery.

• Watch the replay from last month’s live event mode & method and don’t miss Cole’s tactics for powerfully presenting data (and viewer Q&A) during this month’s live event on September 29 (open to premium subscribers).

how many words should I put on my slides?

As the workplace shifts to more remote communication, a question we’ve been receiving frequently in our virtual workshops is “How many words should I put on my slides?”

The answer? It depends on how your audience is consuming the information. If you’re sharing your data live—whether over video conference or in person—that typically calls for minimal text. You can provide the words verbally, which enables your audience to focus on what you’re saying rather than trying to read your slides. If you’re not presenting live, however, you’ll need more words on the page because your audience is left to understand the data without you there to help them interpret it. 

In a perfect world, this would call for two distinct deliverables: a presentation with sparse slides and a written report containing more detailed content. In reality, this rarely happens. Because of time constraints, we often create a “slideument”—it’s part presentation, part written report, and not exactly meeting the needs of either scenario. This term was originally coined by authors Nancy Duarte (Slide:ology) and Garr Reynolds (Presentation Zen), who have written about this unfortunately all-too-common communication in the workplace. Below are some example slideuments. 

Source: Google image search for “slideument”

In this post, I’ll discuss some tips for avoiding slideument, with an example excerpted from storytelling with data: Let’s Practice!  

Imagine you work as an analyst for a large healthcare system with medical centers in several states. Your analysis has uncovered a trend—a recent rise in patients’ diabetes rates and a forecast showing a continued trajectory—which you believe warrants a closer look to assess whether additional resources are needed. You’ve been directed to prepare one slide on your findings that will be passed along to administrators, first in a live meeting, and then emailed around for those who weren’t able to attend. The data used in your analysis is visualized in the graph below.

Assume we want to focus our audience’s attention to the projected rise in diabetes between 2020-2023 and prompt discussion on hiring additional staff to remain accredited within national standards. How might we design a communication that meets both the needs of a live meeting and an asynchronous reader? One approach: first design a single summary slide for those who might miss the meeting. Then adapt it—using animation—so that in the live setting, the presenter can highlight just a few components at a time. The beauty of this method is that it allows the speaker to lead the audience through a data-driven story by controlling the amount of detail shown at any given time but ends in a fully annotated view that can be distributed. 

My single-slide summary might look like the following visual.

For the live meeting, I’ll transform it. While the amount of words on my summary slide might be fine for an audience to consume on their own, they could distract on the screen. When we attempt to use visuals like this in our presentations, many people will tune us out as they read. We’ll explore this topic further during our upcoming live event mode & method (open to premium subscribers).

To keep my audience’s attention in a live setting, I’ll break this slide into several components and talk to one piece at a time. I used the SWD community exercise storyboard your project to construct the following narrative flow. 

This approach does take time, but it’s time well invested in ensuring that the hard work I do behind the scenes to find the interesting things doesn’t get lost when I share my findings. Premium subscribers can watch how I narrate these slides for the live meeting in the learning video, differentiate between live and standalone stories. 

The next time you’re struggling with how many words to put on your slides, pause and reflect on how you’ll be delivering those slides and design accordingly. Be sure to check out my follow-up post: how white text boxes in PowerPoint can improve your virtual presentations.