“Where should I place the labels on my graph?” It’s question that comes up in many data visualization discussions. Although the decision about where to place your labels is largely an aesthetic preference, I do think there is an objective logic you can follow.
Let’s start with this simple line chart of the share of people in the labor force by generation—a graph that I saw in Axios and in Philip Bump’s newsletter. In this basic chart, we have a legend at the top of the graph.
There’s nothing inherently wrong with using a legend, but it’s disconnected from the data. As I argue in my Better Data Visualizations book, a better approach is to integrate your labels with the data, making the graph easier for your reader to understand. In this next graph, I remove the legend and directly label each line.
I see two issues with this revision.
First, the reader has to work to figure out which label goes with which line. I’ve arranged the labels to be as close to the associated line as possible, but there are some cases (e.g., Gen Z and Boomer) where the label is close to multiple series.
Adding color to each label is a simple change that can add clarity.
Second, depending on where the labels are placed, certain series may appear more important than others. For example, Because the “Gen X” label is closer to the title—which we believe people tend to read—it might be perceived as being more important and is more likely to be read.
Instead, let’s place the labels off the right side of the graph. This approach neatly orders them along a single vertical column, with the color-coordination integrating the text with the data. Keeping the labels aligned along a vertical line allows the reading process to be easier and faster.
In some cases, you might have missing data or an incomplete data series, as Philip Bump did in the original graph he published in his newsletter. Notice how the labels are aligned along the right-outer edge of the graph, but only the “Pre-silent” label sits by itself. Because the data are incomplete, that’s where that label has to go–but the rest of the labels are aligned and easy to read.
As you think about labelling strategies, try integrating the text with the data and making it as easy as possible for your reader to navigate through and around your graph.
Hi Jon, thanks for this food for thought. I agree with the rationale for your enhancements, but for me, a little something is left unresolved in the final version. This is because with the series labels to the right, and the axis markers to the left, the reader has to flick their glance backwards and forwards between left and right to gauge a series starting value and ending value.
I would alleviate some of this flicking by adding in final data values. In my file uploaded here, the LHS image adds both series name and value to the last marker for each series. To my mind, this version reduces the cognitive load for the (left-to-right) reader because it lets them take in the scale of the vertical axis first, and when they then trace a line rightwards to its final data point, they don’t have to flick back left again to the axis to remember the scale and get a sense of the final value.
Even better (in my personal opinion of course) is my RHS image in the attached. This is because it sets the scene early – with series name and colour priming at the left-hand side, close to the vertical axis marks – so those aspects are now connected for the reader, along with a view of the starting rankings. Then the reader traces each line rightwards through to its final data point to see the change (without having to flick back left again to remember the scale of the gridlines). What do you think? I’d be interested in your feedback!
In both cases, the Alpha and Pre-Silent series are small anomalies in placement, but as you say, this is justified given they are incomplete. Also, I’ve plotted these charts using my rough calcs from the data, so any errors mine etc…
Attached image:
Jane… if the values are important (not just trends comparison in which case you may not need them) than your solution on the left is elegant. the solution on the right is a mess driving me back and forth finding labels and numbers and wondering what the emphasis is. even looking at them zoomed out together you can see order versus chaos. not a fan of that one.
Effective labeling ensures that viewers can quickly grasp the key insights without confusion, making data interpretation a breeze. Whether it’s using descriptive labels, color coding, or incorporating legends, thoughtful labeling strategies enhance the overall readability and usability of graphs. Here’s to mastering the art of graph labeling and creating visuals that speak volumes
Great labelling strategies, thnaks for sharing