Overlay Bar Chart

In this post, I will discuss the overlay bar chart. There is plenty of material on stacked and grouped bar charts, but relatively little written about overlay bar charts. This post attempts to address that gap.

There are many ways to visualize a dataset. Some approaches are better suited than others depending on the nature of the data. The same applies to overlay bar charts. They work well when certain conditions are met. We will start with a generic chart and then see whether it can be improved.

The data we are going to look at contains two metrics that are highly dependent on each other. The example here is ad impressions and ad clicks. There is a strict relationship between these two metrics: clicks can never exceed impressions, because a user must first see an ad before clicking it. This type of relationship appears in many real-world metrics where one action can happen only after another. Examples include emails received vs. emails opened, page views vs. sign-ups, product views vs. purchases, searches vs. bookings, and video plays vs. likes. In all these cases, the second metric represents a subset of the first.

In this post, I will use ad impressions and clicks as the example, but the same idea applies to any pair of metrics with this type of relationship.

Grouped Bar Chart

Let’s start with a grouped bar chart. This chart makes three comparisons easy: you can compare impressions across days, compare clicks across days, and compare impressions and clicks within the same day.

Let’s increase the number of data points. The chart now starts to look a bit noisy. Can we do better? Can we make better use of the available horizontal space?

Stacked Bar Chart

This looks visually cleaner and makes better use of horizontal space. However, an important trade-off appears. We can still compare impressions across days, but we can no longer easily compare clicks across days. Since clicks are stacked on top of impressions, the visual baseline changes, which makes height comparisons difficult.

We also lose the ability to quickly compare impressions vs. clicks for a specific day. For example, if clicks are 30% of impressions, that relationship is no longer obvious because the bars are stacked rather than placed side by side. Instead, the chart emphasizes the sum of impressions and clicks, which is usually not a meaningful metric in this context.

Overlay Bar Chart

Given that clicks can never be greater than impressions, we can place the bars directly on top of each other. The impressions bar forms the full height, and the clicks bar sits within it because it is always smaller.¹

This layout retains all the advantages of the grouped bar chart. You can still compare impressions across days and clicks across days because both metrics share the same baseline. You can also compare impressions and clicks within a specific day, since both bars start from zero and their heights remain directly comparable.

At the same time, the chart uses significantly less horizontal space than a grouped bar chart. Instead of placing the two bars side by side, the overlay approach combines them into the same position while preserving the relationships between the metrics.

Variations: Inset and Shift

You might wonder how to visually distinguish this from a stacked bar chart, especially if both appear in the same report. One simple approach is to use a slight visual offset.

For example, instead of using a full-width bar for clicks, you can render it slightly narrower. This creates a small inset, making it immediately clear that the clicks bar is overlaid inside the impressions bar rather than stacked on top of it.

Another option is to apply a small horizontal shift. A large shift would effectively turn it into a grouped bar chart, but a subtle shift keeps the bars visually connected while still signaling that one bar overlays the other.

Conclusion

Overlay bar charts work well when two metrics follow the rule A ≤ B, where A is always a subset of B. In practical terms, this usually means measuring how many users completed an action out of the total number who had the opportunity to do it. For example, out of all ad impressions, how many resulted in clicks; out of all emails delivered, how many were opened; or out of all product views, how many led to a purchase.

In these situations, the smaller metric is typically the outcome or success event, and it is almost always interpreted in relation to the larger metric. The overlay bar chart keeps this relationship visible while using less horizontal space than a grouped bar chart. When you have many categories or time points, this makes the chart more compact without losing the ability to compare the two metrics.