Small multiples are a popular concept in data visualization and dashboards and have been used to present information for centuries. A small multiple is a series of graphs or diagrams presenting the same measures over several or many categories, time frames or variables. Placing similar diagrams next to each other helps to identify trends, similarities and differences at a glance.

To build something like this in Tableau you need just one click – drag a dimension you want to break down your small multiples by to the row or column shelf. If you drop it in the right place (on the left of any existing dimensions) Tableau will automatically create multiple similar plots representing all categories across the selected dimension.

The only problem is that if the number of categories/values in the selected dimension is more than 3-4, it breaks the core idea of a small multiple – Tableau will arrange them into a single row or column, and the more of them there are, the harder to capture all of them in one glance. A perfect small multiple arranges the categories into a compact grid, trying to pack them as tight as possible. Let’s build a simple grid-based small multiple in Tableau.

For example, I will use the Sample Superstore dataset to build small multiple of trading trends in the nine most important cities.

First, enumerate the categories. If the number of categories is not huge, we can do it by a simple calculation.

(If the number of categories is larger – i.e., you want to represent all the States, – we can use Tableu Prep, SQL, Excel or another external tool. The nice thing about small multiples is that visually they can serve a huge number of categories.)

Next, calculate smalls’ positions in the grid.

(I converted values into strings within the expression, so my calculations are placed to the Dimensions automatically.

Finally, build the viz.

Start with a basic viz we wanted to see – i.e. sales over time (quarters here – filtered for one last year). Then add our SM position dimensions to the corresponding shelves. As there are more cities in the dataset, we will need to filter them out too.

Finally, put CITY into the label mark, so our SMs are properly labelled.

Isn’t this picture telling a story?

You could achieve the same result by building nine separate graphs and combining them in a dashboard. But when they all are in one sheet, they are easier to maintain, apply formatting, calculations, transformations etc.

Happy multiplying!

Eugene Kutilov
Author: Eugene Kutilov