I have a series on Instagram where I check dataviz palettes for ideal text contrast (WCAG 4.5:1 contrast ratio) and simulate how they appear in protanopic, deuteranopic, and tritanopic vision. Ideally, direct labels should have enough contrast to be read easily by people with lower vision, and all colours should be reasonably distinct for people with the most common forms of colourblindness.
Keep in mind that, despite contrast checking and colourblindness simulation, not all screens display colours in the same way. Some display in true colour, but many have different contrast, saturation, or brightness settings, polarizers or other filters to reduce eye strain (particularly on enterprise hardware), ’night light’ reading modes that shift colours away from blue, and so on. Test palettes as much as you can within your own business environment, and allow users to modify the palette in interactive environments where possible.
Here’s a summary of my recommendations so far.
Recommended Palettes #
Tol Muted (Modified) #
This is a modification of Paul Tol’s Muted palette, with the olive colour removed and four of the colours modified to increase their distance from one another. An organization selected the Tol Muted palette to use in reports where colour accessibility was a large concern. On testing, some colours were too close for comfort and I modified the palette for them. This palette is not branded, so feel free to use it yourself.
It is meant for use with qualitative or categorical data. The final grey colour is intended for ’no data’ or non-highlighted areas.
Not Recommended #
Excel: Office 2023+ #
Excel’s palettes are famously bad for colourblind users, and since Excel is many people’s entry into the world of data visualization this palette is found everywhere. In protanopic and deuteranopic vision colours 2 and 6 look identical. Colours 1 and 5 look identical in deuteranopic vision, and colours 1 and 3 look very close in tritanopic vision. If you must use this palette, skip colours 1 and 6.
Note that if you have more than 6 categories the palette repeats with slightly different shades (causing even more issues for low-vision or colourblind users).
See the original post on Instagram.
Power BI: Default #
Power BI is found everywhere in the business world these days, but if you’re using it make sure you change the palette from the default. Colours 2 and 4 look identical in protanopic and deuteranopic vision. Colour 1 looks very close to colours 5 and 6 in protanopic vision (but 5 and 6 themselves are reasonably distinct). Colours 1 and 6 also look too close in deuteranopic vision. In tritanopic vision it’s very difficult to tell the difference between colours 3, 5, and 8. If you must use this palette, avoid using colours 4, 5, 6, and 8.
See the original post on Instagram.
Tableau: Tableau10 #
I love the muted colours of this palette, and you don’t need to switch label colours back and forth. However, muted colours tend to create problems for colourblind users. In protanopic vision colours 3 and 9 look identical, as do colours 4, 8, and 10; colours 1 and 7 might be too close for comfort. In deuteranopic vision colours 3 and 5 look identical (9 is a bit too close), and colours 4 and 10 also look very close. In tritanopic vision colours 7 and 9 are not reasonably distinct. If you must use this palette, avoid using colours 5, 8, 9, and 10.
This leaves you with 6 reasonably distinct colour options, which is enough for many visualizations.
See the original post on Instagram.
Qualtrics: Default #
Rainbow palettes are notoriously difficult to make colourblind-safe, and this palette is particularly bad in protanopic vision. Colours 1, 4, and 6 are not distinct, and colours 7 and 8 look identical. The palette generally looks acceptable in deuteranopic vision, but be warned that minor screen modifications can make colours 1, 4, 6, and 8 blend together easily. In tritanopic vision colours 1 and 8 are nearly identical, and colour 6 is close to both of them. I love the vibrancy of this palette but I don’t think it can be salvaged unless you have only two or three categories.
See the original post on Instagram.
SurveyMonkey: New Default #
This palette looks drab in full colour vision, which also hinders its accessibility. In protanopic and deuteranopic vision colours 1 and 3 look identical, and colour 8 is fairly close to both; colours 2 and 10 will be difficult to distinguish on many screens. In tritanopic vision colours 1, 6, and 8 all look nearly identical to one another.
I much prefer SurveyMonkey’s old palette. It wasn’t perfect, but at least it was visually interesting.
See the original post on Instagram.