![](https://static.wixstatic.com/media/8fa98a_c47968ffe07e45b48c96fba73d94da36~mv2.png/v1/fill/w_449,h_445,al_c,q_85,enc_auto/8fa98a_c47968ffe07e45b48c96fba73d94da36~mv2.png)
I have seen a lot of graphs lately where colors are used to convey a message of comparing groups.
Here are some examples:
![](https://static.wixstatic.com/media/8fa98a_6a459371b4c14149b47df199d0166db7~mv2.png/v1/fill/w_980,h_520,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/8fa98a_6a459371b4c14149b47df199d0166db7~mv2.png)
Both graphs convey a message of ranking and grouping into categories.
The categories are shown in both cases in a color-coded manner instead of in a hierarchical format. As graph creators, why do we tend to create graphs with color categories?
1. The fear of being boring, one color seems uninteresting, and here we have both colors and icons. This is an "excellent" attribute for a storyteller.
2. Visually representing a group with similar characteristics makes sense.
But can such graphs tell us anything about groups? Are they easy to understand?
Let's discuss a few aspects of those cases together:
1. Color vs form
In which of the two can you see the size differences faster?
![](https://static.wixstatic.com/media/8fa98a_9ede7fcf263547b8aac4063fccb3b3fb~mv2.png/v1/fill/w_609,h_233,al_c,q_85,enc_auto/8fa98a_9ede7fcf263547b8aac4063fccb3b3fb~mv2.png)
As far as I can tell, the first one is easier than the second. A person's mind notices color before noticing the shape. Therefore, as soon as we paint the bars in different colors, we will first try to figure out what the colors mean. In addition, this creates an unnecessary load.
2. "Games" with the Legend
Once the person who created the graph used different colors to classify categories, our brain needs to perform several functions:
1. Compare sizes without being distracted by the colors.
2. Understand which color belongs to which group based on the legend.
3. Repeat step 2 if there was an interruption in the middle or if I am not able to recall all the colors.
![](https://static.wixstatic.com/media/8fa98a_f49d34b36c6c46a3a544d3acc50aaf41~mv2.png/v1/fill/w_577,h_177,al_c,q_85,enc_auto/8fa98a_f49d34b36c6c46a3a544d3acc50aaf41~mv2.png)
The Result:
A lot of unnecessary effort and the possibility of making a mistake. 3. Colors meaning Colors have meanings. Colors are interpreted by our brains in accordance with a system of stereotypes we have developed over the years. Red, for instance, conveys a warning or a negative message. Is there anything negative about the balance between work and home? Probably not...but understanding it is yet another use of energy unrelated to the message being conveyed by the graph.
![](https://static.wixstatic.com/media/8fa98a_431fa44eeb05428db6f8ba14e7545518~mv2.png/v1/fill/w_605,h_275,al_c,q_85,enc_auto/8fa98a_431fa44eeb05428db6f8ba14e7545518~mv2.png)
4. Finally, the groups
If I ask questions at the group level and show two styles in the pictures, will it be easier to identify patterns?
![](https://static.wixstatic.com/media/8fa98a_0cd9f079e40d4f03a62acc2bd426ffca~mv2.png/v1/fill/w_418,h_321,al_c,q_85,enc_auto/8fa98a_0cd9f079e40d4f03a62acc2bd426ffca~mv2.png)
As we saw above, the answer will be more obvious in a uniformly colored image. Conclusion: The whole idea of highlighting groups using colors does not work.
What is the preferred alternative? By using a matrix and conditional formatting, it is possible to create hierarchical visualizations with clear groupings and interactivity that allow conclusions to be drawn both at the group level as well as at the item level. Bringing together a developer's thought process and Power Bi's capabilities yields the result shown below. In a matter of seconds, you can view and analyze the rating at any level desired.
![](https://static.wixstatic.com/media/8fa98a_00c287a9cb8942b4b57c3562600971a9~mv2.png/v1/fill/w_865,h_419,al_c,q_90,enc_auto/8fa98a_00c287a9cb8942b4b57c3562600971a9~mv2.png)
Link to sample Pbix: https://github.com/MadeiraData/MadeiraToolbox/blob/master/Power%20bi/Groups%20and%20colors%20encodiding.pbix
Graphs sources: 1. Statista 2. https://blog.curiosity.ai/how-to-fix-a-terrible-mckinsey-chart-in-3-2-easy-steps-3481821a24fc
Thank you for sharing, Rita! I love your posts 😄