It’s year 3 of Makeover Monday, the Community project, and we have a new partnership with, a social data community where we host our data sets, provide a place for people to share their visualisations, comment on others, and have a better conversation than can be had on Twitter.

The week began with a webinar for anyone new to Makeover Monday. If you missed it, you can watch it here. Wednesday we had our usual Viz Review and this time we decided to go through the vizzes posted on Over 75 minutes, we got through as many as we could, focusing on people that were new to Makeover Monday or those that hadn’t asked for feedback before. We didn’t get through them all; there were simply too many to get through. Hopefully, though, if we didn’t get to yours, you still learned something.

One of my favorite outcomes of this week was seeing so many different data visualisation tools being used. Remember, this isn’t a Tableau project. This is a data project; all tools are encouraged to join. This week we saw vizzes created with Power BI, Data Studio, R, Datawrapper, Infogram and Tableau. What was even better was how welcoming everyone was, even though we work with different tools. I think this is where moving the discussion to really makes a difference.

This year, we’re going to make these recap posts more succinct, including two lesson per week: one for design and one for analysis. However, since this is week 1 and there were an overwhelming number of visualisations created (which is great), I’m going to break this rule and include a few extra lessons learned.



This week’s data set was easy to work with and the visualisation to make over was quite simple. I really liked how so many people kept their vizzes simple as well. Many people decided to take the original chart, give it a bit of polish to make it more effective, and they were done. Let’s look at four examples where the creators kept the viz simple, added a bit of text, made the overall chart look better and/or used a different take on the metrics.

Author: Tushar More
Source: Tableau Public

Author: Yanning Wang
Source: Tableau Public


There are definitely different camps when it comes to stacked area charts. There are those that love them, especially how aesthetically pleasing they can be. Then there are those that dislike them because the patterns of the individual sections are hard to interpret.

Consider this example from Heather Murphy, which Eva and I talked about during Viz Review:

When reviewing this, I mentioned how I found it very visually appealing at first glance. The colors work well, the title is nice, it looks sharp, the icons aid understanding. As we talked about the chart more, I didn’t understand the message because of the stacked area chart.

  1. The pork trend is easy to understand because it’s the bottom most section, so all you have to do is follow the line with your eye to see the pattern.
  2. The dark line at the top represents beef because of its color, but what the line really represents is the total consumption. This can be easily misinterpreted.
  3. The chicken segment, which is the primary focus of the analysis, is stuck in the middle. This means that you have to always account for the size of the pork segment when trying to understand how big the chicken segment is. The line at the top of the chicken segment is impacted by the line of the pork segment. Therefore the pattern that you see with the chicken line, doesn’t accurately represent the trend of chicken consumption.

This chart by Daniel Caroli is an example of a stacked bar chart (could easily be an area chart) that works well.

Daniel’s chart works well for two reasons:

  1. The bars are stacked to 100%.
  2. There are only two colors, which makes the pattern of one versus the other very easy to see.

For more reading around this topic, check out this blog post by Stephen Few and the accompanying discussion between he and Cole Knaflic.


The data set started with the year 1960 and ended with 2018. There were a couple of gotchas in the dates that I was wondering if people would notice.

  1. There is no data for the years 1961-1964. What should you do? Should you include or exclude 1960? If you want to look at the rate of change, what year should you calculate the difference from?
  2. 2017 included actuals and estimates. 2018 was a forecast. Should you include these?

Natasha Kurakina, one of my Data School students, created this beautiful viz that included 1960.

Notice the gap between 1960 and 1965. Natasha liked that this showed off where data was missing. Without including some markers, it could easily be assumed that there was an identical year over year increase from 1960-1965. I “suggested” removing 1960, which she reluctantly did.

Excluding 1960 makes the viz look cleaner and it tells a more effective story. Notice that Natasha also decided to exclude 2017 and 2018 since they were not complete actuals. This helps with consistency in comparing the metrics. Well done Natasha!


Another viz we reviewed during Viz Review was this one by Mehsam Raza Hemani.

If you don’t look closely, you might not notice that the left and right axes are not synchronized, despite both being measures of consumption. Why does this matter? Well, when looking at the two lines, you naturally assume they are on the same scale and, with this chart in particular, you focus on where the lines cross. In fact, this is what the entire viz is about and there’s a nice circle around the intersection point.

The problem is, with the axes not synchronized, the intersection actually isn’t in the right place. This is misleading. His second chart is even more misleading. Check out how far off these axes are off. The annotations make it seem that seafood (right axis) overtook chicken (left axis), yet the consumption of seafood is significantly less than chicken. Don’t believe me? Look at the scales.

The lesson here is to pay careful attention when creating dual axis charts. If not, your analysis could be very, very misleading, which in turn could hurt your credibility.


Adding text and annotations to visualisations is one of the best ways to aid your audience in understanding the data and explaining your analytical story. This week,  created this masterpiece that demonstrates exactly how to use annotations effectively.

In this connected scatterplot, notice how  used simple annotations to explain what each area of the chart means.

I don’t believe I’ve ever seen a connected scatterplot explained this way. It’s not much text, doesn’t take up a lot of space, yet it adds so much value for understanding the chart design. I’ll definitely be stealing this idea!



Source: Power BI

Source: Tableau Public

Source: Tableau Public

Source: Tableau Public