You're contributing to an investor presentation and your boss is simultaneously happy and unhappy. You found a killer insight, but presented it in a poor way. Nobody ever told you how to make it better!

UPSERT OFFICES – THE DAY BEFORE AN INVESTOR MEETING

I need to talk to you about a chart
First, I just want to say I thought about it and this is one of the most important insights in the whole presentation!
Knowing that 2/3rds of our ad spend goes on variations we eventually turn off is very important for our discussion with investors
It shows how much we need to spend to find a winning variation
That said, the way you presented it... needs some work
I'm not sure if you've ever taken a data visualization course, but I'm about to blow your mind
Let me send you a blog post on the 'data ink' ratio

The concept of data-ink comes from Edward Tufte, in The Visual Display of Quantitative Information. He described data-ink as the "non-erasable core of the graphic", meaning stripping away anything redundant to get to the core of the story you're telling with the data.

Most charts suffer from too much redundant information or noise, which can make them hard to read. Most people, in particular important decision-makers, don't spend as much time with data as the person who made the chart, so there's a constant danger of miscommunication that must be addressed.

Data-ink is what's left over when you remove anything that's not directly supporting the message of the chart. The default charts and popular features available in Excel or GSheets often run counter to best practices for conveying information. By removing redundant labels, increasing font size, getting rid of special effects, and other simple rules, you can make an immediate impact on a chart's effectiveness.

Data Ink Ratio

Hey, we're going to improve the data increase, share of first shop. So you see this sort of thing all the time.
People who don't know about data visualization will make a child that they think looks nice, though. I have lots of special effects, you know, colors and we're going to remove all of that and we're going to see how much better it looks.
Side-by-side. So if you just copied the child, I just clicked on it and then press control C and then paste it here.
This is in Google sheets. All of the same things are possible to do in Excel as well. Although it will be slightly different for you to do them.
Okay. And we have a list of different things we need to do on the side here. We're not going to cover all of them.
In this tutorial, I'm gonna leave some of these exercises for you to do at the end. Okay. So this is a pretty bad looking pie chart and the, the number one thing you really want to do that will help improve your visualizations in Excel or Google sheets as well as actually, this is good advice for PowerPoint presentations too, is remove all of the special effects.
You're much more likely to get them wrong than you are to have them improve your presentation of data. You know, you're not the New York times.
You don't have a full-time data visualization specialists or to really like pull this off. So just going to change this do we click into the chart and go let me see pie chart.
We're going to take it off 3d. Okay. So straight away that is easier to read. It's not at an ambulating more, and you can see that this smaller green section of removed and now actually kind of looks a little bit bigger.
So that's the first thing we do remove special effects, remove redundant labels. So we have, you know, 33% here, 33% here enabled here.
And then this data, we don't really need to know the actual numbers necessarily. We just need the percentages. So let's go in here and instead of value in percentage, we're just going to change this to percentage, to this customized pie chart percentage.
Cool. So now we just have the 33% hair, but we also have these labels too. So this is part of the legend, and you can see that if I click on it, it says legend and it changes instead of auto.
Let's say let's put that at the top or we can put it on the left. Okay. This looks pretty good.
Cool. So now it's already much cleaner. You can see, we have the legend that we have the percentages here. All right.
Removed backgrounds. So this yellow background is just distracting. It doesn't add anything to the data and anytime you use color, it should be really intentional.
You should be drawing attention to a piece of the chart. So you get to chart style, background, color, and then just click non cool remove borders as well.
You know, you can see this looks not great because this border like having is all wrong. You can spend on a lot of time messing around with it.
But actually let's just go in and remove that border and it should look better straight away if you have borders on the actual pie segments as well.
I would remove this to re bolding.  So both texts is a little bit actually harder to breed on these types of charts.
So I'm going to go in and we're going to remove the bold from the labels. So a little bit easier to reach straight away.
We're going to remove it from the chart title and we're going to remove it from the legend as well. So look at that, it's already looking a lot better remove redundant pixels, so this isn't always possible, but you can see in this case with a pie chart, we have the option to make it actually a donut chart, because right now this is a huge block of color and we don't necessarily need that huge bulk of color in order to you know, visualize the data.
It becomes a little bit easier to read less mentally taxing if you make a hole in the middle and turn this into a donut chart.
All right. And then so that's removing redundant pixels, or we want to improve the colors. And the reason I want to do this is these are the worst colors you can pick for someone who's color blind and actually 8% of males.
I think it is a color blind much less likely in females, but but it's a, it's a problem that everyone deals with.
And actually you know, it's the good default colors to use the, the the calibrate color blind, friendly colors which are like blue, orange, gray because they have higher contrast less likely to get mixed up.
So we're going to change to that. The there are different color palettes. You can look up online and it, Tableau has a really good palette that you can search for.
But in this case, we're just gonna kind of go with blue for enabled paused. We're going to change that to orange and then removed.
We're going to change that to gray. Cool. So already we have a much better chart. It's much easier to read.
There's more that we can do on a light in the labels. We want to increase the figure size and font size when it adds some source attribution as to, we want to say you know, where did this chart come from?
It's from the Google ads data. We want to say what, what date range it came from as well? I would compare that underneath the chart.
Don't be even enemy to attach it to the chart you can just put text underneath. And then we want to think about maybe a more appropriate chop to use as say, if you Google why a pie chart is bad, he should be able to find a good alternative.
Okay. Hope for this was a.

When would it make sense not to reduce the data-ink ratio?

What would be a better choice of colors to aid color-blind people?

What would be a better headline for this chart that 'revealed the punchline' i.e. made the key takeaway clear?

Why might it make sense to use a bar chart rather than a pie chart?

Data looks better naked

You should spend twice as much time on the presentation of the data as you do finding insights to present. Otherwise your data will never drive important decisions because the people that make those decisions are busier and less technical than you are.

Vexpower Learning Certificate of Completion

Congratulations relieved-machine-75 💪