New Chart: awareness/adoption delta

[SachaG]

It would be nice to visualize how a feature (or tool)'s adoption or awareness have evolved over the past year.

One way could be a simple ranking of percentages. For example, if a feature goes from 22% to 30% awareness from 2022 to 2023 that would be a 8% delta, and we just rank deltas for all features in a simple bar chart.

But maybe there are other, more interesting ways to visualize this?

cc @ShaineRosewel

[eric-burel]

Looking at some charts they seem to compute simple bar charts indeed. What you can play around with is perhaps the unit, here you are describing using points 30-22=8pp but you can also a relative difference (30-22)/22=+36% or indices 100*30/22=136.

A fun visualization could be an animated one, the relative difference is also the speed at which your value change

[eric-burel]

Thinking a bit more about this, if you want to show the value + the variation, you can perhaps rely on "paint brushes", like on paint apps where the velocity is taken into account and can change parameters. It can make line charts more lively, we could draw lines taking the values into account. The difficulty is in 2 parts: getting the maths right (but there are probably tools in the wild), and getting the rendering right. In SVG that's probably not hard, you can make the line a path and the path would be it's contour. In HTML + CSS I don't know if it is feasible.

[SachaG]

We have this kind of "paint brush" effect in Amelia's "arrows" chart. It's true we could explore it more.

But in the meantime I think just having this chart, but for features, would already be pretty good:

[ShaineRosewel]

I suggest that we keep the viz as simple as possible. Percentages could be prone to confusion.

Imagine say that we have 1000 respondents in 2021 and then 500 in 2022. If 50% said yes in 2021 and 100% said yes in 2022, that means a difference of zero percentage points - what does that mean for us given the different number of respondents? 50% in 2021 means 500 and 100% in 2022 is also 500. In reality we know, it could happen that the 500 who responded in 2022 came from say a very similar subset and getting outside that subset would easily change that percentage. We have different base number of respondents.

What I am trying to say is let us keep it simple and if possible, explained: bar chart is okay and a note that the number of responses vary across the years will be helpful and probably a reason on why this is the case. Also a note on what is common across the years (ex: mostly professionals) could be helpful.

[SachaG]

Well, in any case I did implement the line chart and rankings chart for features too: https://css2023.onrender.com/en-US/features/