I started auditing Medium’s lame Stats redesign, and realized why all their UX sucks so bad.
All the stuff Medium missed or screwed up
Line graphs for time-category data
This is the first thing that jumped out at me. Most of the goodwill they engendered by forgoing the aforementioned pie chart they have squandered by this misuse of chart type. A line graph is for comparing changes in a single category over an independent variable (usually time). A bar graph is for comparing categories. That category can be time intervals, such that each bar represents a day, a year, an hour, etc. That is the way Medium’s old Stats design works, and in this aspect it is actually a step backwards.
A line graph implies that there is a value that is continuously changing, and the line is tracking that value. A chronological bar graph implies that there is a change being tallied at a regular interval, and the total for each interval becomes the length of the bar. Mixing these purposes up is confusing to the person reading the graph, even if they aren’t already taught how graphs work.
If that were not bad enough, as you can see that the line area graph is the stacked type. This means that, for each day, the number of views from Medium determines the height of the dark area, and the number of views from other sources is the distance between the dark area peak and the light area peak. Take the graph below:
If this were not a stacked graph, then it would indicate there were more views from outside of Medium because the light green peak is higher. But because it is stacked, it indicates that there are fewer views from outside, because the area between the color peaks is smaller.
As you can see, it is really hard to compare the change in external sources over time because it is so heavily distorted by the height of the Medium source graph at the bottom. This is already a problem when you have stacked bars, but stacked line graphs? That’s even worse. That is two reasons that the choice of line graphs was wrong.
No way to compare read rates or fans for each traffic source
One of the most important things I might want to know about my traffic sources is who is bringing in the most quality traffic. If I get 10,000 views in a day from some site, but the read rate is less than 1%, I will probably not spend a lot of time marketing to that demographic, or optimizing my content for it. If I consistently get a high read rate and lots of “claps” from another site despite modest traffic, I will try to serve that market.
Data is still limited to stupid 30 day intervals
If you want to see the performance over time of an article, you’re SOL. You can only look at it in increments of 30 days, and the height of the graph adjusts to the data, meaning that even scrolling through won’t allow you to see that change over time.
Still no explanation of traffic sources
There are lots of mysterious names that are a frequent source of traffic. They are frequent enough that Medium must certainly know what they mean, and could offer an explanation. But they don’t, leaving us in the dark as to a good portion of our traffic. Moreover, it still does not give a visual indication of which sources can be clicked on, so you can see where those links are (this can be done on Twitter and Facebook), and which sources only lead to the top level domain (such as Google or LinkedIn), or which don’t do either.
Mobile still sucks donkey balls
The above is from the Help page. Do they think that we aren’t interested in accessing our Stats when we are on the go?
The back button is on the right
I suppose I should have included the presence of a back button at all in the good news, considering it used to not even have one. But then, it was such a pathetic omission in the first place, that it doesn’t really count. But really, you put the back button on the right? Who the fuck even does that?
Because right means forward, putting the button up there suggests it links to a page with richer data on that particular article, rather than a zoom-out back to the Stats home. The bad, ambiguous microcopy doesn’t help.