This was originally written on July 17, 2014 and published on Hatch, our internal version of Medium. Read more about this collection at Inside Medium.

homepage experiments

The fundamental difference between Medium and any other blogging site is the network. So we say. But we need to work to make it a reality. When can we say that we are really adding value with our network of connected posts? When being on Medium has a noticeable, measurable, significant effect. For writers, this means writing on Medium consistently and noticeably sends you more traffic than you would get just posting on your own site. We can measure this with network TTR. For readers, this means Medium itself is a destination, not just the container for the post you read (and bounce from). This means wanting to go to Medium, on purpose. We can measure this with 3 in 7. To improve our 3 in 7 numbers, we recently discussed preventing churn: widening the gap between when you gain a new active 3 in 7 user, and when you lose them. This means going beyond just hoping they see a tweet every day with a Medium post they might want to read. This means giving them a reason to come back. Features that make the site feel fresh. Features that encourage the formation of the habit of typing into your browser every day.

So how can we make the site feel fresh without resorting to displaying all posts sorted by time in descending order?

One way could be having a theme of the day. Every day, there’s a new theme, curated by editors, User Happiness, or the Dewey contextual feature platform (i.e., based on trending themes or your own topics of interest). Every day, you come back to Medium because you want to know what the new theme is. But the posts associated with the theme could still be “evergreen content,” not necessarily new stuff from the week or whatever.

I actually got this idea from clothing and fashion sites. Even ones that don’t have new clothes all the time can change the theme — the types of items being highlighted on the homepage — frequently to make it fun to visit more often than once a season. The featured items on the homepage, like the sandals and sunglasses here, are not necessarily new, but they might be interesting to browse anyway.

You might not like this idea. Maybe this is bad because what if no one cares about Travel or whatever random thing the theme is. You might say we should just make the homepage look like Compass, because how is the web inherently a different problem than mobile? Or you may assert that the best homepage when we put Brad’s face on it.

Guess what. Maybe these are bad ideas. But the existing homepage is also bad. And it gets a lot less traffic compared to the rest of the site (i.e., post pages), so we don’t have very many users to upset if we change it. See where I’m going with this? Let’s just try some of my bad suggestions, and your bad suggestions too, and we will be able to get data on whether they are actually as bad as we think they are.

I’m thinking we churn out a bunch of stuff really fast. I mean 1–2 days to design, 1–2 days to implement, for each idea. Then we can get a lot of ideas out there quickly, and we won’t feel torn up about it if we have to ruthlessly kill the ones that aren’t performing.

In fact, let’s have a contest. Let’s run some A/B/C/D/E/F/G/H/I/J tests together (with one of them being the existing homepage). I’m thinking of using any free time I have in the last hacking day of Hack Week to start working on some of these (Maybe not the one with Brad’s face). Maybe Ev will even let us merge the stuff into master. And then let’s see which one actually wins.

Former Medium Engineer. — Find me on Twitter: @infraredflower