
Exploring Small Improvements Of Medium’s Biggest Features
I love Medium and I use it every day. Right now, I’m halfway around the world, and use it to connect back to everything happening in Silicon Valley. This is how I would improve Medium for me and my friends.
The first two laws of Human Centered Design are “know thy user” and “thou art not thy user”. Taking those laws to heart, I decided the best place to start would be calling some friends that use Medium, specifically on iOS. After six 10-minute informal interviews with friends about their experience with the Medium app, a few things bubbled to the surface:
- They bookmark a lot of stories, but never get around to reading them.
- They have trouble finding the newest stories for specific tags, because “I don’t know where the ‘latest’ button is.”
- They feel overwhelmed by the amount of content presented to them.
I asked “Why?”
“I want to read great content.”
Well, that makes sense. Medium does one thing extremely well — it provides a platform for amazing content to make its way to interested readers.
So instead of improving the app for myself, I came up with a blended persona of the people I interviewed to guide my thinking throughout this process…
Meet Skyler.
- Skyler is afraid of missing a great story by not bookmarking the ones that pique his interest, even if they’re only mildly interesting… Therefore, he bookmarks a lot of stories.
- He knows he has too many bookmarked stories to read, but continues to bookmark more stories before reading the ones he has.
- Bookmarks can easily be forgotten because the button (and only reminder) is hidden 2 clicks away.

Based on that minimal qualitative research, improving user engagement with content seemed like an interesting direction to go. But what purpose does improving engagement serve? Well, content is at the core of everything Medium does. The more that users engage with content, the more that Medium can leverage that to its benefit as a business.
With engagement as my true north for this exploration, I would need to find a way to provide users with Medium content that is the most relevant to them. Some KPIs (Key Performance Indicators) that might reflect this direction are:
- Trust: the amount of time users spend reading stories compared to how much time they spend in the app. The ideal would be that users trust Medium’s recommendations enough to be reading the whole time they are in the app. When the ‘gulf of satisfaction’ is wide, they start searching. Quality of content curation.
- Connection: the number of times a user highlights content in a session. Highlighting is a reflection of users connecting with content. Engagement with content.
- Excitement: the number of times a user shares a story per session. If users share stories, the content is exciting enough for them to think that their network (friends, followers, etc) will enjoy it. That sounds like engagement to me. Engagement with content.
- Confidence: the number of times a user un-bookmarks a story without reading compared to the number of times a user bookmarks a story. I would want a user to be confident that every story he bookmarks is worth reading. Quality of content curation.
KPIs that I wouldn’t track because they don’t provide as much value:
- The number of times a user recommends a story. I’ll sometimes like my friends’ writing just to support… or I’ll like 50 things in a session because I’m in a weird mood. Vanity metric.
- The percentage of stories read that were received by a notification on mobile . A user can turn on notifications for every story published by people they follow. It’s out of my control if that content is any good.
I started by trying to generate as many ideas as possible before going deep on any specific one. Here is a picture of the possible solutions — good and bad — that came to mind.

A few things I took away from the above list of possibilities.
(I’ll be using the royal ‘we’ here)
- How might we move ‘reading bookmarked’ stories up on Skyler’s priority list?
- How might we get Skyler to only see content he wants to read?
- How might we remove ‘analysis paralysis’ from reading stories?
The Finalists
After a brief review of the possible solutions, I’ve chosen the best options that work best in this specific situation.
Super Recommend ❤️
Medium meets Tinder🔥? You only get 1 a day, and it says a lot about what content you are into.
If Medium wanted to show this Super Recommend publicly, it could give a user insight into how many other users were willing to recommend this story over anything else that day. Something like…
💚 1.5k ❤️ 68
This would give Skyler a second dimension of information with which to decide if he wants to read it or bookmark it for later.
If Medium wanted to keep Super Recommend data private, it could be used to create a tighter data profile on users, so they can have fewer and better recommendations on their homepage.
Medium recommendations are the love-child of Twitter’s favorite & retweet button. Though elegant and simple, it’s hard to decouple enjoying a story from wanting to promote it. When users are encouraged to be discerning with recommending content, it gives readers more confidence in opening the curated content that appears on their homepage.
Tiered Highlights 🖌
In addition to the “Top Highlight”, it would be nice to have different shades of green highlights showing which parts were highlighted with what frequency, so Skyler would know how often parts were found to be invaluable.
This would minimize “buyer’s remorse” when starting a long article. This also asks for the least change of behavior from Skyler and the Medium community.
The highlighting tool that exists right now is already pretty sweet. I was initially thinking about how we might get users to highlight more often. This ended up not being the right question. Asking users to highlight more stuff is asking them to lower their standards of what they think is quality material, which grinds against what I believe Medium is trying to create — a space where high quality content is applauded and organically distributed.
Sharing with specific users 📲
Right now, there’s no interface to share Medium stories with specific Medium users. But the homepage is an interface to see those stories — it’s just that it hasn’t been implemented yet. Intuitively, when I think about who could recommend the best content to me, it would be people that know me personally.
In my experience, recommending doesn’t feel like I’m recommending to my followers, but more like just recommending in general to boost the popularity of a story. It would be awesome if the ‘For you’ tab on the home screen included stories that my friends had specifically shared with me.
In Conclusion
I would have to A|B test these improvements individually to see which ones performed best against the KPIs I defined. Determining if a success metric is statistically relevant can only be done once I know how a feature moves the needle on the KPIs.
Of course, were I part of the Medium team, my approach would be informed by data and knowledge of the avenues that have already been explored. Medium is doing so many things right, and this was my humble offering of how I might contribute to that trend.