Product Management Lessons: How Medium replacing ❤️ with 👏has made the product much worse
Recently Medium announced a key change in how users interact with their product in a blog post titled Show authors more ❤️ with 👏’s:Introducing Claps, a new way to react on Medium. This is a classic example of a change that optimizes for vanity metrics instead of improving the experience of users (both authors and readers) of the product.
In the announcement, the Medium team wrote
Today we’re hoping to change that. Rolling out to Medium users over the coming week will be a new, more satisfying way for readers to give feedback to writers. We call it “Claps.” It’s no longer simply whether you like, or don’t like, something. Now you can give variable levels of applause to a story. Maybe clap once, or maybe 10 or 20 times. You’re in control and can clap to your heart’s desire.
To understand why this change makes the product experience worse, it is useful to first understand what the purpose of “likes” are in a social media product.
The Purpose of Likes in a Social Media App
The ability to a like a post on a social media platform like Facebook, Twitter, Instagram and until recently, Medium, serves a few distinct purposes for users and creators of content on these sites.
- A like is a way to encourage authors to create more content via the dopamine rush of being told people appreciate their post and shows them the type of content their audience likes.
- The number of likes sends a signal to readers of how interesting or not the content is especially for content that may take time to digest like a video or long form post so they can decide if it’s worth investing in clicking.
- Likes provide a way for the reader to “thank” the author for their content and signals to the platform’s relevance algorithms the type of content the user would like to see more of.
Medium switching from allowing users to provide a single ❤ to allowing them to provide 1 to 50👏 breaks pretty much all of the above.
How 👏’s Muddy the Waters
Let’s look at each of the benefits of “likes” on social media platforms and see how Medium’s old system of ❤ compare with 👏.
- In the old system as an author I could tell that a post with 25❤ was appreciated more by readers than one that got 8❤. In the new system any user can give 1 to 50 👏 (claps) . So it is now possible to have the same two posts where one has 25 claps and the other has 155 claps because three of the people who read the second post gave it 50 claps. So as an author the number of reactions is now not a useful guage of interest. This is further complicated by Medium deciding to pay authors based on how long people hold down the 👏button (i.e. number of claps) instead of how many people liked the post or how many people read it. This is very counter intuitive for authors.
- Using the previous example, a reader may thing an article with 155 claps is more interesting than one with 25 claps yet in reality 3 times as many people liked the latter article than the former. This means the number of reactions on a post is no longer a meaningful measure of how interesting or popular a piece of content is. For a real world example, this post I wrote in response to James Damore’s gender manifesto currently has 830 claps while this post I wrote last year about diversity hiring practices at tech companies has 799 claps. However the older post was actually appreciated by twice as many people. 🤔 XKCD captured the problem of users overrating content in this classic comic
3. As an end user, instead of just clicking like I now have to decide what score out of 50 I want to give a Medium post as if I’m some sort of 10th grade English teacher. It actually now turns what should be a lightweight and fun experience into somewhat of a cognitive load on end users.
So why did Medium make this change?
From the announcement post
The Recommend — our version of a Like or upvote or fav — has been our explicit feedback signal since almost day one. Explicit feedback is the most valuable signal, both for authors and the Medium system. But a simple, binary vote has its limitations. It shows you how many people thought something was good, not how good was it?
A system that allows 2 people to leave 100 claps as a score for a post where most other services would give it 2 likes sounds more like a recipe to game the system than a useful way to tell the quality of an article. As it stands, I’ve already started mentally discounting the number of reactions on new posts on Medium since it only takes two of the author’s friends for a post to have a 100 claps. Medium took a feature that billions of people all over the world understood and have become familiar with then replaced it with a system that is less clear, more complex to use (how do you remove claps on a post?) and untrustworthy.
The product management team had the right goals but ended up building the wrong feature.