Here, I’m reviewing medium strategy on tag feature

Hypothesis and KPIs

Qualify and link the information

Medium took a disruptive approach to information by creating a mix between microblogging and webzines. In both scenarios, categorizing and quantifying the information is key to drive user engagement.

The mission of Medium is to give a voice to people who have something interesting to say even if they don’t have millions of followers on twitter or a blog on the NY Times website. So while they are working hard to give everyone a place to write about what matters to them, they need to make a great deal of effort to qualify the information so that visitors get to see what they are looking for and stay as long as possible on Medium.

Medium initially solved the issue by introducing channels which was a great solution at the time. With success they reached thousands of stories per channel; each of them became saturated, and Medium started to have less user engagement — the stories in each channel became more and more heterogeneous and the information was not qualified enough.

Therefore, they introduced Tags. This allows users to click into more targeted categories, while also giving authors more exposure to their work.

Increase user engagement

By specializing and qualifying the information, Medium aimed at getting users to engage more and ultimately increase member retention.

To measure the success of the way they’ve implemented the new feature they need to monitor:

  • the number of tags created
  • the number of users following tags

The last one should increase with the number of tags created.

Tags will create a stream of stories that are more closely related, which will add business value. Medium should expect an increase within:

  • Scroll depth and time spent
  • Bounce rate
  • Number of clicks on tags per visit

Page per visit may lead to the wrong conclusion. Tags as well as the new streamed feature will make the content easier to access and via fewer pages — the number of click on tags per visit is therefore more relevant.

To optimise the tag feature the following parameters need to be looked at

  • how the engagement indicators evolve in correlation with the number of stories per tag; for example, I would monitor the engagement for all tags with 10–50 stories, 50–100, 100–200, 200–500, 500–1000, 1000+
  • Medium is converting existing channel themes as tags —They need to use it as a reference

Both of the above would tell what the optimal ratio is and how the tag feature is improving user engagement.


Next step ?

Help the user choose the right tag

The current Medium tag suggestion feature seems to be only semantic. Based on the successful outcome of the tag feature, they should focus on helping users to select the right tag so that related articles are better connected together.

This would contribute to increase user engagement and the number of users following tags as they become more relevant. With that new feature, Medium would be able to automatically push some tags forward so each tag would reach their optimal ratio of number of articles per tags per month

Tag Suggestions

I would build an algorithm that:
· Finds the stories that are related to my story depending on the content of the text and the tag I’m typing

  • Get all the associated tags
  • Score them depending on :
    - how many times they appear within the related stories
    - how closely they meet the optimal ratio of articles / tags / day
    - how many followers they have
  • Rank them
  • Return the top tags

This will improve the following situation:

If I write an article about codes, I would probably get more readers using a tag such as “Learning To Code” and “Learning” seems saturated. On the other hand, tagging “Learnings” would probably be useless.

Mock up To keep the clean look of Medium I would re-use the existing tag suggestion tool and make it appear as soon as the user clicks on a tag

Once the user starts typing, existing semantics suggestion results will appear first

This new feature combine with the semantic suggestion will give great ideas to writer and help them to find the right tag so that their stories reach the most relevant readers.


Failure?

What if the tag feature is not increasing my user engagement ? The feature could have failed for different reasons depending on my indicators.

Implementation of the feature

If the number of tags created remain low medium should first do small changes on the UX design and do some a/b testing by positioning the tag field differently, make it more visible or even envisage a tool to add tags when writing inside the in-line editor.

If the number of users following/clicking tags remained low; as well as doing UX a/b testing, they should consider adding tags to the stream to make them more visible.

Engagement

If users are adopting the new tag feature but the engagement indicators are not increasing; They need to analyse how the new tags are performing compared to “channels tags”. If they are not generating more engagement; depending of the number of stories per tag, this would tell if they need to optimise theratio or if they need to ensure that tags remain generic or more specific.

The tag feature can have the opposite effect of the channel feature — it’s diluting the content and there are not enough articles per tag.

Medium can increase or decrease the number of tags per article to restrict or expand user choices. Also, the feature suggested earlier will help users to choose the right tag and they should make sure that suggested tags are less or more specific.

Qualify and link the information

If there is no difference of engagement between tags from my initial channel and new tags and no improvement of user engagement in general then tags may not be the right answer. There are other ways to link content such as suggested articles or hashtag. The first one is easier to implement on a selection of users (beta testers) whereas hashtag requires a high number of users to make the functionality relevant.

Hypothesis

Finally; if all of the above fails, Medium should review their hypothesis to drive user engagement: Early adopters may have been fully engaged in Medium and ready to read long articles without knowing the subject (because they are generally quite enthusiastic); while a recent, wider, audience may want to know a bit more before starting an article. Offering a summary (written by the user) of the article (and not necessary 140 characters) in the stream may give a better overview of the article and increase user engagement.


Question

What do you think ? would that be your approach ?

I would love to hear some people from Medium about that !