Twitter, one of the most popular social media sites in the world, was founded in 2006 by Jack Dorsey, Noah Glass, Biz Stone, and Evan Williams. Six years later, in 2012, Evan Williams realized that the platform they created wasn't giving people the ability to create quality content.

At the time, tweets were sent out in 140-character messages. That's hardly enough space to write something meaningful, impactful, or of high quality. Williams had such a problem with this that he quit Twitter and started a blogging platform called Medium, with the goal of creating higher-quality content than what was available on Twitter.

To pull this off, Medium decided to ignore user growth and unique website visitors and focus on the metrics that the blogging platform claims really make a difference. Medium's focus was on metrics related to website engagement — things like time on site, time spent reading articles, and number of repeat visitors.

This, of course, raises one huge question: Is Medium actually pulling this off?

In a word, no. By using Nacho Analytics, we were able to examine Medium's website-engagement stats and compare those numbers to Twitter's. It turns out that Twitter is beating Medium in nearly every engagement category.

1. Website Engagement: A bad start for Medium

The best place to start when comparing website engagement between two companies is two simple but crucial website metrics: bounce rate and average session duration.

Bounce rate is important because it gives you a quick snapshot of your “one and done” users. If someone lands on a page and quickly hits the back button, you can safely assume that they didn't find something of interest or value on your website.

Average session duration is also important for Medium and Twitter because they both have the same goal: keep people around as long as possible. Since Medium's goal was to create a higher-quality website, you'd expect people to have a long average session duration on the website.

So, how did they stack up?

  • Bounce Rate: 34.04% for Twitter vs. 58.21% for Medium
  • Average Session Duration: 04:20 for Twitter vs. 01:40 for Medium

This wasn't a great start for Medium, the platform built on quality. We expected Medium to outperform Twitter on at least one of these two metrics. Unfortunately, they didn't pull that off.

Round 1 winner: Twitter, because it has a lower bounce rate and a higher average session duration.

WHAT DOES THIS TELL US?

Medium's bounce rate is higher than Twitter's, but according to data from custommedialabs, Medium's bounce rate falls in an acceptable range for a website of their type.

Medium's average session duration, on the other hand, is low. Databox reports that marketers typically see average session durations in the 2- to 3-minute range.

Since Medium didn't outperform Twitter on either of those metrics, the early results point to Medium being a less useful platform for visitors.

2. New vs. Returning Users: A sign of life from Medium

If your website is producing quality content, then you will have a large percentage of returning visitors. Website visitors want quality content. If you supply it, they're likely to keep coming back for more.

This holds especially true for a website like Medium, which isn't stuck producing content on one topic or for one industry. Their wide range of topics makes them comparable to a newspaper — something that people buy and read every day.

The new vs. returning visitors metric looks pretty good for Medium:

  • Twitter: 58.6% New vs. 41.4% Returning
  • Medium: 54.8% New vs. 45.2% Returning

Medium has a healthy percentage of new visitors, which make up 54.8% of their website visitors. But, more importantly, 45.2% of their visitors come back for more. That's a very good ratio to see. Twitter's metrics are also quite good, with a split of 58.6% new and 41.4% returning.

Round 2 winner: Medium, because users return at a higher rate.

WHAT DOES THIS TELL US?

Medium might be on to something. The fact that such a large percentage of their visitors are returning suggests that the company is publishing quality content on their platform — the type of content that people want more of.

3. Returning User Stats: Twitter keeps 'em coming back

We need to dig a little bit deeper into the returning users. Returning users should be the highest-quality users because they've been on the site, they know what to expect, and they're coming back for more.

If Medium is really producing a higher quality experience for its users, then the most engaged users — returning users — should have great engagement stats.

For this, we'll focus on three stats: bounce rate, average session duration, and sessions per returning user. We've covered the importance of the first two previously, but the third one, sessions per returning user, is new. The reason we're looking at that stat is because it helps us figure out how many times a user returns. The more a user returns, the more engaged they are, which means they're finding value in the website:

  • Returning Visitors' Bounce Rate: 33.57% for Twitter vs. 54.52% for Medium
  • Average Session Duration: 04:17 for Twitter vs. 01:57 for Medium
  • Sessions per Returning User: 10.3 for Twitter vs. 3.8 for Medium

Round 3 winner: Twitter, by unanimous decision. Twitter's returning visitors have a lower bounce rate, spend more time on the website, and return more often than Medium's returning users.

WHAT DOES THIS TELL US?

Medium got destroyed in this round, and, honestly, we're not surprised. Twitter has the benefit of never-ending scrolling. You can go to Twitter's homepage and scroll for hours. Each entry is one tweet, and each tweet asks for the commitment of a few seconds to read, so jumping into more content is an extremely low-friction act for every reader. Even if links are attached, it's easy to scroll through Twitter to cover one topic from hundreds of perspectives or hundreds of topics from a few perspectives.

When we first published a version of this article on another page, Medium's articles ended with a footer of topic tags, giving few opportunities to explore actual article options. We suggested at the time that Medium take notes from Twitter.

Since then, Medium has updated its end-of-article actions. Readers can choose from related articles or more from the same author. It's less abrupt than Medium of before, showing that they are a site that embraces testing and user experience adjustments.

But Twitter still wins this one.

4. Engagement on Top Posts: Twitter's bounce-rate domination

Twitter has a huge advantage over Medium: site structure.

Twitter is, essentially, one page. A visitor can go to twitter.com and scroll forever. Medium, unfortunately, doesn't have this feature. If a visitor goes to medium.com, they still have to navigate to another page to read anything. The end result is that there are more opportunities for a visitor to get bored and leave the site.

Each website has a different goal for users. Twitter's goal is to keep people on the homepage as long as possible. Medium wants to get people off the homepage and looking at specific posts for as long as possible. That's why the metric we need to compare for the two companies is the engagement stats on the most visited Medium posts versus Twitter overall.

Our theory is that even though Medium is losing on overall site-engagement stats, they should win when it comes to the performance of their most-visited posts.

To figure this out, we looked at the top 50 Medium posts from the past 30 days, sorted by page views, with engagement stats alongside. Here's what we discovered:

  • Average time on page for Medium T50: 02:09
  • Average time on page for twitter.com: 01:39

That's a good start for Medium. Their time on page is 30 seconds longer than Twitter. Now, let's look at bounce rate:

  • Average bounce rate for Medium T50: 52.12%
  • Average bounce rate for twitter.com: 10.33%

Twitter's bounce rate is so low that it's almost unbelievable. Medium's T50 bounce rate of 52.12% isn't bad either.

Round 4 winner: It's a mixed bag, but only because Twitter's bounce rate is unbelievably low. We'll award half a point to each platform.

WHAT DOES THIS TELL US?

It seems like we've finally found some signs of quality from Medium. The site's overall stats are likely dragged down by the fact that anyone can create a Medium post — meaning that a lot of the posts on that platform are low quality.

You could make the same argument for Twitter, but the difference is that Twitter has the benefit of appealing to short attention spans. Read a tweet you didn't like? Keep scrolling, you'll find something else of interest in a second.

The final score

Twitter won rounds 1 and 3. It also picked up half a point for having a lower bounce rate than Medium in round 4. That gives Twitter a final score of 2.5.

Medium didn't fare so well. It won round 2 and split round 4. Its total score was 1.5.

Really, no matter how we split the data, engagement stats for Medium were almost always worse than Twitter. This suggests that Medium's founder, Evan Williams, still hasn't achieved his goal of creating a product that is objectively better than Twitter. Considering that Medium is still struggling to figure out a path to profitability, it might run out of cash before it can ever create a better product.

Update| A word about our access to the analytics and experiments shown here:

This original article ran in support of a marketing service called Nacho Analytics. At the time, we had access to behavioral data about how people interacted with websites, and we visualized it through the Google Analytics tool. All of that information has since been removed, and the service is no longer available.