One of the core benefits of publishing on Medium is reaching a broader audience than elsewhere. By matching Medium readers with stories we think they’ll enjoy, we help writers connect with an audience they might never otherwise discover. These network effects surface in many ways, both directly within the system and via sharing beyond, so they’re difficult to quantify precisely. But one way to start is with referrer data.
When someone reads a Medium story, we track the preceding page that referred them, whether that’s another Medium page or an external website. This allows us to see if Total Time Reading (TTR) is arriving from within the Medium network (which we call “Medium Network TTR”) or beyond. Medium Network TTR is important to us because it represents those elusive network effects: the additional audience that writers can’t easily connect with anywhere else. And we can make a rough calculation based on it too; if a story has 20% Medium Network TTR, then that’s 20% above the other 80% the writer may have still achieved by publishing elsewhere. Effectively, that’s a 20/80 = 25% boost from the Medium network. It’s not a perfect calculation, but it gives us a pretty good idea of how well our network is performing.
I calculated what this looks like across Medium. For the most common type of story (older lower-traffic ones, described below), we see 39% Medium Network TTR on average. That’s a 39/61 = 64% increase. Or, in other words:
The Medium network expands total time reading on older, lower-traffic stories by 64%.
Overall referrers
To start, I looked at the overall referrer distribution on Medium. Here’s TTR broken out by referrer:
We can see that social media sites (including Twitter and Facebook) refer the majority of TTR. Clicks from Medium pages (such as collections, user profiles, and the “Read Next” section at the bottom of stories) represent another 11% of TTR.
But this first chart didn’t tell the whole story. To better understand what’s happening, I examined how this distribution differs across two important factors.
Traffic Level
Since Medium stories see very different traffic levels, they likely see different traffic distributions as well. I bucketed stories into one of three categories: top, middle, and tail. “Top” stories on Medium are the ones that see super high levels of traffic, often going viral on sites like Reddit, Twitter, and Facebook. “Middle” stories are the middle chunk that don’t necessarily go viral but still see decent traffic, often due to engaged audiences that continue sharing the story long after it is published. The “tail” end category comprises the rest.
Recency
The second major factor that affects traffic is the recency of a story. Since people typically share their stories immediately after publishing them, about half of a story’s traffic tends to arrive within a week of publication. So, I categorized referrer traffic by whether or not it’s arriving within that first week.
Break-outs by category
I now had three traffic-level categories and two recency categories to total 6 distinct categories. To start, I broke out the total distribution into the traffic-level categories:
And then I broke them out further by recency:
We can now see a clear trend: older, lower-traffic stories tend to see a much bigger percent bump in TTR from other Medium pages. That bottom bar shows 39% Medium Network TTR, which gives us the 64% boost described above. And since the data has a typical tapering-off pattern in total TTR, that 64% boost represents the majority of stories.
Relative totals
That 64% bump applies to the majority of stories, but it’s doesn’t represent the majority of site-wide TTR. To keep these numbers in perspective, I considered these same distributions in terms of total hours (vs. percent-of-total):
It’s from this view that we see the Medium network hard at work. Regardless of a story’s “viral” potential, our goal is to match readers with the stories they’ll enjoy most. That’s the middle group—great engaging stories that might not appeal to everyone but are perfect for the right audience if we can find them. And sure enough, in absolute terms these middle stories see the most Medium Network TTR, even more than the top group. This suggests some initial success in the network’s ability to circulate many decent-quality stories to the folks who will find them engaging.
Still, we want those green bars to grow. Medium Network TTR is important to us because it represents the audience that writers can’t easily find on their own. But it’s also the engagement we can affect most directly by continually improving our recommendation algorithm and product design. We’ll always see impressive traffic spikes for great, unexpected stories that go viral across the web. But as we keep improving, writers will consistently see more and more engaged readers arriving internally through the Medium network.
