Hi, I'm Michael Langenmayr. You've probably never heard of me.
I've worked here at Daily Kos for more than three years, but I spend far more time reading blog posts than writing them. In fact, a pretty big chunk of my day is spent reading what you write.
I put together the Daily Kos Recommended email—a daily newsletter of our top stories that goes out to more than 1.1 million subscribers—and send it out every day. But most people don't know how or why it comes together the way it does, so I want to take a minute to explain that process.
Choosing the stories
I start every day at 8 AM ET by diving into your stories. The volume of stories written by staff and community members is too large for me to read everything, so I cull the stories using three metrics:
-
Pageviews—literally the number of times a page has been loaded on a computer, phone or tablet to be read—from social media (mostly Facebook), as reported by Google Analytics;
-
Pageviews from internal traffic; and
-
The number of recommends a story receives on the site.
Now, if you're reading this on a desktop, I'd like to point out that "pageviews from social media" does not refer to the number you see next to the Facebook badge at the top of this or other stories. That badge tracks how often a story is liked or shared on Facebook—not the number of people who are coming to Daily Kos to read that story from Facebook. This is an important distinction because the number of likes and shares a story gets is not always indicative of the number of pageviews a story gets from social media. We often have stories get liked or shared thousands or even tens of thousands of times but get very few pageviews from social media. (I would guess this is because people like the headline they see on Facebook, not because they're interested in reading the story.)
Pageviews from internal traffic is the number of times a page has been loaded by someone who came to that story from another page on the website. For example, this would count someone reading the front page who then clicks on and reads a story in the Recommended List.
The third metric—recommends—should be self-explanatory to the users of this site.
Using these metrics, I can usually compile a list of at least a couple dozen stories. Then, I cull the list again by reading each of these stories.
I look for several things when I read through these stories, and you likely won't be surprised by the list of things that disqualify a story for inclusion:
-
Conspiracy theories
-
Excessive typos and grammatical errors
-
Inappropriate use of other people's material
-
Lack of original material, i.e., stories that are just links to other sites
-
Old news, or stories on events that have changed since publication
-
Poor sourcing or reliance on dubious sources, which has become especially important with the rise of fake news.
My goal is to find 15 stories to test, though that’s not a firm number—some days it might be 12 and others 16 or more. I can usually find 15 stories just by looking at the highly trafficked and highly recommended stories as explained above. But, on days when I can't, I will include a large number of cartoons or low-trafficked/low recommended stories that may otherwise not have been considered. And on really slow news days (especially around the holidays), I will look back at some of my favorite diarists, like Major Kong and Haole in Hawaii, who write historical and photo posts that are always fun to read, no matter when they were published.
Testing
After the stories are compiled, the actual email work begins. All 15 stories are put into our email client, Action Network, with a “liquid code”—don't ask me what that is, I basically just copy and paste things at this point—that randomizes the email to determine which stories people most want to read.
Each story is tested as the top story in the email in groups of 10,000 people—so, with 15 stories, we send 150,000 tests—with the other 14 stories randomly ordered below it.
For example, let's say I have 15 stories, and let's call them Story A, Story B, and so on all the way to Story O. If you receive an email with "Story A" in the subject line, then Story A is the first link in the email with the other 14 emails completely random below it. So, your email may look like this:
-
Story A
-
Story O
-
Story F
-
Story B
-
Story H
-
And so on ...
Or your email may look like this:
-
Story A
-
Story C
-
Story G
-
Story O
-
Story I
-
And so on ...
In short, all 10,000 people in the Story A test group will see the email with Story A on top, but the rest of the email will look differently for each of them. The same is true for the Story B test group, the Story C test group, and so on.
We let the tests run for 90 minutes before putting together the final email, which goes to the remaining 950,000-ish people on the Daily Kos Recommended Email list. (In other words, if you get a test version of the email, you won’t also get the final version.)
The final email
After the tests have run, the final email is put together based on the number of clicks each test received, with the stories that get the most clicks on top. We have tried ordering the email a number of ways over the years and found that this is the best way to maximize the number of readers to the site from email.
For example, let’s say the tests come back showing this many clicks in each group:
-
Story A: 1,088
-
Story B: 1,166
-
Story C: 1,080
-
Story D: 994
-
Story E: 959
-
Story F: 1,040
-
Story G: 1,234
-
Story H: 1,080
-
Story I: 1,568
-
Story J: 1,172
-
Story K: 899
-
Story L: 838
-
Story M: 1,075
-
Story N: 831
-
Story O: 931
Story I tests got way more clicks that any other test group, so it's the first story in the final email. Story G tests got the second-most clicks, so it's the second story. And so on, until the final email looks like:
-
Story I
-
Story G
-
Story J
-
Story B
-
Story A
-
Story C
-
Story H
-
Story M
-
Story F
-
Story D
-
Story E
-
Story O
-
Story K
-
Story L
-
Story N
And that's exactly how the final email received by everyone on the DKRE list who did not receive a test email that morning will look.
How can you tell if you got a test email or the final email? Just look at the time stamp. Test emails are sent at 10 AM ET, so you should receive them sometime between 10 and 10:30 AM, depending on how fast our email program is working that day. I put together the final email at 11:30 AM and schedule it to send at noon, so anything you receive after noon is the final email, which is put together as described above.
I won’t be around this weekend, but feel free to ask questions in the comments. I’ll happily answer them when I come in to work on Monday.
Now, for reading all the way to the end, here's a picture of my new puppy.