The email newsletter in 2017 is a core product for national newspapers, digital media brands, platforms, and local news because it’s simply the best available channel to directly reach users on a daily basis.
But how to fill a newsletter with content is an open question. Do editors or algorithms work better? Is a personal touch better reading than a list of headlines? Do you empower users to choose their interests or deliver a curated product?
Over the last year Tracy Clark has run an experiment to help answer these questions in partnership with the Reynolds Journalism Institute and the Austin American-Statesman. We published her findings on Friday, but I had more questions for Tracy so I called her up. Here’s our conversation.
RELATED: Check out our DigitalEd online training: How to Launch a Killer Newsletter!
How does your research — in which personalized newsletters performed better — fit with the trend of hyper-voiced newsletters, with an editor and a point of view?
Tracy Clark: That’s an interesting question to start out with. Because with the whole Trump-questioning-media thing, it’s a debate whether journalism is the voice of the media or the voice of the people and who’s telling the truth. So that’s why I almost felt with people mistrusting media these days or the hype around that, that the personalization is important. They’re able to get the information they want so they have the power. They’re actually taking ownership over their news so hopefully getting a more trustworthy relationship with media.
Going to your specific questions, there’s definitely that trend of publishers today having dozens of different newsletters on different topics written by specific editors. That’s why I want to figure personalization out because the more newsletters these people are deploying the more time and effort to get it all done. They’re trying to deploy these different newsletters on different topics, which in essence is personalizing the newsletters to try to get more niche audiences, what if a platform could automate that?
How do you know what newsletter metrics to measure and optimize for?
Clark: Obviously you want to optimize your open rate, maximize it. You want to optimize your click-through rate, you want to decrease your unsubscribe rate and your bounce rate. Because your bounce rate implies you have a really bad list of email addresses, which harms you and your deliverability score. Those are the ones we looked at for this study in particular. From there you could do a lot cooler things, like click-through rate to purchase path.
Do those metrics move together or do you have to choose?
Clark: If you’re delivering good quality content day-in and day-out, your open and clicks should be strong and positive and your unsubscribes should be decreasing.
For our test we didn’t manipulate the subject line, which was just Statesman News for You and the date. But you can certainly personalize it or try to get reader attention which would drive up open rates. But if it’s a really market-y title just to get attention but the content inside isn’t good, then your click rate would go down.
Your test got an open rate twice as high using the same subject line format as the control?
Clark: It’s kind of cool, right, when you think about it? Literally, we tried to do an apples to apples comparison. Their email subject line was Statesman Midday Break and the date. We did Statesman News for You and the date. The main takeaway from that is that people were looking for that email because they’d built up an understanding that they’d get valuable content within it. Once they saw it, they had a brand affinity to it and over time they would then be willing to open it more so than the other one.
Are News for You readers already more engaged because they’ve shared their interests?
Clark: Totally. Since they took the time to fill out the form they for sure had more investment into the product. The interesting thing we found because of that, our unsubscribe rate was actually pretty significant, percentage-wise. You look at that and you’re like, bummer. But that tells me immediately that there are people who went in and filled out the form and didn’t immediately see results that matched their preferences so they unsubscribed. They wanted immediate satisfaction.
With the algorithm, we were just pulling content that the Austin American-Statesman published. If it didn’t find matches, it would just give you recently published content because we had to fill the email. The other option if you have a personalized newsletter and you don’t have anything that matches their interests, just don’t send anything. Which is probably the smarter way to do, but for our comparison study, we had to send something every day.
That’s my hunch why our unsubscribe rate was high. People didn’t put in enough interests or they didn’t give it enough time to see the personalization algorithms actually find content over a few days.
Stepping back, there’s a lot of talk about breaking out of filter bubbles. It’s a philosophical question, but are you excited about personalization? How do you think about that?
Clark: I know people are concerned. They say the whole reason of journalism and the role we play in society is to raise people’s awareness and knowledge about different topics. But in my opinion, media is always going to play in that space. That’s the 30-minute nightly newscast and the NPR broadcast. You’re already getting those quick-hit updates. The homepage of a news site, for example. The main things will come across your social feed if they’re big enough.
But for any push mechanism where you’re actually intruding into someone’s life via a mobile push or an email, where you’re coming to them instead of them coming to you, I really do feel like with technology the way it is, where everyone wants personalized food and travel arrangements, why is media not providing that personalized opportunity and capitalizing on that? So I don’t see it as a fear factor or negative at all. It’s just a component that should be added into a media strategy.
I hear you differentiating between different kinds of news products and seeing personalization as a tool to serve a specific audience.
Clark: When I first built the platform it was a web interface, a mobile app and an email. I think there’s a lot of testing going on to figure out the right channels to optimize. I think there’s an opportunity for a personalized news homepage, just like your Amazon.com is personalized. Why would I put the New York Times as my homepage if it’s just the same homepage everyone else saw?
I would push the news industry to actually personalize other channels, not just what I mentioned about email newsletters and push apps. But it’s a difficult balance. I do agree there needs to be some oversight of the people who do want to see everything, to get away from that tailored view. The tailored view is not for every channel, every publication, all the time.
Talk about the value proposition here. Can personalization lead to revenue growth?
Clark: The way I see it, and the way I pitched it to Austin and Cox [Media Group], you have all these properties across the U.S. If I live in Austin and I’m a huge Braves fan, I’m not going to subscribe to the Atlanta Journal Constitution just to get Braves content, nor am I going to remember to look at AJC.com because I don’t live there.
But if I tell you I’m interested in the Braves and I live in Austin and you pull in Braves content from AJC.com, you’ve given me a better experience and provided a value-add for me. Depending on how you can price that out, if it’s per-article or otherwise, why aren’t newspaper chains aggregating all that for their users and creating a new product?
Is there a feedback loop built into this process for editors to become better editors?
Clark: We could see analytics based on each person every day. So taken at the aggregate level you can see what people are actually interested in. The homepage editor is always going to say that they’re choosing the stories based on quality and importance, but maybe showing engagement trends could help other email newsletters that aren’t personalized. Especially for story placement. Because in an email newsletter what’s placed at the top versus down below, there’s a lot of strategy that should be going on there. I’m sure at different levels it is, but you can always take in engagement data to improve your suggestion for those layouts.
Finally, the promise of personalization seems like it’s been around a long time. Is it here for real now?
Clark: I sure as hell hope so. I started thinking about this in 2007 when I was at the Roanoke Times and it was driving me mad that the product was so one-size-fits-all, delivered by hand. It was like, we have to do better than this.
Now all the big technology partners have the code and the technology to personalize. News publishers are really concerned about giving up their content to these technology players, do they really trust them? Well, then do it yourself. Tech companies aren’t willing to wait. With all this great content using some data and some basic code, you can deliver a better product.
If anything it’s a kick in the butt to publishers and editors that says, experiment. Let’s really see the data and see how it works. With this research, my assumption was right: if media would be more willing to use their content in different ways than they do day-in and day-out then other opportunities could arise that they never would have thought of before.
And I’ll keep pushing it.
Jason Alcorn (@jasonalcorn) is the Metrics Editor for MediaShift. In addition to his work with MediaShift, he works as a consultant with foundations, non-profits and newsrooms.