The Sample publisher report (21 Feb)

New stuff from last week:

  • New web UI. After people rate or subscribe to a newsletter, we now take them to a web page which shows the previous newsletters we’ve sent them. The main page shows newsletters which haven’t already been rated/subscribed, and there’s a “History” page for the ones that have been. (Screenshot)
  • Emails now include a “Read another issue” link under the rating buttons, which lets you see up to three previous issues. I added this because we’ve had several people mention that they’re often on the edge about subscribing, and being able to receive a few more issues might help. Originally I was planning to add some kind of button which would cause the next two or three issues to get forwarded to you as they come in—however this was much easier to implement (which is why I finally got around to it), and hopefully it helps. (Screenshot)
  • Fixed a bug which made it so new users’ preferences (i.e., the checkboxes you select when you sign up via the landing page) were completely ignored on their first email. Yeah...
  • It’s been very helpful to have regular contact with all of you via these emails/your replies/Discord. However I don’t have as much of a relationship with regular readers. I’d like to fix that. So now as I write these publisher reports on Mondays, I’m also going to write a condensed version with stuff relevant to readers as a Tweet thread. (Here’s the one for this week). I’ll include a link to it at the top of the emails on Tuesday. Hopefully this will help me get better intuition about what to do with the product!
  • We made a tweak to how organic forwards are selected. We do some “wealth redistribution”/”popularity smoothing”—if we’re picking between two newsletters for a particular user and we think the user will like them both about the same, then we prefer whichever one has received fewer forwards so far. i.e. we want to spread organic forwards around as evenly as we can without hurting the user experience/reducing the total number of 1-click subscribes. However we think the popularity smoothing may have been too aggressive, which would cause high quality newsletters to not receive as many organic forwards as they should. We’ve turned it down a notch for now. At some point we should wrap it in an A/B test so we can measure how many 1-click subscribes we’re losing because of it and dial it up/down accordingly.

In progress:

  • We’re making some more algorithm changes that should improve performance for readers who come from our Facebook ad.
  • Last week I mentioned we wanted to do an A/B test where we send people a list of 5 newsletters instead of one-newsletter-per-email, and just send two emails per week. A few people have mentioned they like the current format. OTOH I’ve also had positive feedback about having a list of newsletters on the new web UI. I guess this is what A/B tests are for 🤷‍♂️. BUT, don’t worry: either way I’ll leave in a setting so you can control how many newsletters you get per email (and if you leave it at “1” then nothing will change). I suspect that this’ll perform best as part of a reactivation campaign: we start people out on one newsletter per day, but if they drop off, we switch them to a list twice a week and send them a heads up about it.

Backlog:

  • Might start nagging readers to set preferences if they didn’t select any checkboxes when they signed up. e.g. maybe put a little banner in the web UI.
  • Maybe we can start experimenting with the web UI. e.g. right now it shows a reverse chronological list, so on the main page you’ll see the five most recent newsletters you’ve been forwarded. Maybe instead we should randomize it, to help resurface recommendations you got long ago? We could even have them show the most recent issue instead of whatever issue was forwarded to you.
  • Still planning to do publisher console improvements once I get through reader experience improvements.

Finally—we’re pretty excited about how our FB lead ad is going. Previously Facebook was optimizing it for number of new signups, without taking into account how active they are after signing up. However, as of last week, the “conversion optimization” option has finally been enabled. When our readers from FB subscribe to any newsletters we forward, we report back how much revenue we get from that 1-click subscribe. With conversion optimization enabled, Facebook will do a better job of finding people who actually engage with email. We’ve put about 500 bucks into the new setup so far, and the results are looking really nice. For example—one of our main metrics is “day one conversion rate,” which is the percentage of people who subscribe to the very first newsletter we send them. For our previous ad, it was usually 6% - 8%. With the new ad it’s already over 10%. Cost-per-signup is higher on the new ad, but it’s not terrible, and it’s still improving.

So, with that improvement and the product changes we’ve been making, we’re crossing our fingers that we’re close to making the pipeline efficient enough to scale up ad spend profitably.

Speaking of which—if you want to help us get this going faster, feel free to try out paid forwards! At some point we may need to put more marketing work into that... perhaps some content marketing, e.g. a post that talks about growing newsletters with FB ads and how it’s easier/cheaper/requires less scale if you just go through us. 🤷‍♂️. Let me know if you have any ideas!

Jacob

Published 21 Feb 2022

I write an occasional newsletter
about my work and ideas.

RSS feed · Archive

𝔗𝔥𝔦𝔰 𝔰𝔦𝔱𝔢 𝔦𝔰 𝔭𝔯𝔬𝔱𝔢𝔠𝔱𝔢𝔡 𝔟𝔶 𝔯𝔢𝔠𝔞𝔭𝔱𝔠𝔥𝔞 𝔞𝔫𝔡 𝔱𝔥𝔢 𝔊𝔬𝔬𝔤𝔩𝔢 𝔓𝔯𝔦𝔳𝔞𝔠𝔶 𝔓𝔬𝔩𝔦𝔠𝔶 𝔞𝔫𝔡 𝔗𝔢𝔯𝔪𝔰 𝔬𝔣 𝔖𝔢𝔯𝔳𝔦𝔠𝔢 𝔞𝔭𝔭𝔩𝔶.