How do we pick which newsletters to send you?

If you're reading this, I'm deeply sorry for whatever terrible newsletter we forwarded you. The Sample is quite young and we're actively working to improve recommendation quality. Thanks for your patience.

With that out of the way, our algorithm mainly uses collaborative filtering. This means we compare your feedback ratings to those from other users, and we source recommendations from people who have similar tastes. Since many people have a wide variety of interests, this tends to produce recommendations that are similarly diverse. In general we consider that a Good Thing, but a side effect is that when you get a bad recommendation, it often looks completely random.

We combine that with content-based filtering, which is based on the interests you select when you sign up. We compare those words to the crowdsourced tags that other users supply when they rate newsletters, and we try to guess how similar each newsletter is to your interests. Newsletters that we think are similar to your interests get a higher chance of being recommended.

Besides that, we do a handful of things to boost long-tail recommendations. We try to help you find the hidden gems instead of just whatever's popular.

Untitled

One-click subscribes

This is our most amusing feature. Whenever you click "Subscribe in 1 click," we get an email alert. Then at some point, typically a few hours later, I or my wife will paste your email address into the signup form for you. (Sadly this does mean you'll still have to click a confirmation link if the newsletter uses double opt-in.) Besides making things a bit more convenient for readers, this also gives us a rough measure of how many people subscribe to the newsletters we forward, which is incredibly helpful.

Untitled

Sponsored newsletters

10% of the emails we forward are sponsored. These are picked by a combination of how much we think you'll like the newsletter and how high the newsletter author set their bid price. Sometimes sponsored newsletters will be sent close together, but over time it'll average out to 10% (it was easier to write the code this way). Whenever a newsletter is sponsored, the email will say "Today's newsletter (sponsored):"

Feedback, questions and updates

You can tweet at @the_sample_umm. I tweet about what we're working on and other frivolous things. You can also join our Discord server. Most people there are newsletter authors, but everyone is welcome. And finally, you can always reply to any of the emails from The Sample, or you can email me directly at [email protected].

About us