Thread by @asherthoughts

I've been inspired by @wired's 5 level series, in which experts explain what they're doing to a kid, a teenager, a college student, a grad student, and a fellow expert.

As an artistic exercise, I'm going to explain the idea of attribution in growth marketing the same way👇

7 y/o: Have you ever used a coupon to buy toys? Did you find it in a magazine? It tells the toy store which magazine you read - because they use different coupon codes for different magazines. Websites use codes too - to say where you came from. They call those codes UTM tags.

16 y/o: When you buy Nike shoes through an Instagram ad, how does Nike know that ad worked? What if you saw other ads first? Or a friend told you about them first? It's really hard to figure out what *actually* got you to buy the shoes - to "attribute" why you bought.

College student: There are many ways to figure out what actually caused you to buy: there's last-touch attribution, which gives the credit to the last thing you saw. Linear attribution gives every step the same credit. Or you can weigh each touch differently with a custom model.

Grad student: Big companies use machine learning to make custom attribution models. You can use features like time of day, acquisition channel, and channel order to train on millions of purchases, and figure out what fraction of CAC to assign each touchpoint.

Growth marketer: We've had more success figuring out attribution by literally turning a channel on and off than by making a complicated model. We've also forced users to tell us how they found us in a form, right before they buy; surprisingly, that almost never hurts conversion.

I challenge more folks to explain things this way:

- @LauraDeming: explain a computational bio concept?
- @eriktorenberg: explain community-building?
- @ShivaKilaru: explain video production?