GTM Sauce

Duolingo · B2C · language learning app

Use the end-of-lesson ad slot to advertise your own paid plan

Guest says house ads beat programmatic ads; no numbers

Unknownupsell-expansionpaywall

What they did

At each lesson end an ML model picks between a programmatic ad and an internal ad (Super, Max, Duolingo English Test), maximizing LTV with purchase likelihood as the objective. If a user ignores the Super ad about three times it backs off. Internal ads are character-driven videos that explain the subscription features.

What happened

Guest prefers them to programmatic ads as a better experience; video is one of the few places to tell the subscription story.

In their words

Why Your Free Users Are Your Real Growth Engine – Cem Kansu, Duolingo CPO

Play the episode

Cem KansuAnd we decide, like you said, if we should show you a programmatic ad versus one of our internal ads, which could be an ad for a subscription, or it could also be an ad for our other products like the Duolingo English test. But we call those basically internal ads that we decide we've produced and it is a Duolingo product.

Cem KansuMajority of them are going to be either Super Duolingo ads or Duolingo Max ads. The decision logic is mostly now controlled by ML actually, but the ML uses various signals that is trying to figure out LTV maximization at its core and it is looking at every signal that we feed it that could be an indicator for someone that is more likely to buy Super.

Cem KansuOr it also balances out, of course, if, for example, it's shown you the same, or it doesn't really show you the same Super ad three times, but it's shown you Super ad three times that clearly you're not interested in, it backs off and shows you other ads.

Cem KansuSo it is basically built to optimize LTV with a likelihood to purchase being the objective function.

Get tactics like this every Monday

New tactics from the week's founder interviews, each linked to where it was said.

More from this episode

Tags: internal-ads, ml, ltv, video-ads