As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting — plus creative upscaling tools that transform existing assets into CTV-ready video ads. Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

In 2018, after years of explosive growth, Duolingo's Daily Active Users (DAUs) stagnated to single-digit YoY growth rates. What happened next is one of the most compelling case studies on how to align data, product, and marketing that’s publicly available. As proof, between 2018 and 2024 Duolingo grew their DAUs by 900% with growth rates higher in recent years than previous ones (a very rare feat).

To get there, Duolingo did 4 things arguably better than anyone else:

Let’s dig in and learn the language of growth!

… see what I did there?! Duolingo.. language…okay I’ll shut up.

When Duolingo began its journey to solve slowing growth, they did what most companies do. They jumped into experiments and product building. First, they tried gamification:

However, to speed up iterations they borrowed what other companies were doing and tried to apply it to Duolingo. The result? From the mouth of their CPO at the time “The first attempt to reignite growth through more gamification resulted in a dumpster fire.” The strategy and vision effectively imploded and was dismantled.

Then, they tried referrals. Referrals is always a go-to for companies because there's the assumption that because it works for other companies, it'll work for you.

Duolingo took inspiration from Uber and tried to install their own referral program. While more successful than gamification, it only moved moved new users by 3% and that couldn’t justify the effort they were putting into it.

What happened next, though, is a Masterclass in how to use data to rally an organization.

The team took a step back and modeled their growth to better understand:

This led to the introduction of Duolingo’s growth model:

The model above classifies users into 7 mutually-exclusive user states per day:

Note: This is very similar to what I recommend for how to properly calculate MAUs (although Duolingo focuses on DAUs).

They then created rates and ratios between these states to understand how users were flowing in and out of segments.

For example:

By calculating these rates and ratios they were able to build a projectable growth model because you could use the last time period to predict the next one.

T_-1 → T₀ → T₁

Knowing they could build a projectable growth model they could also build a sensitivity analysis to see which metric they needed to focus on to move the needle. Specifically, they wanted to know what would happen to DAUs if each rate (CURR, NURR, etc.) grew 2% MoM over 3 years:

Improving the current user retention rate (CURR) had by far the biggest impact, so that's what they doubled down on. They created a new team, the Retention Team, with CURR as its North Star metric and got to work.