Duolingo's Journey: From Gamification Misfires to 350%+ User Growth

To set some context, Duolingo was launched as a private beta in November 2011, and in June 2012, they opened it to the public with 300k+ people on the waiting list. They raised $3.3M (Series A) in October 2011, even before their Beta launch.

They cracked the 0-1 growth and even the 1-10 growth with a unique two-sided business model. The plan was to provide free language learning to users while relying on those same users to translate texts for paying customers. However, this translation service failed to gain traction, with only two customers signing up between 2012 and 2014.

From 2014 to 2016, Duolingo had no revenue and relied on the venture capital they had raised to continue developing the product. 

‍

‍

In 2017, they introduced Google and Facebook ads and launched an ad-free subscription option, generating $13 million in revenue by the end of the year.

In 2016, Duolingo also started working on the Duolingo English Test (DET) as another potential revenue stream, aiming to compete with the TOEFL by offering a cheaper and more accessible alternative.

Despite these efforts, by 2018, Duolingo found itself in a challenging position, with Daily Active Users (DAUs) growing only in single digits. This stagnation prompted the company to enter a new "growth" phase, where they focused on reigniting user growth and engagement.

The journey began with a closer look at gamification, a strategy that had proven successful for many mobile apps.

Phase 1: Gamification Misfires

Duolingo's initial foray into gamification was led by a dedicated team tasked with improving user retention. Drawing inspiration from the success of mobile games like Gardenscapes, the team implemented a "moves counter" feature, which aimed to replicate the engaging gameplay experience by limiting the number of moves users could make to complete a lesson.

However, this initial attempt failed to deliver the desired results. The "moves counter" feature did not translate well to Duolingo's learning context, as it did not account for the fundamental differences between the two apps. 

While Gardenscapes required strategic decision-making to overcome obstacles, Duolingo's language learning process was more straightforward, with users either knowing the answer or not. The added complexity of the "moves counter" felt like an unnecessary hindrance to the learning experience.

Duolingo learned a valuable lesson from this misfire: the importance of intelligently adapting features to their product's unique context. The simple insight was that simply copying successful elements from other apps without considering the underlying differences could lead to a poor user experience.

‍Phase 2: Referral Inspiration

After the gamification setback, Duolingo turned its attention to user acquisition. The team sought inspiration from Uber's successful referral program and launched their own system, offering a free month of Duolingo Plus as an incentive for users to invite their friends.

Despite the initial excitement surrounding the referral program, the results were underwhelming. The main issue was a misalignment of incentives with Duolingo's most engaged users, who were already subscribed to the premium plan. As a result, the users most likely to spread the word about the app were unable to benefit from the referral incentive.

This experience helped them understand the importance of aligning incentives with target personas and use cases. They realised that a one-size-fits-all approach to referral programs might not be effective and that they needed to consider the specific needs and motivations of their most valuable users. 

Translation Framework

The disappointing outcomes of the gamification and referral initiatives prompted Duolingo to take a step back and reflect on its failures. They recognized the need for a more systematic approach to adapting ideas from other products to their unique context.

To address this, the team developed a feature "translation" framework, which provided a structured process for evaluating the potential fit and impact of borrowed features. This framework helped them assess the underlying principles behind successful features in other apps and determine how they could be tailored to enhance the Duolingo experience.

Jorge Mazal (former CPO of Duoling) explained that this framework was basically around 3 parameters: 

1. Why is this feature working in that product?
2. Why might this feature succeed or fail in our context, i.e. will it translate well?
3. What adaptations are necessary to make this feature succeed in our context?

Phase 3: Modelling the Way Forward

Armed with a renewed focus on data, Duolingo's data science team began exploring engagement segmentation models to better understand user behaviour and identify growth opportunities. They created a custom Duolingo-specific model that allowed them to run growth simulations and predict the impact of various initiatives on key metrics.

Through these simulations, the team discovered that the Current User Retention Rate (CURR) was the highest-impact lever for driving growth. This insight led to the formation of a dedicated Retention team, whose primary goal was to optimise CURR and keep users engaged with the app.

The Retention team became central to Duolingo's growth efforts, with CURR as their North Star metric. Gina Gotthif, Duolingo's VP of Growth (and employee number 20), shared that this focus enabled a targeted approach to growth, prioritising features like leaderboards, push notifications, streaks, etc.

Leaderboards became one of the first successful experiments. Inspired by Zynga's FarmVille 2 leaderboards, Duolingo adapted the concept to fit their learning context. Users were matched with others who had a similar level of engagement, and the top players moved up to a higher league the following week. 

‍

This feature had a significant impact, resulting in a 17% increase in learning time and tripling the number of highly engaged learners.

Push notifications were the next motivation factor for the users and they were able to drive substantial gains in engagement and retention.

To avoid the pitfalls experienced by companies like Groupon, they set a guardrail to protect users from over-notification. They even did extensive A/B testing and implemented bandit algorithms.

Through analysing their notifications over time they learned that not only did some of the notifications work better than others, but the effectiveness of notifications was also dependent on the language and the kind of notifications (reminders, leaderboard-related, etc)

Duolingo used the forgetting curve to determine how often to repeat practice reminder notifications to maximise their effectiveness in motivating users to continue practising.  

The forgetting curve describes how information is lost over time when there is no attempt to retain it. 

In the context of Duolingo's AI algorithm, the forgetting curve helped them determine the optimal spacing between repetitions of the same notification.

The team also discovered the power of habit formation through streaks. They found that users who maintained a 10-day streak had significantly lower churn rates. By optimising notifications, UIs, streak freezes, and rewards, they encouraged users to form lasting language learning habits. This led to some impressive results, with one user even maintaining a streak of over 1,400 days.

Streaks have emerged as a significant driving force for many education to motivate users to maintain daily usage, but it seems to be working extremely well for Duolingo. 

‍Community-Driven Growth and TikTok Success 

In addition to these product-led growth strategies, Duolingo also leveraged the power of its community to drive engagement and expansion. They empowered passionate volunteers to create language courses using the Duolingo Incubator, fostering a sense of ownership and investment in the platform's success.

Duolingo also established online forums where learners could engage with one another, share knowledge, and provide peer support. 

They have adopted a robust review system to continuously improve the quality of its language courses. One key component of this system is the Report Quality Estimation Tool, which allows users to report concerns and issues they encounter while using the platform. 

It created a sense of community and engagement among its user base.

Recognizing the importance of social connections, Duolingo also organised offline events where users could practise speaking together face-to-face. In 2019 alone, they held an astounding 600 events per week across 113 countries, spanning 820 cities and encompassing 34 distinct languages.

More recently, Duolingo has found remarkable success on TikTok, among other channels, thanks to its playful and engaging brand personality. With over 8 million followers, Duolingo's TikTok presence is characterised by its iconic green owl mascot, Duo, and its humorous, self-aware content.

‍

According to Zaria Parvez, Duolingo's Global Social Media Manager, the key ingredients to their TikTok success are:

1. A memorable brand personality and voice that doesn't take itself too seriously and makes learning fun.
2. Staying on top of the latest pop culture trends and putting their own twist on them.
3. Authentic interactions with fans through comments and video responses.
4. Collaborations with other brands and creators to expand their reach and create buzz.

The Multiplier Effect 

Duolingo's relentless focus on product and growth optimization, combined with its community-driven initiatives and social media success, has led to remarkable results. The company's CURR increased by 21%, translating to a 40%+ reduction in power user churn. DAU grew by 4.5x, and the number of users maintaining a 7-day streak tripled.

The improved user base size, engagement, and monetization propelled Duolingo to a successful IPO, driven by record-setting growth. As of 2022, Duolingo boasts 56.5M active users, a quarterly revenue of $96.1M, and a valuation of $3.7B – a testament to the power of its growth strategies and community-focused approach.