Gina Gotthilf is VP of Growth at Duolingo – the wildly popular language app that brings gaming and learning together on a mission for ‘free language education for the world’.
We caught up with Gina to understand how Duolingo came to be one of the most popular language learning tools in the world, and what startups and other product teams might be able to learn from their success…
The growth toolkit
At Duolingo, every new feature is A/B tested before it’s spun out for all users.
“I think people see PR as old school or boring communications. I see it as a growth hack”
“There have been so many times in my career where I’ve been so disappointed or pleasantly surprised by A/B tests that I thought would be significant or that I thought would change everything” Gina says. “Some people are good at having a hunch on whether something is going to work out or not… but you can still get proven wrong.”
If you want to read more about Gina’s experience with split testing, check out this interview with First Round, where she shares the details of four of the most revelatory A/B tests she’s worked on at Duolingo.
Alongside A/B testing, one of the main tools in Gina’s toolkit is PR.
“I think people see PR as old school or boring communications. I see it as a growth hack because if you can get the story right, if you can get it to the right reporters and right market, the effect on your numbers is extremely visible,” Gina explains.
“Focusing on brand identity and a story that people can connect with is key to making a splash with PR… Our brand is about free learning education for the world. Externally, its very important to build out what it is we stand for, because then we’re no longer a language learning app, we’re a movement. I think that helped us create an identity that really fuelled word of mouth.”
“None of this content matters if your site or app are structured in a way that doesn’t allow search engines to find you.”
Duolingo has only recently started to use paid ads on Facebook and Google etc. “We’re taking a very engineering-driven approach, which goes against everything I’ve learned in my career” says Gina. “We have engineers who have PHDs in machine learning writing code to help us automate ad bidding, kind of like it works in the stock market.
“We did the same thing with SEO,” Gina explains. “We took an engineering-centric approach because none of this [content] matters if your site or app are structured in a way that doesn’t allow search engines to find you.”
Gina says that because Duolingo was flash based “there was no actual text anywhere for Google to find and our URLs weren’t slugified.”
Essentially, Duolingo hadn’t been built with SEO in mind, but once they were really focusing on rapid growth, it became time to re-examine the product in a way that made it super simple for people searching for language learning tools to find Duolingo.
Gamification and coming up with ideas
“I think that people see gamification as magic wand” says Gina. She explains that though Duolingo was originally built to feel like a game, the process of gamification is slow and steady. Every week, they run different experiments to add game-like mechanics to Duolingo and see how they affect human behaviour.
Not so long ago, Gina assigned each member of the growth team one of the top 10 games on the App Store to play for a week so that they could “find what’s really interesting and captivating, as well as what we didn’t like.”
“If we are going to redesign the signup screen, that can have a very high ROI because most new users are going to see it.”
At the end of the week, the team got back together to share a couple of slides on their opinions of the game to “identify things that work for other apps and games and think about what that could do for Duolingo.”
“We do a lot of brainstorming” says Gina, “and I try to make it as focused as possible.” This could be looking at particular screens and thinking about what could be redesigned, or asking the team to share their main frustrations with the app that haven’t been addressed yet. “Then we have a backlog of ideas. I would say the vast majority are not useful but even then, they help you think in a different direction or come up with other ideas.”
Gina says a key part of deciding which ideas to test is assessing the potential return on investment. “If we are going to redesign the signup screen, that can have a very high ROI because most new users are going to see it.”
Gina says a lot of people in the growth team wanted to reinvent the experience when a user completes a lesson on Duolingo. “It’s not very exciting and nothing really happens,” she admits.
The issue here is that the number of users who actually get to the end of a lesson is comparatively small. “Do we really want to focus on that small percentage that has already finished the lesson?” asks Gina. Wouldn’t they see a higher ROI by optimising the onboarding experience that is seen by so many more people? “It’s a trade-off, since users who reach the end are also your most committed users and potential brand advocates.”
Next is the “actual investment part, how much time would it take to make this happen? How many days of engineering, how many days of translation, do we need back-end implementation or not?” Ideas can then be prioritised based on doing the smallest amount of work to have the biggest impact on the most users.
Monetising a free app
Duolingo’s mission is free language education for the world, so “one of the main criteria we look at is how to monetise without hindering people from learning a language” says Gina. “Are we putting up a barrier for someone that has no money?”
“A lot of the time, something that is going to help us make money is going to hurt our retention”
Anything to do with monetising at Duolingo is down to the Midas team (as in King Midas and his golden touch) and Gina points out that “a lot of the time, something that is going to help us make money is going to hurt our retention.”
One example is streaks – a popular gaming mechanic where users are encouraged to play every day to keep their daily streak.
“From a retention perspective, I want everyone to keep their streak because having the streak is a great motivator” says Gina. “However, from a monetisation perspective, we’ve now added something called the ‘streak repair’ which allows users to pay if they lose their streak to ‘repair’ it and thus maintain their motivation to continue learning.”
This difference in goals for the Growth and Midas teams understandably creates friction. “They’re interested in having people losing their streaks” says Gina, “but I’m like ‘no that hurts our retention numbers!’ That conversation is always interesting.”
Currently, the answer to this conflict is compromise and understanding what’s best for users.
“There’s a lot of negotiation between the Growth team and Midas” says Gina, “like sometimes they say ‘if you let us launch this, we’ll do this and help you.’ We both have metrics that we really care about and I think that’s really important, you have two major agents in the company fighting for those metrics rather than letting them slide.”
In the future, Duolingo hopes to build an equation to understand the effects of monetisation on retention and vice versa, but for now, the focus is on user experience. “We don’t want to do anything that is going to hurt the user and is going to be frustrating,” says Gina.
How would all this be different for startups?
Duolingo has around 200 million users around the world so we asked Gina how her approach to growth would change if Duolingo had 50 or 100 daily active users.
“there’s no point putting a load of water in a leaky bucket”
“Oh I would change everything” she says, “you can’t really use A/B test to reach statistically significant results with that number of people. If there was a way to get 100,000 users so that we could then a/b test, even if it’s through paid ads, I’d maybe try that, but if it’s a product that does not retain anyone, then that’s also not a very scalable solution.”
Gina says all companies need to focus on retention but for early stage startups in particular, there’s no point “putting a load of water in a leaky bucket.”
Gina recommends tools like usertesting.com to see where real users struggle most with your product. “It’s interesting because often, when you build a product everything may feel extremely obvious to you, but that’s not necessarily the case to others.”
She also recommends the book The Design of Everyday Things by Donald Norman.
“I don’t want to make light of the work, but it’s not rocket science”
“I’m not a design expert at all but I really do think that this book helps you think about design in a different way… It’s often the case that engineers will build something, and if users don’t use it they’ll think ‘those users are so stupid.’ But the reality is that more often than not, the product simply isn’t simple or intuitive enough, which can be solved with smart user experience design.”
What advice does Gina have for aspiring growth hackers? “I don’t want to make light of the work, but it’s not rocket science. I’m a philosopher, I studied philosophy and am not an engineer or a designer. It’s pretty crazy and random how I got here, and I don’t think I’m a genius, I just work really hard, read as much as I can, hustle and try to figure it out.”
If you’d like to discuss your startup or product, get in touch with Simpleweb today.