Leave a comment
Get the GH Bookmarklet

AMAs

Hi, I'm Gina the VP of Growth at Duolingo, the most downloaded education app in the world. Over the past 3.5 years, I've helped build our user base from 3 to 150 million. I run the cross-functional growth team and am in charge of all marketing and communications at Duolingo.

Duolingo is a very interesting company growth-wise because we've never done any paid advertising. The company has raised $83.3 million from Google Capital, USV, NEA and KPCB and selected by Apple as "iPhone App of the Year" and by TechCrunch as "Best Education Startup." Our mission: free language education for the world.

Previously, I lead Tumblr's growth in Latin America and developed digital marketing campaigns for brands like Louis Vuitton and American Express. I have a Bachelor's degree in Philosophy from Brandeis University and am originally from Brazil.

Ah, and last year I had the honor of presenting to President Obama. That was the bomb. AMA!

Connect with me on Linkedin (linkedin.com/in/ginafrombrazil) or Twitter (@ginag)!

I will be live on Oct 6 from 930 AM PT for one and a half hours during which I will answer as many questions as I can.

  • TS

    Tom Ste

    10 months ago #

    Hi Gina,

    I am Thomas and run a health care startup which is a two-sided marketplace. We have the chicken-egg problem and have almost zero budget for paid advertisement.

    This is it: https://curafides.com/en/

    How would you start reaching out to potential clients and grow the business?

    What I plan is:

    + Ask health care bloggers for coverage
    + Send articles to press agencies and hope for the best
    + Facebook is our best channel to reach targeted audience

    Do you have any recommendations?

    Thanks.

    Tom

    • GG

      Gina Gotthilf

      10 months ago #

      Hi Tom! I’ve had that chicken-egg problem many times since both at Tumblr and Duolingo, paid marketing isn’t an option. : )

      Really quick note that a lot of the headers on the webpage seem to be broken. Share your email and I’ll send you screenshots.

      Press has been a huge ally in growth career so far, but it doesn’t work by magic so you need to be strategic. Mainly, remember that you don’t get a lot of chances to make a first impression with bloggers and journalists because they get tons of emails and pitches every day. Find stories that would *actually* interest their readers (successful / touching user stories, data / trends you’ve come across in the experience / anyone famous who has experienced the problem you’re solving and might be willing to advocate / why you’re changing something in a significant way including examples) and make sure to personalize messages to each and every one of them. This doesn’t mean just writing Dear : it means knowing what they’re interested in and letting them know you’re informed about their work.

      One quick trick for press that has really worked: if you want people to visit your website, and I’m assuming you do since that’s where you’ll see conversions, it’s really important that publications link to it so that readers know how to get there. People are too lazy to read about something and Google! So when you send out the message or release, make sure to include key phrases, hyperlinks and image assets that you’d like to see included in those stories.

      Facebook and Twitter (targeting messages to influencers who might want to spread the message) are good channels, but make sure you’re tracking analytics. That means: how is each post actually driving traffic or awareness in a quantifiable way? Facebook ads, by the way, are also pretty effective and cheap, so worth considering. If anything, they can help you discover what drives people’s interests in terms of images and copy.

      You can also contribute to publications you respect in the industry. You don’t have to write about your business, like an advertorial, but you can start positioning yourself (or someone else in the company) as thought-leaders in the space. I hope this helps!

      • BS

        Bhaskar Sarma

        10 months ago #

        Excellent suggestions, @ginafrombrazil , especially about channels and analytics.

        About Facebook ads, it's way cheaper to boost a blog post than it is to get someone to directly download your app or sign up for your list. Instead, use the post to provide value and THEN use something like a content upgrade or a lead magnet to build your list.

        @tomixus I had a look at your site and you have the potential for a number of human interest stories that would be attractive for reporters writing on technology and healthcare.

        You could create an infographic and preferably use internal data to highlight best practices in terms of care for the elderly. Include details around physiotherapy, nutrition, post operative care etc. Also tie in publicly available data like the number of elderly patients, number of households where senior citizens live on their own, the percentage of hospital population whose age is over 65 etc to put your data in context.

        And then, promote it using @backlinko 's Guestographic method, http://backlinko.com/how-to-get-backlinks.

        If you have the resources, create a regular video series and upload it on Facebook or YouTube. it doesn't have to be even professionally edited: initially, simply using a good quality phone camera with a decent microphone with do the trick. Interview your top caregivers, and ask them about what the elderly should do to stay fit and healthy, or how often a dressing should be changed after a surgery.

        You could also pitch to authority blogs in niches like ageing, palliative care, nutrition for elderly etc and gain significant traction.

      • SE

        Sean Ellis

        10 months ago #

        I'm getting a lot of broken images too (for example the images under "how it works").

      • TS

        Tom Ste

        8 months ago #

        Hi Gina,

        Thank you very much for sharing your thoughts. It's really helpful. :-)

        Regarding broken images, can you send me an email with screenshot and details to: thomas.stermole@gmail.com ?

        We recently made some changes to our landing pages.

        Thanks.

    • AG

      Anshul Garg

      10 months ago #

      I am curious to hear @ginafrombrazil's answer on this.
      @tomixus - For two-sided marketplace, Andrew Warner of Mixergy interviewed Jerry Nettuno of SCHEDULICITY. I really liked that interview. Checkout:
      How to launch a two-sided marketplace from day one https://mixergy.com/interviews/schedulicity-with-jerry-nettuno/

  • SE

    Sean Ellis

    10 months ago #

    Hey Gina, thanks so much for doing this AMA. I'm really impressed with what you guys have done with Duolingo. I'm personally addicted to the service. My question is what is your typical day like now vs what it was like two years ago?

    • GG

      Gina Gotthilf

      10 months ago #

      Hi Sean! It's so cool to hear that you're addicted to Duolingo, and thank you for the kind words. : )

      My day is vastly different: 2 years ago we didn't have a growth team. My focus was mainly on communications, PR, partnerships, and anything that would really get the word out there. The entire dev team was working on optimizing the platform. Now, I'm overseeing both the marketing organization across all Duolingo teams and also heading our growth team. I find it interesting how marketing and product roles are blending in tech startups, and how important it has become for marketers to develop the skills necessary to evaluate and analyze designs, as well as to communicate with and lead designers and engineers.

      Thanks for having me!

      • SE

        Sean Ellis

        10 months ago #

        Agree - effective marketing has definitely become a more integrated function than it used to me. There are just too many important growth levers that sit outside the traditional areas of influence of a marketing team.

  • SA

    Shaker A

    10 months ago #

    Hi Gina,

    Thanks for doing this!

    1)How do you decide how long to run an experiment for? You could take forever to analyze the data (always another piece of data you could look at). How do you know when you've done enough analysis to declare it a winner or loser? Declaring a loser would probably be harder since if you look hard enough you could probably find something of value.

    2)What is your process for figuring out what channels you should focus on for customer acquisition? How do you evaluate them? How do you come up with a plan of action to attack that marketing channel? Do you have a process for evaluating channels, if so could you share said process?

    3)How do you think about scaling traffic growth? What changes for how you use a channel from when you first identify it's viability,to using the traffic channel at scale? What question do you ask yourself to make sure that the channels stays viable as you scale it up? What are the pitfalls to avoid as you scale a channel? How do you overcome said pitfalls? Do you have a process for scaling channels, could you share said process?

    Excited to learn from you!

    • GG

      Gina Gotthilf

      10 months ago #

      Heya Shaker! Here we go.

      1. We usually run experiments for 2-3 weeks and the way we decide it’s enough is if results have been stable for many days. To analyze data, you need to know if those results are statistically significant. If they’re both stable and statistically significant, then it’s time to make sure you’re checking all relevant metrics. For example, we’ve run experiments that increase a metric like “trial account conversions” but actually hurt other metrics, or even long-term retention. Something that helps streamline this process is putting together experiment spec docs in advance that ask these questions and set everyone up to look at all relevant metrics before making a decision.

      5 Share
      • GG

        Gina Gotthilf

        10 months ago #

        2. This is probably going to sound obvious, but it’s all about ROI, and the “I” can be defined by resources in general. At Duolingo, we actually don’t do any paid marketing, so our “I” is often time (which costs money!). We’ve done things like PR, social media, events, partnerships, bylines and a few others, and after a few runs with each, it becomes pretty clear what deserves more or less time. For example, we’ve found that Facebook posts (which were taking about 30 minutes a day to craft) brought us very similar results whether we posted once or twice a week… so now we only post once.

        For us, return means traffic to Duolingo or downloads (directly or indirectly) because we’re focusing on growth. In most cases, return will mean money, or the LTV (lifetime value) of your users: how much does it cost me on this channel to bring in one user, and how much is the user worth? One thing that’s important to remember is that the “I” part can always be optimized so you can’t judge before you know how low it can go. For example, email marketing requires time: how can you optimize your team’s process to help cut down on the hours that get poured into each email? How low can you bring the CPC of an ad by experimenting with keywords and targeting?

        The same thought process is applied to growth a/b experiments.

        4 Share
      • GG

        Gina Gotthilf

        10 months ago #

        3. When thinking about scaling a traffic, first you need to define if it’s worth it (ROI) and if the channel is actually scalable. That means considering available inventory, required resources, audience size and fatigue. Then, optimize the process: write it down so anyone in the team can replicate and people can operate without bumping into each other or going back and forth.

        The main things to watch out for are audience fatigue, message targeting and, relevant to my experience, internationalization. Sometimes things work because there’s a novelty effect. This applies to traditional marketing and growth a/b experiments as well. You need a little time to see if results can be maintained and scaled. Once you know something is working, scalable and bringing in legit returns, you can crank it up but then sensitivity to different audiences comes in. Messages and images will work really well for one group but will need significant tailoring and/or localization for others. Scale carefully and keep your eye on metrics.

        2 Share
    • BS

      Bhaskar Sarma

      10 months ago #

      @shak Riffing off of @ginafrombrazil 's answer to your question #2, you should also decide on a Northstar metric and set KPIs before figuring out and optimizing which channels to use for customer acquisition.

      Are you looking at unique visitors to your website or qualified leads? Is it the size of your email list that actually matters, or should it be CAC? The answers will vary based on your niche, the growth stage you are at, and your business model.

      Knowing your Northstar metric, and then deciding on a stretch goal, like @noahkagan 's focus on hitting 1 billion unique visits to sumome.com in one year can gift you plenty of clarity.

  • AA

    Aldin A

    10 months ago #

    Hey Gina,

    Great to have you here at GH!

    1)How do think about retaining users, if your user only need to use your app occasionally by nature (ex shopping app)? If your app isn't used frequently building up the habit is hard, which makes it even harder to retain the user. How do you go about trying to stay top of mind so when the user has a need that your app solves they think of you?

    2)How do you increase your chances of determining that a feature that you are thinking of building is of genuine value to the user vs just leading down the path to a more bloated product? Do you have a process for vetting features? Additionally as your product grows how do you make sure your product doesn't become bloated?

    3) What are some of your favorite resources pertaining to not only growth, but business in general (books, podcast, blogs, courses, etc)?

    Thanks

    • GG

      Gina Gotthilf

      10 months ago #

      Hey Aldin! 1. Good question, and of course it depends on what you’re offering and who your audience is. With Duolingo, we actually are actively trying to help users create a habit and come back as often as possible, so it’s a fairly different problem. With a shopping app, I’d look into creating a messaging/email campaign that allows you to stay top-of-mind in a relevant and non-intrusive way, using a/b testing to determine timing and messaging. Relevant means: what can you bring to your customers based on your company’s knowledge / field that can be relevant more often than they actually need to go shopping? Trends, suggestions, user stories, that kind of thing -- test and see what results in higher open rates or click-throughs. There are also really interesting ways to target location which would be particularly relevant to that business: you’re close to a store, come on over! Social proof: connect with friends somehow and have users share ideas, likes, opinions, anything that will allow you to ping customers in a way that doesn’t feel like spam.

      • GG

        Gina Gotthilf

        10 months ago #

        2. We’re extremely careful about that and to be honest, most feature ideas die. It’s really important that Duolingo stay simple and intuitive even though everyone and their brother will say “why don’t you have this,” “why don’t you have that.”

        The features that get built are chosen by a combination of team experience and intuition, but unless they have a significant impact on a significant metric, we opt for simplicity. We also highly consider ROI in the sense that bigger features can take a lot of time so the question is: are there simpler a/b tests that can bring similar results that can be implemented in less time?

        Right now, the growth team is actually balancing “inch-by-inch” type experiments with 1% gains and allocating time to defining bigger feature projects that can contribute to growth in more significant ways, but other teams on Duolingo are dedicated to larger features like the chatbot, which literally just launched! (chatbot.duolingo.com)

  • LS

    Logan Stoneman

    10 months ago #

    Hey Gina,

    Thanks for being here today! I used to work in growth/marketing in education and found the bureacracy of higher education to be tough to penetrate as a young startup. It appears as though Duolingo has not had that problem as you went directly to the users/learners. Now that you've grown so rapidly through this D2C channel, do foresee doubling down on this approach moving forward, or are you looking to work your way into public/private institutions?

    • GG

      Gina Gotthilf

      10 months ago #

      Hey Logan, we know what you mean! We decided to go straight to learners in a bottom-up approach for that reason, and also simply because a lot of people don’t have access to educational institutions (due to time, money or location) and are interested in learning. Duolingo has now grown to 150M users and we only started offering a product for schools (schools.duolingo.com) fairly recently.

      We’re now doing both, but I’d say we’re really working hard to reach scrappy, dedicated teachers instead of going for administrators because in a similar bottom-up approach, they can advocate for us internally. We have gone for bigger fish, like for example the Ministry of Education in Colombia and the Department of Education in New York City, and have put together projects we’re really proud of, but these take a lot of time and we’re a tiny team so it’s difficult to scale.

      It’s exciting to see a platform used by celebrities and people with SERIOUS resources like Bill Gates or Chelsea Handler using Duolingo, while ALSO seeing public institutions in poor countries like Guatemala use the same tool To us, that’s educational equality: when money can’t buy you a better education.

      The rest of Duolingo still focuses on making Duolingo itself great (again, just kidding) so that we can continue growing through word of mouth, in and outside of classrooms.

      4 Share
  • GC

    Gary-Yau Chan

    10 months ago #

    Hi Gina,

    What is the average DAU / WAU / MAU for education apps?

    I assume they are not very high. What other metric you use to measure overall Duolingo KPI?

    And what do you with the large amount of users who churned or inactive from Duolingo?

    Thank you,
    Gary

    • GG

      Gina Gotthilf

      10 months ago #

      We actually don’t know for sure, but from what we’ve seen Duolingo is way ahead of the curve. We don’t disclose DAUs publically but can say we’re at 20M MAUs (and 150M total users).

      DAU is our key metric at Duolingo but a lot of sub-metrics feed into it directly. We look at retention (day 1 and recurring, which is a lot harder to measure), use site visits and downloads to measure the impact of marketing and PR efforts and work on optimizing those metrics every single day at Duolingo. It’s pretty fascinating to see how far we’ve come since our launch!

      6 Share
    • GG

      Gina Gotthilf

      10 months ago #

      We send them a photo of our management team crying.
      Just kidding! We actually started sending a day-30 email with our mascot, Duo the owl, crying. It turned out to be pretty effective and got some attention in the media. :) Guilt FTW! We’ve also tested with messaging to try and re-engage people who really wanted to learn a language but got distracted, which happens often. This is all through a/b testing, of course.

      We contact inactive users once in a very long while (every few months, sometimes once in 6 months) with very cool new updates via newsletters to see if we can get their interest back.

      4 Share
      • HQ

        Hila Qu

        10 months ago #

        Haha, you should test different people from management team crying against the owl and see who wins :)

    • SE

      Sean Ellis

      10 months ago #

      Sample size of one, I'm definitely a DAU for Duolingo. I get pretty bummed when I lose a streak (but not bummed enough to pay to retain it).

  • RB

    Ry B

    10 months ago #

    Hi Gina,

    Thanks for doing this ama.

    1) What are the most important lessons you've learned in your career? What are the biggest lesson's you've learned about growth?

    2)How do you look at hiring? Can you talk about some of the mistakes you've made hiring (and also seen others make)? What have you learned about hiring A+ talents?

    3)What do you think are the top skills/traits that a manager needs to have to bring out the best in their employees?How do you go about empowering employees and what does that look at duolingo?

    4)In your opinion what are things early stage startups have to do to not only survive, but thrive? Conversely what do you see startups messing up that they can't afford to, and how do they fix them?

    • GG

      Gina Gotthilf

      10 months ago #

      3. Top traits a manager should have: I'm learning every day! So far, this is my assessment:
      - honesty and transparency. I think showing you're human and being pretty clear about motivation helps the team trust and rely on you.
      - Reliability. Your team should know that they can look to (and depend on) you for advice or help at all times.
      - knowledge of whatever it is you're trying to do and/or a proven track record of some sort. If the team doesn't look up to you or trust you in any way, they will question your existence.
      - Flexibility and compassion. Shit happens.
      - Confidence
      - Communication skills. Things can turn sour fast without communication and a manager needs to keep everyone in tune.
      - Organizational skills. Everyone needs to know what they're doing... and you need to know what everyone is up to in order to help everyone stay on schedule and organized. I wish my elementary school self had known that!
      - Listening skills. That means being able to actively listen to each individual in your team to make sure eveyrone is motivated and on track.
      - Be nice. There's no need to be mean to people. Strict-ish, stern, yes. Unpleasant, no.

      6 Share
      • GG

        Gina Gotthilf

        10 months ago #

        4. In my opinion the most important thing for an early-stage startup is the team (for any startup, actually!). Early on you should set up a rigorous interview process, and only hire A+ players. If you don’t know how to interview them (perhaps because they work in an area that you don’t know much about), ask somebody who does or even consider hiring a consultant to help you interview them. Also, look for ones that are passionate about your mission, and as soon as possible hire somebody with previous startup experience.

        Finding product-market fit sounds obvious but is often overlooked. Just because you and your friends like something, it doesn’t mean there’s a scalable need for that so startups need to move as quickly as possible to get there. Otherwise it’s like planting a seed in a bed of desert rocks.

        The other problem I see a lot is a tendency to outsourcing engineering. That may work for some but if you’re building an app, you better have experts on your team (with high stakes in your company, like partner status or serious equity). In my experience, product isn’t something you can really “finish” building -- it’s an on-going process and you need to be able to optimize as you learn and understand that your assumptions were wrong. And trust me, a lot of them will have been very wrong. : )

        Getting too excited and building something complicated / with too many features. Simplify. To really make something great, it has to be simple otherwise you’ll quickly lose focus.

        This leads to scaling too quickly. Hiring en masse means people will quickly lose direction and focus and you’ll have to sacrifice quality.

        3 Share
    • GG

      Gina Gotthilf

      10 months ago #

      Oof that’s a big one. Biggest lessons I’ve learned in my career:

      Work hard and work smart. Work hard in that I’ve never found shortcuts so putting in time and effort to go above and beyond your job requirements (even babysitting) really well has always made a difference. Work smart meaning: just because you’re spending long hours at the office doesn’t mean you’re doing anything useful. Read, learn, listen, prioritize, reflect.

      Everything is made of people. Offices, careers, companies, industries. It’s all people. Therefore, network intelligently (by helping and reaching out to others) and be nice whenever you can. Remember that people have good and bad days. Curb your own bad days to avoid affecting others and try to give others the benefit of the doubt. I’ve spent too much time thinking people were targeting me unfairly when really they were just struggling with their own insecurities. Finally, pick a place to work with nice people. You’ll spend a lot of time there and you want to respect the people around you, and make sure you’re not crying in the bathroom on a regular basis. Nobody needs that. : )

      It’s important to put yourself out there. You can do the best work in the world but if nobody knows about it, there’s a good chance it will go by unnoticed for a very long time. Self-promotion is frowned upon, but make sure the right people see your wins and that you have a voice.

      Try to avoid complaining. I’m still working on this one but nobody likes complaining.

      Top lessons about growth:
      If it’s not measurable, you can’t grow it. Metrics is our deity at Duolingo.
      A/b test everything with a subset of users.

      Make sure you’re reading results correctly before making decisions. Is it statistically significant? Does it complicate the code more than it contributes to growth? Does it affect other metrics in weird, unexpected ways?

      Simplify whenever you can. When you work with high-achieving, smart people, they tend to be perfectionists. Launching a/b tests quickly and iterating requires a certain level of scrappiness that can feel unintuitive.

      Don’t over-simplify. Often, when we consider an experiment, we try to come up with an “MVP” to test the waters. However, we often end up with minimal-to-no results because as it turns out, the scaled-down version actually depleted the test of important attributes. Take “gamification” for example. Testing a tiny little gamification addition may not actually indicate what you’d see were you to transform the entire experience to make it more fun
      .
      Focus. There’s so much you can do, but you need some sort of structure. Some growth teams divvy up time between “acquisition” (the top of the funnel) and “retention.” Figure out how to structure your team’s time and thinking to optimize performance.

      Don’t beat yourself up. You’ll run dozens of experiments… and the majority will fail. That’s OK, keep moving.

      Learn from other apps and services. What are they doing that’s working for you? How can you extract that idea and implement it into your context in an intelligent way?

      5 Share
      • GG

        Gina Gotthilf

        10 months ago #

        2. On hiring: The biggest mistake is to jump on a hiring decision because you’re drowning and “need” someone right now. Either they’re A+, or you wait. Hiring mistakes are extremely expensive in terms of time, frustration, productivity and team morale.

        A specific mistake I’ve made when I first started hiring was not to check for references (I know, wtf) and not to evaluate the person’s writing. The position was very communications-based and while the writing sample we received was pretty good, we never verified whether it was actually written by the candidate (without outside edits).

        Rely on other people’s evaluations, regardless of whether they’re experts in your field. At Duolingo, we have a number system that very clearly delineates how each person assesses a candidate and any veto requests need to be taken very seriously.

        A+ talents feel right. Many people in your team get excited about them, they can name specific examples of things they’ve done and their results (and prove them) and references (including outside references) will go out of their way to promote them. If you’re unsure about someone, wait.

        Also, hiring mistakes are serious mistakes. Fix them. Learn, have a post-mortem, but don’t procrastinate.

        Finally, “cultural fit” which has recently gotten into a lot of trouble due to its acceptance of blatant discrimination. Be really careful about how you judge people, but ultimately, it’s important to determine whether that person will themselves be happy at your company.

        2 Share
  • AG

    Anshul Garg

    10 months ago #

    What's the best strategy that have worked for you for acquiring first 10 customers ?

    • GG

      Gina Gotthilf

      10 months ago #

      To acquire 10 customers, talk to friends and family! Then try to get press, build a social media presence, start creating a highly-relevant email list of potentially interested people (and spend a lot of time on that one email before you’re marked as spam early on), contribute to relevant media as a writer to start establishing your voice as someone relevant in the space, look for co-marketing opportunities that make sense to both players.

      But unless you really have product-market fit and some semblance of retention, all those people will fall through, meaning they’ll check your product out and “churn” immediately (never come back). So don’t put the horse ahead of the carriage.

  • FS

    Fernanda Saboia

    10 months ago #

    Hi Gina. A couple of questions.

    - Does duolingo have any plans to incorporate messaging-like experiences and interfaces?
    - What do you think the online education sector will look like in 3-5 years? How is duolingo anticipating the upcoming challenges?
    - What are you thoughts on the current state of the general education sector and how digital and online providers are impacting it?
    - What other learning platforms do you use?
    - What language are you currently learning on duolingo?

    Obrigada! :)

    • GG

      Gina Gotthilf

      10 months ago #

      1. Fernanda, you’re a clairvoyant! As you noted below, we launched something awesome ONE HOUR ago, check it out here: bots.duolingo.com

      2. I think online and offline will blend progressively in the next few years. I don’t think online education will substitute offline education in any way, but I do think that teachers and professors who incorporate online tools into their curriculum will replace those who don’t.

      One exciting thing is that for the first time, we can analyze student data at scale. This means we don’t have to wait 1 semester to see if the way we’re teaching something is effective… we can actually observe and analyze learning data and optimize in days or weeks.

      For that reason, I don’t think that static video education is what the future holds for education. People love games and have a natural resistance to studying. They want to be entertained and engaged because that makes dedicating time easier and helps retain knowledge efficiently. We built Duolingo so that it’s completely interactive, meaning not only that someone’s full attention is required (no point in daydreaming since you have to answer to move forward), but also so that we can collect all that data and apply it towards making education more effective.

      3 Share
      • GG

        Gina Gotthilf

        10 months ago #

        3. General education: I think we’re all questioning education at a deeper level and hope to see significant changes, probably driven by data. What should kids be learning? How? We all graduate from high school with years of math and yet, most of us probably can’t add fractions or understand percentages. All my American friends studied Spanish and yet… a very small minority actually speaks the language at any capacity. I learned all about topics that have never been relevant in my life, and honestly, forgot a lot of what I spent hours and hours memorizing. And, of course, college students are graduating with high debt and few job opportunities.

        One major problem is that education institutions tend to be old-school (get it?!) and slow to accept change. I think digital and online providers are working to help change things, but at the same time, I do applaud teachers for being careful when picking what tools to adopt. There are a lot of tools out there and the vast majority are quite ineffective. Measuring education is hard but really important (we have an entire team, mostly of PhDs, dedicated to learning efficacy).

        Another problem is that “fun” and “learning” kiiiiind of go together in the US but in many places, things that are fun aren’t seen as “serious” and effective. We’re trying to break that misconception to help kids become more engaged.

        2 Share
    • GG

      Gina Gotthilf

      10 months ago #

      - What other learning platforms do you use?
      I’ve signed up to classes on Coursera and EDx but haven’t been able to stick to them. The content is amazing, the professors are awesome, but I find it hard to fit it into my life because after a long day of work, I don’t want to sit and watch a long lecture online.

      I do want to commit, though, and haven’t given up.

      - What language are you currently learning on duolingo?
      German! : ) I did Spanish and French but both were languages I already kind of spoke. It really helped me brush up on grammar (implicitly) and vocabulary, but I felt like I was cheating a little. I really wanted to see what the experience was like for someone trying to learn a language from scratch on Duolingo, and I’m on a 160-day streak studying German!

  • AA

    Alexis Ang

    10 months ago #

    Hi Gina,

    Thanks for taking the time to hop on here! Wanted to know:

    - What part of the funnel was the biggest challenge when getting Duolingo to where it is now? Acquisition, activation, retention? etc.

    - What was your OMTM

    - Any analytical stacks you'd recommend that made keeping track of metrics easier to see + understand + leverage?

    Thanks!

    • GG

      Gina Gotthilf

      10 months ago #

      Hallo Alexis!

      1. I think retention is probably the toughest part. You can force people to check out your product / app / service, but unless you can get those people to stick around, there’s no point! In other words, you could potentially buy acquisition… but not really retention, and that takes a lot of time to build up and optimize.

      2. Our holy grail metric: DAUs

      3. We build almost everything internally, but our most important analytics tool is Mixpanel.

      • AA

        Alexis Ang

        10 months ago #

        Thanks so much Gina! PS Love Duolingo. Husband speaks mandarin so I downloaded but haven't upgraded to premium paid yet :) Looking forward to reading your replies today!

  • KS

    Klemen Struc

    10 months ago #

    Hi Gina. Thanks for taking the time to be with us.

    How do you structure, organize & run a cross-functional growth team, especially if a growth team consists of many sub-teams? (e.g Paid Traffic, Organic Traffic, Social Media, Conversion Optimization, Analytics, Copywriting, Design, Tech, etc.)
    How do you organize a meeting cadence in a way where everyone is aligned and on the same page, and things get done and implemented fast, goals and targets are achieved, etc?
    When do you have meetings and who do you get involved?

    Would love if you share anything on this topic.
    Love,
    Klemen

    • GG

      Gina Gotthilf

      10 months ago #

      Klemen, we have people from different disciplines in the team: design, engineering, marketing, and product management. We structure it as a single team and sometimes break out into sub-units, but it’s important for everyone to feel like one united team and have fundamental respect for the other disciplines. If they don't , find a way to build that by allowing people to work together and also giving those who are less appreciated a voice to show off their contributions in a relevant way.

      I do find sub-units useful when meetings feel too big: smaller groups of people can more effectively brainstorm and arrive at conclusions.

      On meetings, it's important to share a meeting agenda in advance. I actually ping each individual before the meeting so they know what's up and also to see if they'd like to discuss anything in particular. I find that a lot more effective than sending an email to the whole team because then, everyone feels like there's an audience.

      In terms of keeping everyone on the same page, it's not easy and sometimes requires 1-1 conversations and being confident. I find that people really want democracy but they ALSO want someone to really help in making decisions and to direct them, otherwise frustration arises.

      • KS

        Klemen Struc

        10 months ago #

        Gina, thank you again for taking the time to share you insights and advice. All the best on your quest.

  • TS

    Tomek Szczerba

    10 months ago #

    Hey Gina,

    We just got excited about opportunity to hear your inputs and brainstormed following questions that help us greatly in fine-tuning our growth process:)

    How big is your growth team and which functions should be represented in it?
    Are you able to implement your ideas independantly (do you have engineers on your team) or book X hours of dev team time a week?
    Who takes a decision on next levers you'd be focusing on and tests you'll be running?
    How do you build synergies and alignment with VP of Product/ other management team members (growth related decisions affect overall product and may be contradicting to their plan?
    Last but not least what is your growth team meeting agenda

    Thanks a lot in advance!

    • GG

      Gina Gotthilf

      10 months ago #

      Hey Tomek,

      1. We have 9-ish people including me, 4 engineers (2 of whom have a PhD!) dedicated to different projects (SEO, Android product changes, iOS product changes, emails / notifications), 2 half-time designers, someone focused on copywriting, communications and PR, one part-time PM and myself.

    • GG

      Gina Gotthilf

      10 months ago #

      2. We’re super lucky at Duolingo in that the growth team has the capacity to operate fully independently.

    • GG

      Gina Gotthilf

      10 months ago #

      3. Who takes a decision on next levers you'd be focusing on and tests you'll be running?
      We collaborate a lot as a team on the decisions, but our PM looks at potential ROI for all experiment ideas before we seriously consider them. The engineers and designers also have a lot of say because they know how complicated or simple an experiment can be and can point out potential pitfalls that are hard to notice if you’re not executing.

      4. How do you build synergies and alignment with VP of Product/ other management team members?

      Communication is really important and it’s something we’re constantly improving at Duolingo. We have an all-hands meeting every week in which everyone announces the most important experiment pushes or projects for the week, and we also make sure that Luis von Ahn, our CEO and interim head of product, knows about everything we’re working on because he has the best bird’s eye view of what’s going on in different teams. We’re also pretty close-knit so colleagues often share information with each other, which allows everyone to point out red flags during meetings in terms of potentially conflictive experiments.

      5. The meeting agenda varies a lot, we don't have a recipe. We've tried different formats and have most recently implemented weekly sprints, which are helping us stay on track pretty effectively. I try to keep all meetings as short, directed and relevant as possible.

  • BW

    Brand Winnie

    10 months ago #

    Hi Gina!

    Thanks for doing this...

    1. You guys have some great design and it's very minimal and easy to use. How important has branding and design been in the success/growth of Duolingo?

    2. How close is the growth team to the design team? Or do you have specific designers that are only focused on growth?

    3. What's the process like when you figure out that something isn't working within the experience and you need to fix the problem based on the data? Can you walk us through that?

    Thanks for your time!

    -Brand

  • HQ

    Hila Qu

    10 months ago #

    Hi Gina,

    Such a treat that you are here, and I love your profile pic :) I listened to a Tim Ferris podcast interviewing Luis von Ahn who I remember is both a co-founder of Duolingo and a professor at CMU, he is definitely a fun guy. Feels like Duolingo is a fun team!

    So first of all, what is the presentation to President Obama about, and what is it like? Really curious to know.

    The growth from 3 to 150 millon users without paid advertising is impressive. What are the most important elements lead to that result?

    What are the challenges in term of retention and can you talk about Duolingo's retention strategy?

    Thank you!
    Hila

    • GG

      Gina Gotthilf

      10 months ago #

      Hi Hila, thank you! 1. On President Obama:

      We’d been invited to attend the White House’s first Demo Day but we didn’t actually know we’d be pitching to President Obama himself! It was definitely a highlight in my life and career and I I know that the cool thing to do would be to play it down and pretend it ain’t no thang. : ) Not only was it nerve-wrecking to be in front of the President, but there were about 15 cameras right in front of us, clicking away every half second as we spoke which made the whole experience even more daunting. It was a huge honor, especially for an immigrant like me. Here’s the video, which is awesome but also a little embarrassing. (http://www.nbcnews.com/news/latino/obama-admits-bad-spanish-he-gets-language-learning-app-demo-n404156)!

      • GG

        Gina Gotthilf

        10 months ago #

        Most importantly, we have a good product. We spend most of our time (even in the growth team) making the product better through optimizations / a/b tests. We measure the fraction of people that come back the next day, the “Net Promoter Score” (NPS), etc. Our co-founder Severin Hacker once said to the company: if you can’t measure something, you can’t improve it. We believe in that.

        In terms of growth efforts, leveraging Luis von Ahn (co-founder and CEO) as a spokesperson and developing targeted efforts to seed news/feature stories for press in countries around the world really paid off. We “launched” Duolingo with a talk by Luis in countries like Turkey, India, Brazil, Japan, China, Korea, Mexico and many others and it really helped get our name out, even for people who had never heard of Luis’s previous work. Developing a relationship with the folks at iTunes and Google Play was also really important since in the beginning, a feature by either one of those makes a huge difference.

        It’s also interesting to note that Duolingo was a website and we’d thought of launching apps as “mobile companions!” Launching mobile-native versions of Duolingo made a huge difference as today, 80% of our traffic comes from mobile.

        Finally, emails and notifications have been extremely effective in bringing people back.

      • AA

        Anuj Adhiya

        10 months ago #

        :raised_hands: :clap:

  • MM

    Matheus Martins

    10 months ago #

    Hi Gina, hello from Brazil (oi, tudo bem?)
    Do you think Duolingo's business model and growth strategies can be applied to any kind of educational business? I mean, not just languages.

  • NL

    Nick Lock

    10 months ago #

    Hi Gina,

    Did you do a lot of growth before the app was launched?

    What are some tactics you recommend for pre-launch growth (wait list/email sign ups)?

    Thanks for doing the AMA on GrowthHackers!

    Thanks,
    Nick Lockard

    • GG

      Gina Gotthilf

      10 months ago #

      Hola Nick! Duolingo didn’t do any “growth hacking” before it was launched. We were lucky that our co-founder and CEO Luis von Ahn gave a popular TED talk (http://www.ted.com/talks/luis_von_ahn_massive_scale_online_collaboration?language=en) in which he announced Duolingo, which got us our first 300,000 users.

      We also focused a lot on building a solid product that people loved and were willing to tell their friends about. That’s what I mean about focusing on product-market fit and optimizing retention before thinking about things like “hacking.” TBH, you can’t really grow anything that doesn’t fully exist yet. :)

      • GG

        Gina Gotthilf

        10 months ago #

        Building up a wait-list and an invite-only ordeal can really help if you have the chance, to create an allure of exclusivity. There are a lot of famous case studies, like Gmail, to learn from. Trying to get the word out in niche communities and seeding the “sign up for the beta” link can definitely help.

        This will sound cheesy, and repetitive, but I think it's an important message. I’d definitely spend most of your effort on making the product great and building up retention (after product market fit). Your marketing efforts will be a lot more effective once you do that because fewer people will drop-off in your funnel.

  • DF

    David Firester

    10 months ago #

    Congrats on all your successes. Growth is very important, but do you have any recommendations on maintain/engaging existing users? Thanks for your time.

    • GG

      Gina Gotthilf

      10 months ago #

      David, at Duolingo, we consider maintaining and engaging our existing users as part of “growth.” In fact, most of what my team does involves improving retention among existing users. We look a lot at the “D1 retention” (fraction of the people who sign up that come back the day after). When Duolingo launched, that number was about 15%. It is now well above 50%. Improving the signup funnel and sending a really good notification has made a huge difference here.

  • NT

    Nahuel Tori

    10 months ago #

    Gina, how do you do to encourage the users to keep using the application? Besides game theory, have you tried anything else?
    Thank you very much from Argentina!

    • GG

      Gina Gotthilf

      10 months ago #

      Hello Nanuel from Argentina, my neighbors! : ) Retention is extremely important for growth or really, any type of success. We don’t use game theory, but instead we use gamification principles (which might be what you meant) to make the experience more engaging. Emails and notifications are also really big players in retention for us. A/B test everything, of course, to improve your D1 and recurring retention rates progressively because those efforts really scale!

  • HP

    Hugo Pereira

    10 months ago #

    Olá Gina!

    Prazer te-la no AMA :D
    Quick question: Most growth methods and tactics are built upon software products and companies with a direct user model. What about growth models via indirect channels (distributors, resellers) since product is hardware. Would the growth methodology apply in the same way? Do you see a difference in approach?

    Thanks!
    Hugo

    • GG

      Gina Gotthilf

      10 months ago #

      Hugo, muito prazer! I’m sorry but I’m actually not the best person to answer this question since it’s outside my scope of expertise since I haven’t worked with distributors and resellers in a growth capacity.

  • ES

    Edward Stephens

    10 months ago #

    So excited to have you on I bloody love Duolingo.

    Some questions from me:

    1) Duolingo has obviously learned a huge amount about how to inspire and continually inspire learning. What do you think the most satisfying think you've understood about the way people learn is and how we can motivate ourselves?

    2) How long does it take to master a language (your typical user) and what languages have you learned?

    3) What challenges have you faces with a huge growth in users and it mentions you never did paid advertising so how did you create a successful referral engine?

    I can't wait to hear your answers.

  • AA

    Anuj Adhiya

    10 months ago #

    Hey Gina - so stoked to have you on!

    I'm going to out myself by saying I don't use Duolingo as often as I'd like to or should (and there's no plan to uninstall it ever) but that's what brings me to my question:

    What patterns have you seen work with being able to resurrect users - especially those like me that tend to constantly relapse into inactivity after bursts of use? Also is that behavior normal or am I (as many would say) a special case?

    Is there a time frame after which you've seen that push notifications work best to re-engage such users? What kind of notification(s) has worked best in such scenarios ie what messaging has converted best?

    Also do you ever give up on folks like me or is there always an nth campaign that will attempt to bring me back?

    Thanks!

  • FD

    Flaviu Dunca

    10 months ago #

    Hello Gina

    What platform do you use for A/B testing a mobile app? Thanks

  • DH

    Dani Hart

    10 months ago #

    Hi Gina,

    So great to have you on here today!

    1. How big is your growth team and what does your process look like? How does this team work within the organization?
    2. What tools are in your growth stack? Do you use any consultants or outside partners on an ongoing or ad hoc basis?

    Really looking forward to learning more.

    Cheers!
    Dani

  • DL

    Douglas Lotz

    10 months ago #

    Hi Gina,

    My app operates under a paid subscription model and I'm in the process of deciding whether to send out engagement-oriented push notifications to paying subscribers who haven't used the app in a while or to just leave them be.

    On the one hand, encouraging currently paying users to reengage may improve retention by reminding them of the value they get out of the app. On the other hand, we may end up just reminding these less-than-active users to turn off auto-renew on their subscription...

    Any insight you can provide would be greatly appreciated, thanks!

    -Doug

Join over 70,000 growth pros from companies like Uber, Pinterest & Twitter

Get Weekly Top Posts
High five! You’re in.
SHARE
95
95