Today's episode is sponsored by public.com.
That's where you can earn a 5.1% annual percentage yield with a high yield cash account.
And while we can't say for certain that's the highest interest rate out there, we can say that at the time of this recording, that's higher than Robin Hood, higher than SoFi, Marcus, Wealthfront, higher rate than Betterment, Capital One, Allied, Barclays, away higher rate than Bank of America and Chase, higher rate than City, Wells Fargo.
Think you get the point here.
If you want to start earning 5.1%, APY on your cash checkout public.com.
This is a paid endorsement for public investing 5.1% APY as of March 26, 2024 and is subject to change.
Full disclosures and terms and conditions can be found in the podcast description.
US members only. This is Business Breakdowns.
Business Breakdowns is a series of conversations with investors and operators diving deep into a single business.
For each business, we explore its history, its business model, its competitive advantages, and what makes it tick.
We believe every business has lessons and secrets that investors and operators can learn from.
And we are here to bring them to you.
To find more episodes of Breakdowns, check out joincollossus.com.
All opinions expressed by hosts and podcast guests are solely their own opinions.
Hosts, podcast guests, their employers, or affiliates, main-antane positions in the securities discussed in this podcast.
This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions.
This is Matt Russell and today we are breaking down Duolingo.
It's hard to find someone who doesn't see the opportunity to build new technology businesses around education.
But it's equally hard to find existing success stories in education technology or edtech.
Enter Duolingo. The learning app built on language learning that is increasingly expanding into other territories like math and music.
My guest is Tyra Winn from Bailey Gifford.
You should assume that most of our guests own the businesses that they are covering on business breakdowns.
But it's important to mention here that Tyra works for Bailey Giffords, positive change strategy.
They invest for returns but also for the impact on society.
And while I personally try to avoid casting judgment on how businesses earn their profits, we have certainly covered some businesses that are deemed to be world class at raising prices.
Duolingo seems to stand out in a different way.
Founder Luis Benon is constantly looking for ways to make this accessible and free to use for many people around the world, while also finding reasonable ways to monetize and make this a profitable longer term business.
It's always difficult to find that balance between your customers, your employees, and your shareholders.
And Duolingo has adjusted its business model over time to ensure that that can happen.
Ty and I cover the fascinating founder story behind Duolingo.
The cover how Duolingo has succeeded in a largely offline market, how they have approached monetization, and how they plan to expand from here.
When we do these breakdowns, we often find some parallels to other businesses we cover, but the themes here are so interesting.
You have a Tam expansion story that maybe isn't quite the size of Uber, but has some similarities to Uber.
You have an organic growth story that feels like Tinder, where so much of this is happening through word of mouth, rather than through paid acquisition.
And similar to many great startups, you have a founder that built something that was incredibly important to him.
So please enjoy this breakdown of Duolingo.
I'm excited for Duolingo today.
And while I would group it in this new wave of technology companies, there's a lot that stands out from both the financial standpoint, the growth standpoint.
It's a fascinating story.
So excited to jump right into it.
And I thought it's a increasingly well-known name, but I think there's still a percentage of our audience that isn't familiar with Duolingo quite yet.
So maybe we could just start with the one minute short description of Duolingo for those unfamiliar with the business and the product itself.
Duolingo is not yet to become a household name, but for people who are learning languages, I'm sure that they all have heard of Duolingo.
The company is most famous for its language learning app that offers over 120 different courses.
Since the founding in 2011, there have been over 500 million downloads.
And Duolingo is constantly ranked as the most downloaded app in the education category globally.
But Duolingo was set up with a broader mission than just teaching languages.
They want to develop the best education in the world and make it universally available.
That is why today, in addition to languages, people can actually learn literacy, maths, and music with Duolingo.
Yes, you mentioned the mission there.
And one of the most interesting things about this business is its founder and the founding story.
So I want to get into that if we could share that right off the top.
I think the history here, it's not a long history, but the founder has quite the interesting past.
So maybe you could share a bit on that.
I always find Duolingo's founding story fascinating and it all started with the founder CEO, Luis Vonan.
He grew up in Guatemala, which is the poorer country in Central America.
And he was fortunate to be born in an upper-meter-class family and was given the best education by his parents.
And that made him really appreciate the importance of education in improving people's lives.
But at the same time, he always felt that education could be among the principal sources of inequality.
The most privileged people can get the best education in the world.
But the poorer may not even get the basic schooling.
So he believes that people should have an equal access to education regardless of their wealth.
And his profile is nothing shorter than impressive.
He came to the US for an undergraduate degree in maths, a Duke University, and later a PhD in computer science at Carnegie Mellon.
And before Duolingo, he was actually famous for his genius inventions of CAPTHA, which I'm sure the audience here know very well about.
She is the annoying task that distinguishes between the humans and both, that pop up when you lock in an internet website.
And interestingly that although CAPTHA was really popular and was really useful, he started thinking that it actually wasted a lot of people's time and his estimation mounted to something like 500,000 hours a day that people wasted on typing those letters.
So he reinvented CAPTHA with the same goal, but with a twist to turn those hours into something useful.
The letters that people have to tie are actually words scanned from books.
And so user would complete the security test why helping digitize the books for the internet archive.
And I think there was genius.
So later he sold that CAPTHA business to Google along with another game that he invented for tens of millions of dollars.
And I also heard a story that he got a personal phone call from Bill Gates that offered him a job at Microsoft, but he refused.
Not that many people returned out and offered from Bill Gates, I suppose, because he just wanted to work in academia and be a professor in computer science and devote his entire life to education.
And that is until he met one of his PhD students named Severin Hecker.
They were both passionate about education and had a strong belief in equal access to education.
So that's why Dueling Go was founded.
And the reason that they started with languages is because Lewis wants to deliver the highest possible impact.
There are nearly two billion people in the world learning languages.
And for lots of people learning a language can unlock tremendous economic opportunity, especially for English.
Lots of people in developing countries who know English that can double their potential income.
But beyond that, there's also benefits of unlocking new experience or building deeper human connection with family and friends from new languages.
I remember Lewis once said that Dueling Go can teach geography, but the impact of that will be quite limited.
And for that reason, they wouldn't teach geography.
But recently they expanded to things like literacy, music and maths.
And you can imagine the impact of reducing illiteracy rate in the world, even just by a couple of percentage, could be quite enormous.
Absolutely. I loved the idea of wasting people's time and finding a way to give it back and really taking tasks that would otherwise be viewed as inconsequential and a necessary evil and turning them into very productive activities.
And I think a lot of what we're doing when we are performing the capture and recapture tests actually have a lot of impact that we don't necessarily tie it to.
But that has evolved quite interestingly.
And that story, I think, is beautiful obviously because it comes again from a personal mission to drive change.
So I'd like to transition a bit into the language learning market as I was growing up the infomercials for the Rosetta Stone, where widely available.
And that was kind of the one vendor that I knew of well besides the private tutoring market, I guess.
But maybe you can give an overview of what that market looks like today, some sense of how it's evolved over time and any players that you think are relevant in that market.
Yeah, sure. Maybe you should start with the size of the market.
Language learning is a very big market.
As I mentioned earlier, there are nearly two billion people in the world learning languages.
And a very interesting aspect to me that there could be some degree of dynamism in the market in the sense that because of the ease of access and affordability that Dueling Girl brings about, they can actually open up their learning market.
And Dueling Girl conducted a survey in 2021, and they found out that almost 80% of the Dueling Girl users in the US were not learning a language when they began using Dueling Girl.
So it's really talks about that idea of if you make something really easy and fun and accessible, then a lot more people would use it.
In terms of the structure of the market, the majority of it is still offline.
So most people would learn languages by attending classes at schools or at language centers.
I guess that the online learning wins over the offline learning from the access perspective because it's cheaper, it's more convenient, it's accessible from almost everywhere.
But what is, I think more interesting is that the Dueling Girl also has a potential to win over the offline channel from the content perspective because in the offline world, learners would receive the same curriculum from textbook, from the teachers, unless they have a one-to-one human tutor.
But on the other hand, Dueling Girl through data and machine learning can truly deliver personalized curriculum, best fits to learners' abilities.
And it is something that really interesting that we can talk more later in our conversation.
But what I mean is that Dueling Girl can really bring that one-to-one human tutor experience to every learner at the fraction of the cost of the offline world.
But having said that, there are still aspects that Dueling Girl is not yet comparable to the offline world, especially when it comes to the depth of the content.
So the app currently only pushes people low to intermediate levels and speaking and writing exercises are still quite limited.
There is something that the company is very aware of and they're going to continue to invest very heavily on to improve the teaching effectiveness.
Can I ask when you say low to intermediate levels, do you have a way of putting context around that if I were to think of someone having an intermediate level understanding of the English language, where would you put them into society or in the workforce, how far could that get you, and just any context around what that would mean to have an intermediate level understanding?
Dueling Girl uses an international skill of language certification and in that skills you would have levels from I1, A2, B1, B2, C1, C2.
A1 is the most basic, C2 is kind of the advanced level and right now Dueling Girl teaches up to the level of B2 in some of the language but most of the language courses are still quite basic.
But the aim is to not to teach people to the advanced level because Lewis thinks that there's going to be diminishing margin or return.
What they want is to be able to teach people to the level that they can get a job with that language skill.
They're not going to teach people so that the students become like a poetry or being purge who ride beautifully is not the purpose.
And so yeah, that's why I say that most is basic to intermediate and they are trying, but with English, which is the most learned language in the world and I would say the most important in terms of improving career and potential earnings, they're trying to put in more advanced content into the courses.
That makes a lot of sense.
Okay. I think you described a bit in terms about the existing market where that offline market is still so large.
I think it's very interesting when you have examples of businesses that expand their markets.
I think Uber being one of the best examples, simply looking at the taxi market was not enough because Uber truly did take that market and multiply it.
Seems like Duolingo is doing something similar here when you mentioned 80% of users or not previously learning a language.
Can you talk about that growth story because to go from zero and there is an existing playing field in terms of competitors, how did they successfully turn on the business and then turn on growth such that they've gotten these users and all that you mentioned in terms of app downloads.
What was the secret to their success there?
I think maybe to set the scene for that, let's talk about what makes Duolingo differentiated in the first place, what makes it stand out in the very crowded language learning apps out there.
So the company always believes that the hardest part of learning a new language is to stay motivated and to keep the study going because learning a new language is hard like a lot of time people just want to give up.
So Duolingo has a very strong gamification approach to keep people engaged.
Action language course it decides to be like a mission in which learners would have to unlock lessons in sequence.
So they must do sufficiently well in one lessons before proceeding to the next and that helps bring out the natural competitiveness in individuals.
There's also a lot of features that are borrowed from the gaming world such as in-app currency rewards, remaining lies or help boosts et cetera.
A lot of the apps are they are more focused on grammatical rules but Duolingo's simplicity, light lessons and methods of intuitive learnings really make it easy and fun for people to study.
So if you use Duolingo, you would know about their distinctive feature called Strik which is how many consecutive days that people are learning with Duolingo and if you use a mister Strik then the Strik will be reset and actually people are obsessed about their Strik and many share their Strik on social media as a proud achievement and millions of users today have Strik
longer than a year. And so features like that really keep the engagement going which is something very hard in the education category unlike Netflix or Spotify which is like entertainment.
And besides gamification, another differentiating feature of Duolingo is that it is free to learn.
Yes, you can learn entirely with Duolingo without paying for anything.
And other competitors allow you to access to their lessons through subscription and because of this freemium model, Duolingo has been able to grow really fast because when the product is free to use people are more willing to give it a try.
And the result is that today Duolingo has over 90% share of the global online language learning monthly active users which I think is quite impressive.
And so that has set context a bit but I think if you ask about what makes Duolingo grow really really fast then I would attribute it to three factors.
So the first thing is the product-led strategy.
Luis Vius is always that customer will stay with you for your product.
So investing in product is really important.
About 70% of the employees are working in engineering, product development and design and all the product features from the design of the button to when and where they are displayed on the app are very well thought out and methodologically tested.
So Duolingo is very famous for their AB testing for every product launched.
On average they run about 500 AB tests per quarter and that really shows how the feature is going to affect engagement, teaching efficiency or monetization.
And the result of that is the very good products that just attract users.
And people may look at Duolingo and think that it is just a little queued up but actually there's quite a lot of sophistication behind the technology.
Another reason is the freemium model which I mentioned earlier if it's three more people are willing to try it.
And lastly I think Duolingo is very good at social marketing.
They don't spend a lot on marketing but their marketing is super efficient.
They really know how to tap into culture and humor to create interests.
So everyone using Duolingo would know about their mascot, the aggressive green hour named Duol.
Duolingo uses Duol as a protagonist to build content which can be a bit silly and stupid but that silliness makes people remember and talk about.
So you can find a lot of memes on the internet about Duol based on the messages that Duol send out to people to remind them to take the lessons every day.
So the brand has become really strong because of that social marketing.
And I remember laughing a lot when I saw photos of Duol dressing up as a Barbie because Duolingo was referenced in the Barbie movie and the mascot appeared at the Barbie premiere which is really fun.
Or recently they just spend about $700,000 on a five-sake and super bollette but receive over 600 million impressions on social media which I think is super efficient.
Yeah it's truly a fascinating example where gamification is used for what I would consider a very good cause here in terms of education and taking something like education which can be very dry and painful and turning it into something fun where it sounds great to be idealistic about that stuff.
It's very rare that you see the case that you can realistically do that but they seem to be achieving that quite well.
One of the data points that I read and I don't know how up to date it is now is that 90% of new users were organically acquired which is a metric that the only other app that I can think of that was anywhere close to that or in the same ballpark was Tinder which again had this organic natural explosion and growth story.
Would you attribute that mostly to the social dynamics of people wanting to post their streaks and wanting to talk about it on social media and other areas where it becomes this social status game?
Is there anything else that you think was playing into that organic growth without having to spend a ton on paid marketing?
As I mentioned earlier I think it firstly it was because of the products.
The product is just very good and it's really fun to use because once you have a product that is good it makes people wanting to keep using it and then recommending it to their friends and family.
And couple with that is that very efficient social marketing so they created a marketing never with the intention to make people subscribe more to Dueling Go but always with the intentions of making people wanting to talk about it.
And so as you mentioned people just keep talking about Dueling Go on social media posting their achievements or just talking about how silly the message the green owl has sent them to remind them to take the lessons.
So the combinations of the product and the efficiencies of the social marketing has created that social bust and allowed them to acquire users in such an organic way.
Before we transition a bit in terms of the business model.
In terms of the market size today you mentioned the beginning there's two billion in the world that are interested in learning a new language.
In terms of the existing market do you have any sense of the online versus offline market and just the existing size of those today or just the general share.
It's really hard to get a precise estimate.
I have seen lots of number of being thrown so the language market size both offline and online estimated to be over 100 billion US dollars.
Online learning is the fattest growing market segment within that.
It was projected to grow from about 12 billion in 2019 to about 50 billion dollars in 2025.
But I think the main point here is that easy is a very large market especially when comparing to Dueling Go's current size of about 600 million dollar annual run rate.
Yes they have a lot of share in the existing market and a lot of share in what could be a growing market to gain in the future that lays it out nicely.
On the business model you mentioned it is a premium model which I think was very invoked at a point in time and it's become more challenging for people to figure this out with subscription businesses.
Not everyone seeing the conversion rates they would like to see when going from premium to offering a premium tier.
Can you talk a little bit about that evolution over time and how they've gone about it?
I think maybe it's worth talking about the evolutions of Dueling Go itself before touching on the evolutions of monetization because it would help you understand and appreciate the monetization evolution of the business.
So the first versions of Dueling Go is nothing like it is today.
The experience of recapture instilled the idea of crowdsourcing into Lewis's mind.
So Dueling Go actually started with a web-based product that helped people learn a language by translating the website.
So on one side you have people wanting to learn a new language.
On the other side you have websites, firms and others who want to translate their content into multiple languages as quickly and cheaply as possible.
So a native Spanish speaker who wants to learn English could be given a sentence from an English website to translate it into Spanish.
And Dueling Go would help users to practice new words.
They stumble upon through examples and flashcards and to pay for the bill Dueling Go would sell that user generated translation to paying clients while keeping it free for the learners.
But they didn't get a lot of success with this translation service and they were able to get only two clients first-feet and CNN.
And if you remember back in 2011 when the company was founded, the websites remained the dominant platform that tech startups was built on.
But the idea of a mobile app has started emerging.
So Dueling Go at that time was thinking about building a companion app and they actually hired two interns to start building an app prototype.
And that process really forced them to think much more strategically about what exactly they want to offer their customers.
And the interns actually came up with the idea of the skill tree that users would unlock to make progress.
And that is why Dueling Go was redesigned and became the Dueling Go that we know today.
And for the most part of the company history, they didn't focus on monetization.
Most people came to work for Dueling Go because of the company mission.
So to some extent, monetization felt like a scene to them.
And it's not until they faced great pressure from investors to make money.
And Lewis himself realized that for them to achieve the mission that they want to achieve, they would need to build lasting profitable business.
And that's when they started to think about how they can make money from this app.
And they came up with the subscription model that we have today that you can use it for free.
But if you want to remove advertising or you want to get access to some of the fun gaming features to help you learn a bit faster, then you pay for the subscription.
But you can see that from the original model, the translation service to the subscription model today, they really stayed true to their mission of making free education.
One thing is that they would never charge for education content.
It's quite easy for companies to just want bad users with the advertising and make the experience worse so that people subscribe.
But that is not how they're going to do it because Lewis thinks that it's going to jeopardize the business in the long term.
So I think that given the pressure of being a public company being able to show profits and growth, the fact that they could resist that temptation to make easy money is really admirable.
Yeah, I'm not one to typically judge negatively in terms of how businesses make money.
But there's something I will say that's admirable about the original approach even trying to work with businesses that were essentially funding away for people to learn and evolving that over time and seeing the different things that you could try out.
What does that split look like today just in terms of the different revenue buckets?
I know there's a third that we haven't gotten to yet in terms of assessments.
But between ads, subscriptions and then to the extent that assessments are a material revenue item, what is the split in those buckets?
So Julingo makes money from three sources.
As you mentioned, the first and the biggest part is through subscription, which is about three quarters of the total revenues.
And the rest could be advertising revenue in purchase revenues and English test certification revenues, which are roughly equal to each other.
That is how the business is made up.
And do you have some sense in terms of the user base that number is quite large in terms of 75% of revenue is coming from subscriptions?
What percentage of users are converted to subscribed paying users?
In the most recent quarters, they disclose that they have nearly 90 million monthly active users and close to 30 million daily active users.
And of those users, nearly 7 millions of them are paying subscribers today.
And that is compared to just over 2 millions at the time of the IPO in summer of 2021, about 4% of the monthly active users are pets.
But today it's close to 8% so that has been increasing.
And I think if you think about the growth rate in the numbers of monthly active users, that makes that number even more impressive because your denominator is growing rapidly as well.
It's really interesting that people often ask the management team, what do they think going to be the ceiling for the penetration?
Because you have businesses like Spotify who get close to or more than a half of the users actually paying for the service.
And the company would often come back and say that honestly, they don't know.
They don't really know what the ceiling is.
You have businesses like Bumble and Tina, the dating apps could have about 10% of their users paying for the service.
And they are on the upper hand, you would have Spotify.
So they think that they may achieve some way in the middle.
But honestly, they don't know what it is.
As long as they continue to deliver the value to the subscriber, then they think that more people would be willing to pay for it.
And what is the cost of a subscription for Dueling Go?
It varies based on different regions.
And also it depends whether you subscribe monthly or annually or whether you subscribe as a standalone users or as part of the family package.
But today, an annual planned in the US, I think costs about 80 to 90 dollars annually, which is really cheap compared to the price that you have to pay to go to language centers.
Yeah, one session with a language tutor.
And you can get an annual subscription.
When you think about the user base today and how that's evolved, I think there's different ways of framing this for a lot of these businesses, whether it's cohorts who initially come in, how long do users typically remain paying members?
Is this something where once they achieve a certain level, they will turn off of the app?
Is it something that they are continuously using?
Is there any segmentation in terms of the audience base that you find particularly helpful?
We don't get that granularity, but I think it's interesting to think about who's actually used Dueling Go.
So most people use Dueling Go because they want to learn new language, especially English, half of the users today learn English.
And the motivations for them are very, some learn because they want to have a career development, some simply want to learn to connect to friends and families.
There are people who simply want to learn some words before they holiday to a foreign country, some language enthusiast who just love learning.
But there are also people who are influenced by trends.
So Dueling Go say that after a sweet game, if you remember the popular Korean Netflix drama, they actually sort of spike in the number of people wanting to learn Korean.
So I guess that given that you have such a very base of users knowing exactly when people terminate is quite tricky.
But for people like who are language enthusiasts, as soon as they finish one course, they could move to the next because they just keep wanting to learn new things.
For people who are a lot more serious about learning to have a better career prospects, Dueling Go say that to finish all of the content and to be very good at it, it could take them a couple of years.
So it's really varies. But I think the more interesting question to ask or to think about is like why more people are paying to use Dueling Go.
Despite the fact that they can learn it for free, I always find it intriguing.
And one of the questions that we have when we first invested in Dueling Go is that if the free products are really good, why would people pay?
And the fact is today they're nearly 7 million people subscribe, right?
And the company say that the first reason is because a lot of people still do not want to see advertising, even though advertising are not that intrusive.
Second is because the subscription is really cheap if people use Dueling Go intensively.
So they have observed that people tend to subscribe to Dueling Go once Dueling Go becomes their daily habit because the price per minute of engagement would be very low if they use it.
The third reason is that they have been developing more valuable features that subscribers good value.
Last thing, which I find really interesting is that people love Dueling Go and actually want to support the company's mission.
So supporting Dueling Go's mission of free education is actually listed as one of the benefits of the subscription.
And Dueling Go's have not found any sophisticated patterns among paid users other than simple things like they tend to use more expensive phones.
They live in a richer areas in a richer countries and they use Dueling Go a lot.
So in a way, Dueling Go's sort of created this Robin Hood model where the richer people actually pay for the poorer one to get access to better education.
And I think that is very amazing.
Absolutely. One of the segments of the business that I found particularly interesting just in terms of the opportunity that it opens up is the assessment and certification side of the business.
It seems like Dueling Go is now actually being recognized as a assessment of record as a certificate that has some validation and is accepted broadly in terms of the workforce.
Can you talk about that mission to evolve and develop the assessment line and the assessment business and how that's gone and the expectations for the future there?
Yeah. So in parallel to language teaching Dueling Go has the English certification business.
And the market is dominated by traditional tests like the Toffles and IELs, but those tests have several limitations.
The first is that they require candidates to go to test centers which are often located only in big cities and candidates often have to book months in advance to attend the tests.
The second is that the test is several hours long and it can take like two days for different sections.
The third is that they are very expensive.
They usually cost 200 to 300 dollars a test.
So Dueling Go is disrupting this market by offering an online test that candidates can take in the comfort of their own homes, cost only $49 and requires only an hour.
And the reason that it can be that short compared to the tradition test is because Dueling Go uses computer adaptive tests as opposed to traditional fixed form tests.
What that means is that as the candidates move through the tests, the questions given to them are determined by their responses to the previous questions.
So they wouldn't waste time on questions that are far below or far above their proficiency level like in the traditional tests.
And today the test has been accepted by nearly 4,000 universities in the US, including the top Ivy League schools.
And they're also accepted by the Irish government as part of the visa process for the immigrants.
In total, there are about 10 million English tests being taken every year.
So at a $49 per test, you can think that the market site is about $500 million.
That is not last and that is not exciting in itself.
But I think the test is important to help Dueling Go increase the credibility and it helps Dueling Go become the standard of the market.
And more interestingly is to get people to learn through Dueling Go.
Because if you think about it, then the market for selling tofu in IELS practice materials and training learners for the exam is multiplefold larger than the market for the testing itself.
This can be quite an important avenue to get people to come in and learn with Dueling Go.
I think that's such an interesting point in terms of the prep market being larger than the market itself and probably the case for many standardized tests.
But to your earlier point, I could easily see that market size also being something that expands when you have a tool like Dueling Go really breaking barriers.
And I certainly from the corporate side of things would not mind paying for candidates to participate in that test.
You could see that evolving in a variety of different ways.
I think it's just such an interesting point in terms of how they've built and expanded into an area that maybe is non-obvious but can have all types of positive ripple effects.
As you were describing the test and it's a dynamic test, I think it gets into the DNA of this business which is machine learning AI driven well before it was as popular as it is today.
So can you talk about how they've incorporated the dynamic learning process by using AI?
Whatever it is about AI that they're incorporating into their model and how you expect that to evolve in the future as well.
Dueling Go has been using AI for a long time even before all the assignments about AI recently.
With over a billion exercises completed every day on Dueling Go, they are able to leverage this data to develop AI models to keep users more engaged and to teach them better.
As an example, they develop an AI model called BertBrain to predict the probability that any learner will answer a given exercise correctly.
Dueling Go uses BertBrain predictions to adaptably construct lessons where each exercise is just right or learner in terms of the difficulty.
So they think that 80% of answering an exercise correctly is about right.
That would help you increase the engagement because if you give users a too easy or too difficult exercises, then that would either bore them or frustrated them.
So that's why as I mentioned earlier, the lessons on Dueling Go are truly personalized based on the learner's proficiency level.
And recently with the more rapid development of AI, Dueling Go has partnered with OpenAI and launched a new subscription tier called DuelMux that incorporates generative AI features.
One of the features that I find really interesting is roleplay in which users can have a real time conversations with Dueling Go powered by generative AI.
So there would be different plot developments suggested to users so that users can emerge in different real-life situations and improve their conversational skills, for example, asking for direction or ordering a cafe in a street of Paris.
And if they are stuck at any point in the conversation, then there will be suggested words, then phrases to continue.
And at the end of the conversation, Dueling Go would give them score based on the variety of the words that they use or areas that they can improve in the future.
And where they want to improve this offering further is to mash the difficulty of the conversations to the users' level of proficiency and also to reduce the latency of the AI's responses.
And with the generative AI, Luis believed that Dueling Go can teach better than a human tutor.
And as the cost of language training model continues to go down, at some point, they hope to be able to give these AI features to lower subscription tier and even to the free user so that everyone can benefit from the power of AI.
It's one of the better use cases that I can think of when it comes to AI.
I think there's often, particularly in the financial market, some frustration when trying to use the new AI features for factual information, but the social dynamics and the advancements that have taken place when it comes to the social interaction with AI is completely different.
And then hearing how that can be used for a user in real-world scenarios is just such a natural way to take the tools and put them into practice.
So I love hearing about that.
I want to get into the cost dynamics that are associated with building those models, but maybe we can just start from the top on the financial side of things and get into the financial model a bit.
You talked about the revenue lines and the different buckets.
This is one of the unique, new SaaS companies that is actually gap profitable.
So it stands out in a sea of other businesses that have been heavily investing in growth.
Very interesting here.
So maybe you can just describe the margin profile or however you would think about the income statement when it comes to do a lingo and how you frame that financial model equation for them.
I think you're right. I think the business really stands out in terms of being able to grow really fast and being profitable at the same time.
And if you think about the structure of the business, then it should be very profitable.
They make over 70% gross margin.
Most of the cost of revenues are platform fees paid to Google plays and Apple stores for having their apps.
The biggest operating expense for them is R&D, which has been around 40% of revenues because they continue to invest heavily in products, not just languages but also math and music.
But sales and marketing, as we said, are minimal because most of the growth is organic and worth of mild and they don't have to spend a lot on that to acquire users.
So at steady stage, this could be a 30, 40% operating margin business.
And actually when we look at Tinder, which has a similar subscription model, then that business has achieved over 40% operating margin.
So dolingo should be able to get that at some point.
They have just generated profits at the net income level in 2023, which is a very big achievement.
But the business actually has been very cost-generative for quite a long time.
And their free cash flow margin has been closed to card percent.
And I know the dynamic with many of these businesses, you have the free cash flow, but you do have this share-based compensation and the stock-based expense, which is not cash flow, but is viewed certainly as an expense to many.
It isn't expense. How do you think about that as an investor, just with the dynamics of being a big piece of the equation?
How big is it for dolingo?
And is there any expectation in terms of how that line item will evolve over time?
Yeah, I think you're right that stock-based compensation has been quite a common way for tech companies, especially in the US, to pay for their employees.
But your dolingo does use this stock-based compensation that has been coming down as a percentage of the revenues because the company has been growing pretty fast.
And we factor that in terms of the dilution for the shareholders of the long term.
But as I say, they have been able to generate profits at the net income level in 2023.
And we think that over the long term it should be profitable business as all of these get leveraged.
And on the line items from a revenue standpoint, you mentioned again, math and music being another potential area of growth for them.
Do you have some sense of what that represents today in terms of whether how many subscribed users are actually using it for math and music, whether it's mostly people that have used it for language that are then layering on math and music, anything about those two areas of growth?
Math and music are still very early and they probably weren't contribute to revenues for the next few years.
Dolingo is currently at a stage of product development to attract users in just to be more content.
I think you're right that most people today use math and music because they use Dolingo for language.
And math and music are now integrated into the flagship language course.
So is all there in the art that people can cross-learn.
Interestingly, I read an article recently about why the UK has a problem with maths.
And lots of students are schools and not fans of math.
They skate of numbers and as a result, they struggle with basic math concepts in their adult lives.
So if Dolingo can apply the same method of teaching language to maths to make it fun and engaging, then I think it could be very exciting and impactful.
And the company actually believes that there will be a wide audience for maths, including both students and lifelong adult learners.
And they also believe that their brand can transcend language learning.
So they want to be synonymous with education, not just language.
And it's about a unique weight of teaching.
So learning something with Dolingo's way is how they marketing their brand.
I think of math as kind of an extension of language, it's its own language in many ways and music in a similar way.
So we'll be very interesting to monitor how that evolution goes.
As you mentioned, they seem to be having success educating people that are interested in the education, but maybe not historically keen to go through the exercise and the painful process of what education typically entails.
It begs a question of when Dolingo potentially become more featured in classrooms and universities.
Is that something that's happened?
I would think about it being something that mostly users are using on their own free time, but it does feel like something that if leveraged by traditional education and the existing industry could make it even stronger.
So has that been a goal? Has that been something that they've seen any success with?
Where does that stand? Interesting.
I used to ask management team that questions whether they would consider selling to schools or selling to governments or NGOs, etc.
And they say that they have that discussion early on in the history of the company, but they ended up choosing to go down the route of direct to consumer because that would allow them to grow a lot faster and being very customer-centric is true to their DNA.
And that would allow them to really have a close relationship with the customer and incorporating all the feedbacks and then do their AB testing, which I talked about earlier.
Having said that a lot of schools in the US are actually using Dolingo for free.
40% of foreign language teachers in the US, K-12 schools, actually use Dolingo in their classroom in some forms.
And there are more people in the US learning languages on Dolingo than there are foreign language learners in all US high school combined, which I think is quite impressive for companies of this size.
Yeah, I guess you can have that type of adoption without making it a direct mission to have that type of adoption.
And it seems like it's been the case with this business since the start.
Just organic word of mouth can just be one of the most incredible drivers of business growth and probably the best driver of business growth.
So much fun going on here within this business and the way they approach things.
Are there any other interesting dynamics when you think about the growth story and the growth opportunity?
What stands out to me is I would have thought they could have just continued to lean into the language market.
It's clear that they're already expanding with math and music into broader education.
So they're certainly going after a lot and not necessarily just continuing on one specific road.
They're already branching in a few different directions.
I think I remember Louise once say that they're not going to exhaust of things that they can teach anytime soon.
So I think they would continue to think about what they can teach in the future using the mobile app if they would love to be able to teach people coding.
But it's something really quite difficult to be done on a mobile phone.
But yeah, I think in the next five years or so, I think it's going to be about language continues to deepen the language market in terms of having more contents that are relevant for people and achieve their missions of teaching people till they are able to get a job with that skill and also building out the audience from math and music and starting to think about how
they can monetize those users.
So I think in the next five years, I would see that is how they grow.
I think one very interesting thing for me to wash the business is how they can continue to advance their technology so that they can teach people better because one of the questions that I had in the early days of investing in Duelingo is that they have been doing a fantastic job at engaging people and making people wanting to come back and learn with Duelingo.
But do they really teach people effectively?
And I think with the help of technology, they would be able to do that.
So the AI example is something that I mentioned earlier and feel really excited about.
Just on that point in general, do you think that they are largely seeing success, teaching people?
I think the certification certainly felt like that was some form of validation or credibility.
But when you think about the effectiveness of the product, do you have any sense?
Obviously, once you get into gamification, many of these things can be used as heart learning, but a lot is the interest in the gaming side of things.
But any general opinion on there in terms of where they stand today, in terms of the effectiveness and how much better it can get.
Internally, the companies have three most important metrics that the monitor, which is the engagement, the effectiveness of their teaching and monetization.
And a lot of times you would see there would be trade-offs between either engagement and monetization or engagement and effectiveness of the teaching.
But they want to make sure that they are able to balance those things right.
And I think in terms of teaching effectiveness to your question, there have been quite a bit of gap in different courses that they have.
So some of the flagship courses like learning Spanish and French from English, for example, are really good.
There have been a lot of resources being put into developing the content and having different forms of learning for users.
So usually I read reviews of the internet and find out that people are really like the offer.
But there are courses, for example, which I try like learning Chinese from English hasn't been that good.
So there's been quite a bit of inconsistencies between the different courses that they offer.
And they are very aware of that.
But there's limited resources that they can put into everything.
So they're going to prioritize based on the popularity of the languages and how many people are actually using the course.
It's helpful to get a sense like that of where there might be still opportunities to improve on the existing product and how they go about prioritizing.
As you think about risks to this business, I think we've touched on what differentiates them from competition.
There are certainly other startups in the space.
There are the incumbents.
There is the offline market.
Is there anything that's happening when it comes to the competition that you view as a threat to Dueling Go specifically?
Yeah. So I think you're right that there's always been competition.
But when I think about the risk for this business competition is not the first thing that came up to my mind, I don't mean to say that there wouldn't be competitive threat.
But I think with the positions of the company today, having people coming to them organically and Dueling Go becomes synonymous with language teaching, it's going to be quite difficult for someone to come in and attract users.
Provided that they continue to focus to develop the best products.
What worries me the most when it comes to the business is the key men risk.
Dueling Go today is many times larger than it was in the past and has attracted many talents to come working for them.
But Louis remains the key architect behind the company.
He's still very much involved in all the decisions about products.
That is good, but it raises the slight concern about the scalability of the business because if he has to control everything, then does that potentially slow down the business or as the company gets bigger and he cannot touch on everything.
So whether they have a sufficient death in the management team to make the right decisions to still balancing that profitability and growth, to staying true to their missions of providing free education to people, I would say that is more of a concern than from the competition standpoint.
Yeah, certainly. He throughout the research and this conversation today seems like a very important piece to this puzzle.
I did find it interesting.
The management team owns a large percentage of shares, I think still over 15% is owned by the founders, which shows that dedication and alignment with the business, but to your point, there can be positives and negatives and drawbacks when it comes to scaling sometimes, when it comes to those dynamics.
In the world of AI, it seems like a tool that they are very much leveraging and will benefit them as they incorporate it.
Are there any threats when you think about AI and just all the different things that are happening now with our own ability to take this episode and have it translated is truly incredible in terms of the advancements that have been taking place.
But how do you view AI, either as a major tailwind to the extent that it can become a headwind?
How do you frame that? I think so far, Dueling Girl has been a beneficiary of AI, but people are making stone where the AI would be developed to a point where we no longer need to learn new language because we can just have real-time conversations in different languages.
So I can speak in Vietnamese and you can understand because there would be a real-time translation into English, for example.
I think it is possible, but I guess that learning a new language brings a different perspective in terms of understanding more about the culture and the people that come from that culture and through language learning, it can help build human connection in a deeper way.
So maybe even if the technology allows us to do that, there's still demand for people to learn new languages.
But yeah, we'll see. Interesting scenario to think about at the very least, for sure.
Well, this has been an incredible conversation about just a very wholesome business, not a word that I would typically use, but it seems like the mission feels true and truly aligned with the founder who has found his life's work.
When you think about the lessons that you can pull away from Dueling Girl and this specific example and apply elsewhere, what key lessons stand out?
I think for me, I would say two things.
So investing in Atec has not been easy for investors and a lot of investors consider Atec as a great yard.
There's a lot of reasons for that, but I guess one of the reasons could be because of the tricky balance of mission and monetization.
Education is often treated as a public good and it's not easy to make money from education, especially for an Atec company where founders often have a big mission and want to lower costs and improve access for people.
But at the same time, they have to generate profits and returns to satisfy stakeholders, including shareholders and their own employees.
So not many companies can achieve both of those objectives.
I think Dueling Girl so far has proven that there is a way to do that, although it is the early days.
So I think what we need to see here is a trusted management team who is really committed not to fall for the short-term profit and forget the mission somewhere down the line when things turn difficult.
And a big part of the reason for our investment in Dueling Girl is the trust on the management team, especially Louise.
Another lesson is that product really matters.
I mean, that sounds really trivial, but it could be easily forgotten.
Many investors, especially the early venture capital people who gave money to Dueling Girl because they believe in Louise and believe in the product that he created rather than necessarily seeing a clear path of how Dueling Girl would make money.
It is to believe that as long as you have a good product that brings value to your customer, then somewhere down the line you would find a way to make money from it, that is not to say that other things like sales and marketing do not matter, but I do think that they come after the product.
And for a successful business, I would want to see them being obsessed about products and always wanting to make their products better and bring more values to the users.
I think it's an excellent point to close out on and something throughout the conversation we talked about the social media strategy and the status dynamics, but at the end of the day, if the product isn't great, then all that engagement doesn't necessarily mean much.
So this has been fascinating, Taya.
I really enjoyed bringing down Dueling Girl.
And this is one of those companies where I do some research and I just know that I will be watching this one and think it's just such an interesting case study to follow for years to come.
So thank you very much for joining us.
Thank you. To find more episodes of breakdowns ranging from Costco to Visa to Moderna or to sign up for our weekly summary, check out Join Colossus.com.
That's J-O-I-N-C-O-L-O-S-S-U-S.com.
A quick note before you go.
If you are a company hiring or a candidate looking for your next opportunity, make sure to check out Join Colossus.com slash Recruiting.
We launched our recruiting efforts at the end of last year.
We started working with firms that were interested in tapping into our audience.
And after seeing some early success, we want to open this up to additional firms and to candidates who are proactively looking for their next opportunity.
So we are mostly revolving around the investment world and the tech industry.
But again, make sure to check out Join Colossus.com slash Recruiting for more information.
