Around 34 million people use the language learning app every single day, some for well over a decade, making the company around $500 million annually and making it one of the top education apps in Apple's App Store for the past decade.

每天約有 3400 萬人使用這款語言學習應用程序，有些人已經使用了十多年，每年為公司帶來約 5 億美元的收入，使其在過去十年間成為蘋果應用商店中最受歡迎的教育應用程序之一。

And yet, most people who use Duolingo do not pay.

然而，大多數使用 Duolingo 的人並不付費。

All of its lessons are completely free to access.

所有課程完全免費。

Only around 9% of its users pay, and less than 8% of its revenue comes from advertising.

只有約 9% 的用戶付費，而來自廣告的收入不到 8%。

So how does a free educational app manage to also have a valuation of $7.7 billion?

那麼，一款免費的教育應用程序是如何做到估值高達 77 億美元的呢？

And is the company's strategy sustainable?

公司的戰略是否可持續？

The answer lies in AI, A-B testing, and this little green owl.

答案就在人工智能、A-B 測試和這隻綠色小貓頭鷹身上。

This is the economics of Duolingo.

這就是 Duolingo 的經濟學原理。

If you've ever used Duolingo, you'll have noticed it's less like a traditional language lesson and a lot more like a game.

如果你用過 Duolingo，你就會發現它不像傳統的語言課程，而更像一個遊戲。

Users have lives.

用戶是有生命的。

Get five answers wrong in a day and you'll run out of lives for that day.

一天之內答錯五題，當天的生命就會耗盡。

Lessons have progress bars and are laid out in a journey.

課程有進度條，並以旅程的形式呈現。

Players gain experience points.

玩家可獲得經驗值。

And streaks encourage users to play every day.

連勝鼓勵用戶每天都玩。

The company says that's part of what helped attract some 3 million users within the first few years.

該公司表示，這也是在最初幾年內吸引約 300 萬用戶的部分原因。

When we launched, we had a thing that was mobile, free, and fun.

當我們推出時，我們有一個移動、免費、有趣的東西。

And because of that, very quickly, it overtook all methods to learn languages.

正因為如此，它很快就超越了所有學習語言的方法。

But before long, Duolingo ran into an issue with its mission to try and deliver free language lessons.

但不久之後，Duolingo 在嘗試提供免費語言課程的過程中遇到了問題。

One of our board members who had just invested in Duolingo, she said, look, you can't continue doing this.

我們的一位董事會成員剛剛投資了 Duolingo，她說，聽著，你不能再這樣下去了。

You've got to make revenue.

你必須創造收入。

We had not given ourselves permission to make any money from our app.

我們沒有允許自己從應用程序中賺錢。

It's hard to build up new capabilities in an organization.

在一個組織中建立新的能力很難。

We were in a bit of a pickle because we wanted Duolingo to be a free way to learn languages.

我們陷入了困境，因為我們希望 Duolingo 成為一種免費的語言學習方式。

So in 2017, the company introduced advertising, which now accounts for under 8% of its revenue.

是以，2017 年，該公司引入了廣告，目前廣告收入佔比不到 8%。

It also launched a subscription tier where users could pay $10 a month to remove ads and get unlimited lives.

它還推出了一個訂閱層級，用戶每月只需支付 10 美元，就可以去除廣告並獲得無限的生命。

The company went public in the summer of 2021 and saw a boom during the pandemic as users practiced languages while travel was limited.

該公司於 2021 年夏天上市，在大流行病期間，由於用戶在旅行受限的情況下練習語言，公司迎來了蓬勃發展。

Today, analysts say Duolingo is around 12 times bigger than its main rival, Babbel, by downloads, with some 8 million users who are now paying for the app.

分析人士稱，按下載量計算，Duolingo 是其主要競爭對手 Babbel 的 12 倍左右，目前約有 800 萬用戶為該應用付費。

And while that's only a small fraction of its overall user base, The majority of the money is made from this, you know, under 10% of the active users.

雖然這只是整個用戶群的一小部分，但大部分的錢都是從這些用戶身上賺來的，你知道，活躍用戶的比例不到 10%。

They give us about 80% of our revenue.

他們為我們帶來了大約 80% 的收入。

But while Duolingo's growth through gamification and subscriptions may seem pretty simple, there's a lot more going on behind the curtain.

不過，雖然 Duolingo 通過遊戲化和訂閱實現增長看似簡單，但幕後還有很多事情要做。

Every move the company makes is meticulously crafted down to the pixel, tested on unknowing users, and refined using AI, rigorous A-B testing, and constant tweaks to its app.

該公司的一舉一動都精心製作，精確到像素，在不知情的用戶身上進行測試，並通過人工智能、嚴格的 A-B 測試和對應用程序的不斷調整加以完善。

Whenever people are using the app, we measure how long they use it for.

每當人們使用應用程序時，我們都會測量他們的使用時間。

Do they come back the next day, et cetera?

他們第二天還會回來嗎？

And we try a lot of things.

我們嘗試了很多東西。

We try, when do we send you a notification?

我們正在嘗試，何時向您發送通知？

What we say in the notification?

我們在通知中怎麼說？

Do we make the lessons a little harder?

我們會讓課程變得更難一點嗎？

Do we make them a little easier, a little longer, a little shorter?

我們是否要讓它們變得簡單一點、長一點、短一點？

Take Duolingo's push notifications.

就拿 Duolingo 的推送通知來說吧。

If you forget to do your lessons and risk ending your streak, you'll get a little notification from Duo, the company's mascot, reminding you to return to the app.

如果你忘了上課，有可能終止連勝，你會收到公司吉祥物 Duo 的小通知，提醒你返回應用。

The messages on the push notifications started out simple enough, but quickly evolved to become the stuff of memes.

推送通知上的資訊一開始很簡單，但很快就演變成了流行語。

Including one particularly effective guilt trip.

包括一次特別有效的內疚之旅。

We decided that the last notification we would send you was telling you, hey, these notifications don't seem to be working.

我們決定，最後一次向您發送的通知是告訴您，嘿，這些通知似乎不起作用了。

We're going to stop sending them for now.

我們將暫時停止發送。

Sending that got people to come back a lot because they felt guilty.

因為他們感到內疚，所以他們經常回來。

It's like your mother telling you, like, okay, well, I'll stop, like, guilt tripping you.

這就像你媽媽告訴你，好吧，我不會再讓你內疚了。

Duolingo says that this led to a 3% rise in retention.

Duolingo 表示，這使得留用率上升了 3%。

But crucially, the messages that you receive aren't the same as everybody else's.

但至關重要的是，您收到的資訊與其他人收到的並不相同。

AI is used to start figuring out basically, like, which ones of these are more effective for particular users or at particular times of day or whatever, and letting the more effective ones rise to the top and then get sent to more people.

人工智能基本上是用來找出哪些資訊對特定用戶或在一天中的特定時間或其他時間更有效，並讓更有效的資訊上升到頂端，然後發送給更多的人。

To figure out which messages work best, Duolingo built something called a bandit algorithm, named after a reference to a one-armed bandit, a term for a slot machine in a casino.

為了找出最有效的資訊，Duolingo 建立了一種名為 "強盜算法 "的算法。"強盜算法 "的名字來源於 "獨臂強盜"，是賭場老虎機的一種術語。

Say there are, you know, several different slot machines.

比如說，有幾種不同的老虎機。

In this case, think of them as several different notifications we might send people.

在這種情況下，可以把它們想象成我們可能向人們發送的幾種不同的通知。

And we've tried each one some number of times and seen what it paid out each time.

我們對每種方法都進行了多次嘗試，看看每次都能得到什麼結果。

So we have limited information about all these different options.

是以，我們對所有這些不同選擇的資訊瞭解有限。

And so the algorithm has to figure out, do I just go back to the one that has so far given me the most money and just go back to that one every time?

是以，算法必須想清楚，我是否要回到迄今為止給我帶來最多收益的那個地方，並且每次都回到那個地方？

Or how do I balance that with exploring different options to see, maybe they're actually better, but I just got a little unlucky earlier?

或者說，我如何在探索不同方案之間取得平衡，看看也許它們實際上更好，只是我之前運氣不太好？

And each time we send a notification, we track, did it work?

每次發送通知時，我們都會跟蹤通知是否有效？

Basically, did they come back to Duolingo, do their daily practice?

基本上，他們有沒有回到 Duolingo 進行日常練習？

For example, the algorithm found out that the notification, Time for the Language You're Learning, worked better on Chinese learners than English learners.

例如，該算法發現，"Time for the Language You're Learning "通知對中文學習者的效果好於英文學習者。

It's mostly chosen by AI.

這主要是由人工智能選擇的。

We must have about 100 different things that we can send you.

我們一定有 100 種不同的東西可以寄給你。

They're written by people who are pretty funny, and some of them are also written by AI.

它們的作者都很有趣，其中一些還是人工智能寫的。

And that could be part of the secret to Duolingo's success.

這可能就是 Duolingo 成功的祕訣之一。

The company gets so much data back from those attempting its free lessons, that it has a huge database of information that it can feed to its AI algorithms.

該公司從那些嘗試免費課程的人那裡獲得了大量數據，從而建立了一個龐大的資訊數據庫，可以將這些信息反饋給人工智能算法。

They do something like 13 billion exercises every week.

他們每週大約要做 130 億次練習。

We are throwing all that data to AI and using that to personalise the learning experience for everyone to make sure people are learning effectively and staying engaged.

我們正在將所有數據交給人工智能，並利用這些數據為每個人提供個性化的學習體驗，以確保人們能夠有效地學習並保持參與。

At any given moment, the company is running hundreds of A-B tests like these, where one user is given a different experience to another to see which one wins out.

在任何時候，公司都在進行數百次類似的 A-B 測試，讓一個用戶獲得與另一個用戶不同的體驗，看看哪個用戶會勝出。

Some of the tests are AI-powered, others aren't.

有些測試是人工智能驅動的，有些則不是。

And for most tests, around 5% of its user base act as unknowing guinea pigs.

在大多數測試中，約有 5%的用戶充當了不知情的小白鼠。

I don't know how many A-B tests we run at this point.

我不知道我們現在進行了多少次 A-B 測試。

It's probably more than 10,000.

可能超過 1 萬。

We've A-B tested our way into getting more people to pay us, into getting more people to use Duolingo, more people to recommend it to their friends, etc.

我們通過 A-B 測試讓更多的人付錢給我們，讓更多的人使用 Duolingo，讓更多的人向他們的朋友推薦 Duolingo，等等。

So without having run those, we would not be a successful company.

是以，如果沒有這些，我們就不會成為一家成功的公司。

I'm certain of that.

我確信這一點。

So if Duolingo's algorithms and A-B testing mean that the company knows exactly what makes its users tick, why are subscriptions only at 8%?

那麼，如果 Duolingo 的算法和 A-B 測試意味著該公司完全瞭解用戶的需求，為什麼訂閱率只有 8%？

And could they be higher?

還能更高嗎？

After all, while the company may have 12 times more downloads than Babbel, analysts say its revenue is only two to three times that of its rival.

畢竟，雖然該公司的下載量可能是 Babbel 的 12 倍，但分析師稱其收入僅為對手的兩到三倍。

We could tomorrow massively increase the fraction of subscribers by doing certain things.

明天，我們可以通過某些措施來大幅增加用戶數量。

For example, right now you get one ad at the end of a lesson.

例如，現在一堂課的最後會有一個廣告。

We could play five ads at the end of a lesson.

一節課結束時，我們可以播放五個廣告。

That would get a lot more people to subscribe.

這樣會有更多的人訂閱。

We could maybe even force people to subscribe.

我們甚至可以強迫人們訂閱。

And it is true, we would have an amazing next quarter.

的確，我們下一季度的表現會非常出色。

However, that would also stifle our growth.

然而，這也會扼殺我們的發展。

But analysts say that while Duolingo's leading market position may have been achieved by giving away much of its lessons for free, that growth strategy may start to slow, especially as some users leave the app after mastering a language or giving up on it.

但分析人士認為，雖然 Duolingo 通過免費提供大量課程取得了市場領先地位，但這種增長策略可能會開始放緩，尤其是一些用戶在掌握一門語言後離開或放棄了該應用。

And that may lead to Duolingo trying to increase the number of paying customers.

這可能會導致 Duolingo 試圖增加付費用戶的數量。

So the company is exploring some new ways to get users to part with their cash.

是以，該公司正在探索一些新方法，讓用戶拿出現金。

Its new AI-powered Duolingo Max tier costs around $30 a month and allows players to roleplay with AI-powered bots, something the company had previously tried to offer with humans.

該公司新推出的由人工智能驅動的 Duolingo Max 級每月收費約為 30 美元，允許玩家與人工智能驅動的機器人進行角色扮演，而這正是該公司之前嘗試與人類合作提供的服務。

We would say, okay, so you're telling me that if you could on Duolingo, press a button and you could have another human on the other side, be able to practice, you would use it?

我們會說，好吧，那麼你是說，如果你能在 Duolingo 上按下一個按鈕，就能讓另一個人在另一側進行練習，你會使用它嗎？

And I say, sure, I would use it.

我說，當然，我會用它。

We even asked, would you pay for it?

我們甚至問，你願意為此付錢嗎？

Like, oh, sure, I would pay for it.

比如，哦，當然，我會付錢的。

And then came the very revealing question.

然後是一個非常有啟發性的問題。

We'd say, okay, do you want to do it right now?

我們會說，好吧，你想現在就做嗎？

A hundred percent of people would say, no, no, no, not right now.

百分之百的人會說：不，不，不，現在不行。

What I love about AI is that people don't feel like the AI is judging them.

我喜歡人工智能的一點是，人們不會覺得人工智能在評判他們。

Even though it actually is judging them, but who cares, right?

儘管它實際上是在審判他們，但誰在乎呢，對嗎？

It's an AI, it's not another human.

它是人工智能，不是另一個人類。

But for now, at least, Duolingo says it will continue giving away most of its product for free.

但至少現在，Duolingo 表示將繼續免費提供大部分產品。