Duolingo, a super education app with over 100 million monthly active users, has become a leader in the field of edtech with its unique learning features, gamified design, and AI innovation. It not only exceeds the sum of New Oriental and Good Future, but also wins the love of users around the world with its seemingly simple but hidden product design. This article will delve into the design details of Duolingo, from user onboarding and gamified incentives to AI applications, to its unique membership design, revealing the product philosophy and success secrets behind it
Duolingo is an unprecedented super education app with more than 100 million monthly active users, and its market value exceeds the sum of New Oriental + Good Future. Behind its seemingly simple product design lies a lot of in-depth thinking, counter-common sense and its exaggerated educational product concept.
I didn’t use Duolingo much before, and I thought it was a bit pre-emptive, thinking that its design was ordinary and simple, and the gamification was PBL (points, medals, leaderboards) was nothing special. Because of the past Ego, I now regret not understanding the design of this product earlier.
With the accumulation of experience in the growth of the education industry, looking back at Duolingo, we really understand the secret. Its seemingly simple PBL is actually a simple and truly awesome trick.Often like One Punch Man, there are no too many tricks, but the ultimate in basic skills。
This article will start with the detailed analysis of Duo’s learning functions, gamification design, and AI innovation, and summarize the firstness and philosophy of its products.
Part 1 Link and function design details
Who doesn’t like to be praised? Duoer is like a eunuch coaxing the emperor, coaxing users with reason and evidence, and learning is naturally happy and motivated.
1) More is Less: More interactions + more compliments + more faces = less burden
Duolingo is different from the landing page. Almost all domestic education products allow users to log in and register before experiencing product functions. However, it first let users go through a newcomer process that accumulates nearly 20 pages (with a lot of redundant design), which will undoubtedly reduce the “newcomer login rate”. But the benefits are also obvious: the number of users who experience the core value of the product through the newcomer process has also increased.
This practice is rare in domestic products, possibly due to word-of-mouth and design difficulties. It is necessary to do a guide that does not affect the login rate, which will greatly test the details of product design and the reputation before downloading, otherwise the login rate may be very serious.
In addition, when we do product design, we usually think about how to reduce pages to shorten the funnel and increase conversions, but this often brings a greater psychological burden to users. Duolingo may make conversions more efficient by designing seemingly redundant pages with more incentives and simplified operations. (Every step of the decision is easy, interspersed with encouragement).
After 10 years of interaction design, why did I transfer to product manager?
After the real job transfer, I found that many jobs were still beyond my imagination. The work of a product manager is indeed more complicated. Theoretically, the work of a product manager includes all aspects of the product, from market research, user research, data analysis…
View details >
For example, newcomer guidance requires users to provide three information: English proficiency, learning objectives, and learning duration. We usually design three interactive pages,Duolingo, on the other hand, is designed with more than twice the number of pages.(1) Tips before selecting information – > (2) Information selection – > (3) Praise after selection.
2) Application of commitment consistency
Commitment consistency refers to the behavior or attitude when a person tends to maintain a commitment after making a commitment. Duolingo takes advantage of this frequently in its interactions. For example, in the button copy of setting the duration and winning streak prompts, the cognitive reinforcement of “I am great, I can persevere, and I don’t give up” is constantly used.
It is recommended that you can design common “OK, I know, okay” copywriting and buttons in the direction of consistency, which will be of good help to retention.
3) Praise if you can praise: There is always a reason to praise you for speed, quantity, and quality.
The first is that launching the app will encourage the transmission of emotional values, such as: you are here again, come and learn together.
Before practicing, two additional pages will be added to encourage users: “It’s easy to start, it’s hard to persevere, so get more to make you full of motivation to play while learning!” ”
After the exercise: the first is to change the motivation of different tricks. Including passing streaks, gem rewards, achievement unlocks, medal unlocks, etc.Even if the correct answer rate is very poor, it will even be “The answer speed is so fast!” “, “fast progress” as a reason for praise.
Who doesn’t like to be praised? Spoil users like an emperor, and learn to be full of natural emotional value.
4) What conflicts between learning effect and emotional value? Always pay attention to emotions.
What should I do if I can’t read? Correction improvement? No! Duoer tends not to tell you.
In addition to all kinds of compliments and amplification, the methodology here is to minimize negative experiences.
First of all, the Duo unit practice time is extremely short. Generally, it is 1~2 minutes, 10~15 questions each time. Exercises consume energy, and immediately praise the user to recharge after a long time. Moreover, almost in the productThere will be no two identical question types in a row, avoid the boredom caused by repetition.
When we are still weighing whether to score well or evaluate stars, or consider how to do a more accurate sound correction evaluation, Duo’s choice is: what are you struggling with, you can read it is already awesome! And after reading, users will not be allowed to read it again, don’t worry!
In addition, when the user spells the wrong type, Duo will also judge that the user answered correctly, but there is only a typo in the question type. The protection against possible negative experiences for users is really good.
Another counter-common sense, “It is not supported to quickly enter the next level of study, but it is better to complete the course and retain?” Most learning products usually guide users to continue learning on the end page after the exercise is completed. It’s like sales constantly urging you to continue to recharge and apply for a card. But this is the vision of the producer, not the user. From the user’s point of view, it should give enough emotional value and motivation, and the user is naturally willing to continue learning.
Another small detail is that in terms of energy consumption and difficulty of operation, “button clicking” is easier to interact with than “swiping screen” (Duolingo rarely uses swipe interaction), which is also an energy-saving detail.
Part 2 Gamification Incentive Mechanism
Always emphasize passing on to users: you are awesome, you are a person who can persist and love to learn. Inactive users cause cognitive dissonance (how can I not insist on being so awesome).
1) Winning streak system
Core logic: “You are a person who can persevere, and persistence with frozen props is also persistence, and the more consecutive wins, the more unbearable it is to interrupt, which is an enhancement loop.”
A winning streak is a core reinforcement loop in Duo game motivation. In the early stage of cognitive development, the product will force users to feel 2 consecutive wins, and if they don’t learn the next day, they will still give away props to ensure that they achieve a winning streak.
Encourage a long winning streak through the winning streak elite club, a sense of accomplishment + exclusive App icons.
In addition, through the friendship streak gameplay, on the one hand, it strengthens the sense of accomplishment of the streak, and at the same time, it also uses the black box emotion “I don’t want to ruin the friend’s streak” to motivate users to continue learning.
Through the gameplay of honor incentives, friend streaks, and prop protection streaks, the closed loop of the user sense of achievement system has been successfully strengthened.
2) Leaderboard mechanism
I heard that leaderboard rules can only motivate top players? But in Duolingo, all users are the top players.
Duo is unlocked in stages through content, and 9 units must be completed to join the leaderboard, avoiding too much content at the beginning to dazzle new users.
The earliest book “Gamified Thinking” mentioned the concept of PBL and clearly pointed out that the disadvantage of leaderboards is that they can only motivate top users.But Duolingo always allows users of the same level to compete in flow mode by splitting + segmenting.The following three methods are mainly adopted
- Method 1: Leaderboard segmentation.Different levels have different levels of rankings, roll with people with the same level of effort;
- Method 2: Minimize the leaderboard.There are only 20 people of similar level in each list, and you can improve your ranking by completing one exercise;
- Method 3: Algorithmic personalization of leaderboards.If you learn less this week, you will be assigned to a weaker ranking in the second week, which is easier to be on the list;
In addition, Duo combines leaderboard + review game. Several very addictive game types (system 1 decision, the sense of impact is very good
Good), there are even many people on Douyin who share videos of themselves matching the list. “Highest combo record! I must break! ”
Finally, the leaderboard is also a social entrance, adding friends to establish associations, which is easier to stick to; And the leaderboard will prioritize similar users in the same leaderboard, such as peopleIn Turkey, users who are learning French will be on a leaderboard。
3) Task system
Tailor tasks that are highly valuable to the product, have a sense of purpose for the user, and are in line with the flow of the user.
Motivate different people’s high-value behaviors in a stepped and differentiated manner through different types of tasks.
- Stepped:The three tasks are progressively difficult, and the first one is always easy to complete. After one ended, Zegnik’s influence took effect again.
- Differentiation:There are dozens of task types, and the task requirements are different at different stages, and the underlying recommendation logic should be the incremental LTV of different people and different behaviors from high to low
In addition, there are some details in the mission:
- The shorter the countdown, the more incentive: the daily task is limited to 12 hours after the active event, not 24 hours.
- Simple rules and many transmission scenarios: tasks and progress will be passed after users complete exercises and reviews.
- Monthly special tasks are used for mid-term motivation, and the difficulty changes flexibly every month according to the learning situation, for example, if I don’t study for a long time, there are 25 monthly tasks, and I learn frequently
There may be as many as 60 of them. Complete to get an exclusive medal, and if you miss it, it’s gone. In general, Duolingo is thereThe ultra-short-term to long-term incentives have been fully covered:During the practice process, the encouragement is ultra-short-term incentives, daily task short-term incentives, leaderboard weekly short-term incentives, monthly task mid-term incentives, achievement system and winning streak long-term incentives
4) Achievement system
I’m awesome, I can be even better!
Personal best record:Hidden achievements, with dimensions such as the number of perfect units and the highest experience in a single day, there is no fixed gradient, constantly sprinting to the highest and recording the date;
- It’s more about recording limit data to make users feel special (rather than relying on time accumulation).
- It can make users recall that one day they were so awesome and motivated.
Awards:Most of the learning behaviors are growth awards, that is, reaching lv1 and continuing to hit lv2. There are two points worth mentioning
- Not everything is encouraged, the better: Among them, 3 social friends and the top of the leaderboard are one-time awards. discourage users from continuing to brush the number of friends for achievements; At the same time, the first place in the ranking is difficult to complete, which may be considered to not put so much pressure on users;
- Special time awards have been added: Early to bed and early to rise, dark night walker. Document or incentivize high-value behaviors.
5) Summary
“The strongest closed loop of gamification is not the closed loop of function or currency circulation, but the closed loop of increasing user sense of achievement.”
Tasks, streaks, achievements, friends, leaderboards and other gamification modules do not have a complicated flow, and Duolingo’s gamification closed loop ultimately falls into the closed loop of user emotions and sense of accomplishment: “The product has a reason to praise – > I’m awesome – > I insist – > you have more reasons to praise – >I’m better – > I have to persist even more.”
Part 3 Applications of Duo AI
The core application of AI in language education is to provide more open, personalized and scenario-based application.
In 23 years, Duo already had a counter-common sense when it first used AI to do scene dialogues:Results are not reported in detail.Many educational products want their results pages to report as granular as possible, but Duo only has one metric: the vocabulary used. But because the vocabulary is recorded, users will naturally say longer sentences in order to use more vocabulary. This is really seeing the North Star in learning motivation.
As a result, in 24 years, Duo did AI oral dialogue (Video Talk with Lily), which was more extreme and different from competing products:Not only was it not corrected during the conversation, but it was not reported afterwards. Duo’s official reply was: “We think feedback is important, but Lily only has one kind of feedback, and that is whether she can understand you.” But my understanding is that Duo does everything possible to ensure the user’s sense of accomplishment, and if there are grammatical errors in the conversation, it is not conducive to obtaining a sense of accomplishment.
In addition to speaking, Duolingo uses AI to create scenario-based practice Adventures.
This is the ideal implementation of AI+ gamification: the ability to generate content in real-time through AI, truly putting language learning into the scene. And after the user answers incorrectly, the user will not directly give wrong feedback, but the NPC will interact with the scene to give normal feedback, such as NPC asking: How old are you? You chose I like to eat watermelon, and the NPC will reply that the watermelon is indeed delicious, but I still want to know how old you are now? Instead of stereotyping, labeling you with errors.
The fly in the ointment is that this function is still a short-term incentive and a small level integrated into learning. If you zoom in this direction and consider longer-term scenario adventures, it is possible to become an app alone, such as “Reborn I Learn English in Two Dimensions”.
In addition, AI applications also include AI explanations, AI question pushing (to ensure that users are in flow mode), etc., which will not be repeated here.
Part 4 Duo’s Membership Design
Through two counter-common sense (seemingly not cost-effective for revenue) designs, users can buy more with confidence and bring more friends.
Duo members have four core value points: no disappointment in wrong questions, no ads, wrong questions replay, and free legend challenges. moreoverDifferent entrances to enter the membership department, the order of rights and interests is different.
In terms of payment strategy, Duo supports an automatic annual subscription model after 0 yuan and 7 days of experience (lower the threshold). According to a friend of mine from Xmind, after xmind subscribed to 0 yuan, the revenue achieved double, which can be promoted as an important AB test in some membership products. In addition, for users who are active but continue to not pay, this will be relaxed to a free trial of 14 days (more confident in the retention of this group of users)
This“Free + Subscription”In this mode, Chinese users are worried about forgetting to turn it off after the free experience. Duo reinforces this by giving clear notice towards the end. This is also very counterintuitive, it seems to be doing something loss-making, but it reduces the user’s concern about “forgetting to unsubscribe”, but stimulates the improvement of this step of conversion.
The most special and innovative point isDuo guides fission and boosts retention with family packages。 The point against common sense is: 1 person 588/year vs 6 people 798/year, which is a lot of loss from a mathematical point of view (if you pull a full person, the per capita is only 11 yuan/month). However, this kind of welfare has led to a large degree of fission of people, and at the same time, the family has its own deep friendship to enter, which is of great help to retention and repurchase.
At the same time, this also adds a reason to buy. You can buy it or not, but if you learn with a few friends, you have more reason to buy it, and you can also give friends and members a favor.
end
In addition to some inspiration in functional design and AI applications, I also learned from it by summarizing Duolingo’s product experience and gamified design appearance
Abstract implicit product philosophy and design main line:
Praise the user – > make the user feel great about themselves – > more willing to learn
Let’s emphasize this important point:“The strongest closed loop of gamification is not the closed loop of function or currency circulation, but the closed loop of increasing user sense of achievement.”
This is a very simple and effective design principle and guide. Based on this point and deduced downwards, Duolingo seems to be counterintuitive, but it is in line with it
The design of logic is all make sense. For example
- Login postponement reduces the login rate but allows more users to experience the product;
- More is less？ More interactions + more incentives = less user burden. 2 steps can be explained, but it has to be designed into 4 steps.
- Focus on encouragement, occasionally forget to study can also win in a row, and clerical errors are also considered correct;
- Star rating or rating? Duo chooses not to comment to reduce entanglement;
- Redefine the leaderboard, everyone can be the head user!
- The free trial is about to expire, and we’ll remind you in advance.
- Family package, the price difference is not too much, you can bring 5 more friends to study!
These uncommon senses have also greatly helped and influenced my product thinking, and also reminded me of two sentences in the Tao Te Ching:
“The clear way is like a fee, and the way of entry is like a retreat”
The right path seems to be an obstacle, and the real way forward seems to make people step back. Share.
