From Manus to Button Space to Lovart, every time a new product is launched, the circle of friends is always full of calls for “code request”. What kind of logic is hidden behind this phenomenon? This article will delve into the underlying theoretical basis of the invitation system from the perspective of product managers, the six degrees of network theory, analyze its advantages and limitations in the cold start stage of products, and combine classic cases such as Gmail and Zhihu to explore the applicable boundaries of the invitation system.
This year, there have been many AI products that have attracted attention, especially the few that we are more impressed with, almost all of which have adopted the invitation-based operation method.
From the earliest explosion of Manus, to the later button space, to Lovart half a month ago, every time a new product is launched, you can always see a call for “code asking” and “asking for code” in the circle of friends, which is quite a sense of déjà vu in the pirated film and television forum of “the owner is a good person, asking for seeds”.
In the issue of introducing the button space, I sent more than 4,000 invitation codes to the students who got the AI learning circle; When I arrived at Lovart, I sent almost fifty or sixty that time. Every time in the process of giving code to everyone, my classmates always ask me the same question: Why are today’s products using this kind of gameplay?
I believe that we will encounter the invitation code mechanism frequently in the future. Therefore, today I will briefly talk to you about the logic behind the invitation code gameplay from the perspective of a product manager, including its advantages and limitations, hoping to help you understand its role in the cold start stage of the product more clearly.
01 The underlying logic of cold start of invitation codes: the theory of six degrees of connections
Invitation codes are not new to the present.
As early as the 90s of the last century, someone already added the mechanism of “invite friends” to the product. The earliest practitioner I could find was Andrew Weinreich, who was working on a product called SixDegrees.
B-end product manager’s ability model and learning improvement
The first challenge faced by B-end product managers is how to correctly analyze and diagnose business problems. This is also the most difficult part, product design knowledge is basically not helpful for this part of the work, if you want to do a good job in business analysis and diagnosis, you must have a solid …
View details >
The two names may be a bit unfamiliar to classmates, but Weinreich has since become known as the “father of social networks” because he launched the network-based product in 1997 and was acquired in 1999 for seven figures — a full seven years before Facebook went live. And the name and concept of this product come from the famous “six-degree separation theory”.
The more common name in our country is “six degrees of connection”. In simple terms, this theory believes that any two people only need to be referred by up to six people to establish contact. To use a mathematical model, if you know 30 people, each person knows 30 more, and pushes it layer by layer 6, you can theoretically reach more than 700 million people.
This is actually the lowest theoretical basis behind all invitation code mechanisms: relying on the connection between people and using the smallest seed user group to drive exponential network proliferation.
The earliest prototype of this “six-degree separation theory” actually originated from a brain hole in a literary work. In 1929, Hungarian writer Frigius Kalinti first proposed a conjecture in the short story “Chains”: it only takes five people to contact anyone on Earth. The characters in the novel play it as a game, envisioning that there are only five “friends of friends” between any two people.
This idea later aroused the interest of many researchers, and many verification attempts were made. In 1969, psychologist Milgram designed the “letter sending experiment”, in which strangers tried to send letters to specific targets through acquaintances, so as to observe the length of the actual connection path; In 2003, Columbia University conducted a similar experiment via e-mail, further verifying the widespread nature of this network connection.
Although it sounds a bit like a “classical web Internet”, the influence of this theory has not diminished to this day. Many of the seemingly modern features can now be traced back to this set of ideas, such as the common “people you may know” in short video platforms, which is essentially applying this logic: speculating on a wider range of social connection possibilities through the relationship graph between you and your friends.
02 Two classic cases of invitation code mechanism
Let me introduce you to a classic case of invitation code mechanism – Google’s Gmail mailbox.
When Gmail was launched in 2004, it was released on an invitation system. Although we are now accustomed to mailboxes and network disks, we have dozens of GB, or even TB of capacity at every turn, and it is not a problem to store some high-definition movies. But back then, the vast majority of mailboxes had only a pitiful few MB of free space, while Gmail provided a full 1GB of free space as soon as it came out, which was simply a “shocking” treatment at the time.
If you are not in the IT industry, you may not have an intuitive feeling of units such as MB and TB. We can use a slightly more vivid analogy: the difference between MB and TB is probably like the difference between “grams” and “tons”.
For example, a hamster weighs about 30 to 100 grams, while an elephant can reach 5 tons. This is the difference between MB and TB. Of course, the space provided by Gmail at that time was almost equivalent to the weight of a kitten about half a year old, which was also a behemoth compared to Microsoft’s Hotmail mailboxes at that time.
What’s even more interesting is that Gmail actually provided free space at the time, but Google still used an invitation code mechanism to limit the number of registered users, and the starting point was very simple – their infrastructure simply couldn’t handle the massive influx of users.
Ten years later, there was an article called “The Story of the Birth of Google Email”, which mentioned a very classic detail: “The underlying equipment of Google Email is 300 Pentium 3 series computers that almost no one uses anymore in the company, which is enough for the company’s planned limited beta promotion.” They only issued accounts to 1,000 non-Google employees, allowing them to invite a few friends each, and then gradually grew from this small group. To use a bad joke often used in the short video market last year, it is “the gears of fate begin to turn”.
In fact, being able to get this invitation is not only because the free space has become larger, but also because it means identity. This identity is scarce, so the person who gets the invitation will definitely give praise, criticism, or other types of evaluation of this closed product. But no matter which one it is, there is a faint message behind it: “I am different from you, I am invited to use this product.” ”
And those who are not invited will definitely go everywhere to find inviters. The more difficult it is to obtain the invitation code, the more actively it will be promoted and discussed, for fear that others will not know that they have obtained the invitation code. This brings unexpected attention to a new product.
The person in charge of Google Mail at the time said: “Our limited supply release method had to be adopted because of hardware limitations, but I didn’t expect that the side effect it caused was that everyone was more eager to get invited. This practice is seen as one of the most successful marketing strategies in the history of technology, but in fact, it is just a coincidence. ”
How popular was Gmail at that time? On some e-commerce auction sites, its invitation code was once speculated to $150 or even higher. This situation is almost the same as this year’s Manus invitation code is in short supply on Xianyu. Until 2007, Gmail still had to sign up by invitation, and few products had insisted on the invitation system for so long.
Writing here, as an Internet antique, I specially looked through my first email in Gmail and found that the time was August 2005, after all, I was invited in, and I dared to take it out to talk to you about this topic.
However, compared with Gmail, another more classic invitation code case is Zhihu. When Zhihu was first launched at the end of 2010, it also adopted a closed invitation system, and only after getting the invitation code could you register, and even tourists were not open to browsing. The earliest batch of invitation codes were strictly controlled by Zhihu officials, mainly concentrated in the Internet circle, so many of the first people who came in were investors, technology reporters, and product managers.
Although the number of early users is small, it covers many industries. In the first 40 days of launch, although there were only a few hundred users, more than 8,000 good questions were asked and tens of thousands of answers were answered. The emergence of high-quality content has further stimulated the enthusiasm of insiders to get invitation codes.
Moreover, the entire invitation code issuance process was also very manual at that time, all by hand. Old users need to take the initiative to go to Zhihu’s operation staff and say: “I have a friend, what kind of work does he do, what kind of experience he has in the industry, can you give him an invitation code.” “Every user is basically endorsed and screened by old users, and the overall quality is guaranteed.
I remember that at that time, a large number of professionals gathered on the platform, such as Kai-Fu Lee, Wang Xing, Wang Xiaochuan, Xu Xiaoping, and others were answering questions in their fields. There are even some heavyweights, such as Lei Jun and Ma Huateng, who have personally come up to answer users’ questions. Therefore, it was by this invitation system that Zhihu established its own product image.
03 Applicable boundaries of invitation codes
In the AI era, the strategy of “inviting waitlists” or “limited opening” has become more common.
In fact, the one who plays this trick the most is OpenAI, the global leader in AI. They don’t use invitation for the entire product, but whenever they launch a slightly larger feature, they will first invite a small area, and then gradually expand the opening, and finally even unregistered users can use it.
For example, their AI-generated video tool Sora has only invited a subset of video experts and directors to join the first batch of users since its announcement at the beginning of last year. However, due to the fact that this invitation qualification is set too rarely, a number of Chinese AI video software such as Keling, PixVerse to Vidu took the lead in rushing to the global front line while they were still in internal testing.
Of course, for OpenAI, the use of an invitation system can avoid excessive server pressure from getting out of control, and on the other hand, it can also allow the market to continue to pay attention to it, and wait for the product to be iterated and improved before launching.
Later, many AI products such as Midjourney also chose a similar invitation method to launch, including many AI tools in our country have basically adopted this path.
However, among this batch of products, there is a special outlier, which is the AI search tool Perplexity. When it was launched, it was completely open for registration without any invitation system, and as a result, the number of monthly active users exceeded 2 million within four months.
This also leads to a question worth exploring – what are the pros and cons of using invitation codes, and what kind of product form and team is it suitable for?
Let’s talk about a few advantages of invitation codes first.
The first advantage is faster cold start, which is convenient for acquiring seed users. Because it mainly relies on existing users to invite acquaintances to join, it comes with a screening mechanism that can quickly gather a precise user base in the early stage. After all, old users know best which friends around them are suitable for using this product.
The second advantage is that it is easier to control user quality and create a community atmosphere. For example, Lovart uses an invitation code mechanism, at least the first batch of users who come in are people who are interested in AI design, so that the tonality of early users is more consistent, and it is more suitable for verifying the core scenario of the product.
The third advantage is that it can create a sense of scarcity. The invitation code naturally has the communication attribute of “things are rare and precious”. People like me, if they find a new product that everyone is using and invitation codes are flying all over the sky, but I don’t get it, then I feel a little anxious. On the one hand, it will question whether it is outdated in the circle, and on the other hand, it will arouse a very strong curiosity.
In this case, once I get the invitation code, I will definitely share my closed beta experience as soon as possible. In fact, in the final analysis, this kind of sharing is mainly to let others know: “Look, I also got the invitation code.” This mentality has long been written in Eason Chan’s song “Red Rose” – “What you can’t get is always in turmoil”.
The fourth advantage is that the invitation-based gameplay has its own communication power, which will naturally bring user self-propagation and network effects to a product. If each old user can bring more than 1 new user on average, then the user growth will increase geometrically. And because it is recommended by acquaintances, new users will trust the product more and are more willing to try it. This is higher than the retention rate and activity of cold-start users brought about by advertising.
The last advantage is that the invitation system can help technical teams better iterate products. Because only a small group of users are testing in the early stage, the data generated is more centralized and controllable, which can effectively guide subsequent product optimization. This is equivalent to establishing a “risk buffer zone” first, first running on a small scale, and then gradually accelerating expansion. You know, over the past few years, I have seen too many products crash on the first day of launch due to the influx of users, and the scene is very tragic.
Of course, having said so many advantages, it does not mean that the “invitation system” is suitable for all new products.
First, the biggest disadvantage of the invitation system is that it limits the rapid growth of user scale and is likely to miss the best opportunity for product promotion. Although this method has a sense of scarcity and seems to enhance brand tone, it essentially artificially compresses the speed of user expansion.
For products that need to quickly race and seize the market, closing for too long may give up potential users to competitors. Some students may say: Our products have such and such core functions, but other products do not have them at all. But the problem is that if users don’t even know what your product looks like because they are blocked from the door, then no matter how good your function is, it will be meaningless to him. And once the user gets used to the competing product, you have to come up with a “several times better” experience or a real exclusive advantage to attract him again.
Second, the invitation system can easily cause user frustration. Especially when the invitation code is overhyped and ordinary users are unable to obtain access qualifications for a long time, they may turn to negative emotions about the product, simply give up attention, or even become a “contrarian” person.
After all, for many users, downloading a product is “giving you face”, but you still engage in an invitation system and don’t let me use it? Not to mention that when they see that the person they “don’t like” in the circle of friends actually got the invitation code first, and they are still showing off their experiences, they will naturally feel even more unhappy. At this time, the reputation of the product is easily damaged.
A very typical example is Manus. At the invitation stage, they were crazy for a while, but when they gradually let go, all kinds of strange words and black drafts appeared in public opinion. Therefore, for the invitation system, it has become extremely important to grasp the heat.
Third, it is best not to engage in invitation system for entrepreneurial products that really start from scratch. Nowadays, many new products cannot attract new users even if they spend money to launch, so why artificially add a threshold? Many people have seen that I have given so many examples of success by invitation codes, don’t get me wrong, success is not “invitation system” at all, but these products themselves are good enough, and the invitation system is just the icing on the cake.
For a start-up entrepreneurial team, it should be bolder. Even if the online server is crowded, it is a “sweet trouble”; What I am most afraid of is that I am full of confidence and have conducted a closed test, sending 500 invitation codes to 100 users, but there are only five daily active users on the second day of launch, and they are all your own people, which is too much of a blow.
Another point is that the invitation code is actually not “light”.
The whole process not only takes time to develop and puts forward new requirements for products, technology and operations. For example, technically, to develop an invitation code generation, distribution, and verification system, it is necessary to manage lists, permissions, usage times, etc., which is actually not low. Operation students also have to invest energy in multi-party communication, monitor the distribution situation, deal with various public opinions, and maintain various big Vs, for fear that they will not say good things after getting the invitation code, and prevent the invitation code from being resold. These things are all asking for trouble for a small team.
Don’t think that this function is small, there are a lot of details in it. For example, this year’s button space and Lovart’s invitation code issuance method is actually manually issued by operating students, and it is already 2025, and the fully automatic AI intelligent products actually rely on the operation to issue codes one by one, which itself has a very magical color. Behind this, it can only be said that the product technology students who do the invitation code system do not consider the difficulties of the operation students.
Regarding invitation code marketing, I will sort it out here first. When you see this kind of operation in the future, you can say with a knowing smile: “I understand this method.” ”
But in the end, I still have to say one thing: invitation code marketing is actually a typical “icing on the cake” method. The premise is that you must first have “brocade” – that is, a product that is enough to impress people. Otherwise, no matter how clever this method is, it will only become a self-pleasure.
