In-depth Analysis of Kaito: How the Yap Event Ignited the Social Flywheel?

Compilation: Wu Talks Blockchain

This issue features content from Alex's personal YouTube channel, focusing on the recently popular social product Kaito, delving into its product strategy, market background, and development logic. Alexon is the CIO of Ferryboat Research. By analyzing Kaito's choices on the Twitter platform and its characteristics in collecting, processing, and applying crypto social data, he explains the reasons for its high pricing and core advantages. Additionally, he compares the directional explorations of similar projects, pointing out how Kaito breaks through the limitations of traditional data services through API optimization, KOL mapping, and social binding mechanisms, successfully completing a strategic transformation and establishing a unique market position. He also shares entrepreneurial experiences and insights from industry practitioners, directly addressing the challenges and opportunities faced in the productization and commercialization of Web3.

Crypto Traffic Acquisition: Differences Between Advertising and Viral Models

Crypto is a high-volatility, high-risk field with strong financial attributes. You may find opportunities within it, but you should also be mentally prepared for the possibility of losing your principal entirely. Next, let's discuss the first part: why Kaito and similar products choose Twitter as their main battlefield.

First, from the perspective of the consumer goods industry, traffic structure is generally divided into two categories: public domain traffic and private domain traffic. In terms of traffic acquisition methods, there are two main paths: advertising and viral growth. Public domain traffic typically includes Twitter and YouTube, while in the crypto industry, Telegram and Discord belong to private domain traffic. In contrast, private domain traffic is harder to track and has a more singular structure.

Although platforms like Reddit, Instagram, and TikTok are gradually entering the crypto industry, Twitter and YouTube still have the highest traffic concentration. In a domestic context, this might require leveraging platforms like Xiaohongshu, Douyin, and Kuaishou for promotion, alongside platforms like Bilibili for recommendations, and finally using direct promotion methods like CPC or display ads. Afterward, traffic is directed to WeChat and other private domains for conversion and repurchase.

Overall, the traffic acquisition methods in the crypto industry are relatively simple because the advertising logic at the current stage of the industry cannot support sufficient effectiveness. This leads to a relatively singular approach to traffic acquisition, mainly focused on viral growth and distribution.

Comparison of User Acquisition Costs and Viral Effects in Different Regions

More than two years ago, when we were developing our own tool product, we experimented with advertising strategies. I invested tens of thousands of dollars for testing, and while I can't disclose specific data, a clear result was that the cost of acquiring a user in the U.S. was about ten times that of acquiring a user in Vietnam. However, the viral growth rate of Vietnamese users was significantly higher than that of American users. This indicates that American users are less inclined to actively participate in viral promotion, such as creating and sharing a landing page.

In the entire crypto industry, I believe there are fundamentally only two ways to acquire traffic: distribution and viral growth. Although both methods essentially belong to a form of viral growth, their application logic differs. Distribution tends to rely on KOLs (Key Opinion Leaders) or KOCs (Key Opinion Consumers) for promotion; you entrust the product to them for endorsement, and they distribute it to retail users.

Viral growth, on the other hand, involves designing an efficient viral mechanism to create activities that attract users to participate actively. For example, Kaito's Yap activity is a typical case. Users share data from their Crypto Twitter (CT) accounts, such as displaying how many "smart followers" they have, creating a gameplay similar to NetEase Cloud's annual playlist or consumption bill. Essentially, these mechanisms aim to achieve viral growth through voluntary sharing by users, thereby gaining more traffic.

With this background knowledge explained, it becomes clear why we initially chose Twitter as the primary platform rather than a private domain. The biggest issue with private domains is that it is difficult to standardize the acquisition of all content, and the content within private domains is hard to evaluate effectively. For instance, if a community is entirely focused on discussions about Kaito, you cannot accurately assess the true value and influence of this data. Additionally, the decentralization of private domain platforms makes it very challenging to comprehensively gather relevant data. For this reason, it is not a priority choice.

Why Kaito Chooses Twitter as Its Main Platform

On public domain platforms like YouTube, content is typically suited for long video formats. For example, it can be a monologue like the one I'm recording now, an interview format, or content that focuses more on tutorials and interactions, even some mining operation guides. Such content often requires long production and viewing times, suitable for topics that need detailed explanations and learning. Therefore, this type of content medium is inherently unsuitable for scenarios driven by immediacy or hot topics.

These long video contents are generally more suitable for handling PoW (Proof of Work) related themes. So even though we also tried to introduce Kaito's monitoring and analysis logic on YouTube and Farcaster, we ultimately found that the targets that could be effectively observed were usually projects like Kaspa and Helium, while performance for certain short-term meme tokens was completely lacking.

In contrast, Twitter is inherently suitable as a data platform, especially in an environment where social data concentration is very high. Almost all marketing budgets are concentrated on Twitter, forming a high level of consensus. At the same time, Twitter's social graph is also very transparent; for example, your following list, engagement counts, and other data are presented in an explicit form. On platforms like YouTube, it is difficult to obtain clear fan relationships or interaction details.

Ultimately, the reason for choosing Twitter as the main platform is that it is the optimal solution. Its transparent social graph and centralized traffic structure provide us with clear advantages. In contrast, platforms like YouTube make it very difficult, if not impossible, to obtain similar relationship data. Therefore, both we and Kaito are more inclined to prioritize Twitter as our main battlefield.

Two Main Reasons for Kaito's High Pricing: API Costs and Regulatory Constraints

At that time, we used some "tricks"; Twitter had not yet been acquired by Musk, and there were some gray areas in the system. For example, using educational accounts or other means to obtain data, although not entirely compliant, was common in the early stages. For early projects like Kaito, I suspect they initially adopted similar strategies to obtain data through these informal channels. However, once the product began to commercialize, this approach was clearly no longer viable.

Two years ago, when they completed their financing and launched the product, they could only rely on commercial APIs, and after Musk acquired Twitter, many non-compliant channels were blocked. The cost of using commercial APIs is quite high, and as the number of calls increases, this cost grows linearly rather than decreasing.

The second reason for the high pricing is Twitter's regulatory constraints. Even if a company uses a commercial API, there is a limit on the number of calls per month (I can't recall the specific number). This means that if the product becomes particularly popular, the limitations on call volume will make the ToC (business-to-consumer) model unsustainable. Ultimately, both we and Kaito chose the ToB (business-to-business) model at similar points in time, which is the best solution to maximize the economic value of limited call volumes. For Kaito, this was almost the only direction available.

Specifically, due to the fixed call volume, the only way to achieve greater economic returns is to increase the value per user, which simply means raising prices. This is precisely a necessary choice for the product; otherwise, the entire business model cannot be established.

I learned that their delay is about 15 minutes, similar to ours. It is important to understand that the shorter the delay time, the higher the cost required. This is because historical data needs to be scanned at a higher frequency, and this cost increases exponentially. The setting of delay time also directly affects the efficiency and economic feasibility of API calls. In summary, Kaito's high pricing under API calling costs and regulatory constraints is reasonable.

Evolution and Choices of Kaito's Product Direction

Next, let's talk about Kaito's product direction and why they have evolved from "trending" type products to the current KOL type features. Here, I will first provide a small conclusion—it's not about teaching others how to start a business, but sharing our own experiences. We have tried multiple directions and found three paths that can be derived based on this logic.

The first direction is a purely self-use Alpha tool. Kaito's CEO mentioned in a podcast that they had also considered this direction. If the tool is only for Alpha-type purposes, the more it develops, the more it tends to be for internal use rather than suitable for large-scale users. We have encountered similar issues—if it is free, users may not value it; if it is charged, why not just use it ourselves? Such issues make Alpha tools generally more suitable for self-use rather than productization.

We once developed a set of tools using a logic similar to Kaito's. This tool application allowed us to often discover projects before they became popular. We considered using this logic to create a listing tool for exchanges. For example, I once wanted to collaborate with Binance to provide this tool for free to optimize their listing selection criteria. Because certain projects, like ACT, did not show any noteworthy performance in our "God's eye view" based on Twitter data analysis, yet were still listed on exchanges. Such unreasonable choices could have been avoided with a data-driven tool.

Additionally, we also explored applying Alpha logic to quantitative trading strategies. We tested the top 200 or top 100 projects on Badcase, making trading decisions based on text mining, sentiment analysis, etc. The test results showed that this strategy was significantly more effective for smaller market cap projects that are easily influenced by sentiment and events, while its effectiveness for larger market cap projects was limited. I believe Kaito has also conducted similar research, as their CEO has a trading background. From this perspective, we and Kaito share many similarities in our early starting points and logic, but the paths we ultimately chose differ.

Kaito's Exploration of Community News Tools and Its Industry Potential

Under the current model framework, some phenomenal themes, such as memes and NFTs, are very prominent. They show potential for price increases within this logic. However, such phenomena cannot be completely resolved through standardized programmatic trading, as they still require strong human intervention. This characteristic makes them effective but lacking standardization. As for whether Kaito has similar products internally for its own use, I am not sure.

Another direction worth exploring is news and GPT-type products. What does this mean? For example, current Web3 assistants like Alva (formerly Galxe) can obtain all tweet corpus by integrating Twitter's time-series data and process it using the ChatGPT interface. By adjusting prompts on the front end, these data can be output in a more intuitive form, generating many real-time community news.

For a simple example: if you are confused about the "elisa" case, you can directly ask this tool: "What is the reason for the uppercase and lowercase 'elisa' dispute? Who initiated it?" In this way, the tool will summarize the answer based on the latest data. The original GPT cannot do this because its data has a fixed cutoff date and usually cannot provide content from the last six months. You can only crawl the relevant corpus yourself and feed it to GPT, then summarize the logic through prompts. The potential of such tools is enormous and is a direction worth exploring in depth.

Currently, it seems that Kaito is already exploring such products or trying similar directions. The Alva product I mentioned is a good example. It integrates a large amount of industry data by calling APIs related to the crypto field, connecting users with industry information point-to-point. However, Alva has the issue of insufficient data cleaning quality. They spent a lot of time connecting data networks, but there is still room for improvement in data accuracy and the level of detail in cleaning. In contrast, Kaito's advantage lies in its data accuracy, which is beyond doubt.

For instance, regarding the recent "elisa" case, I quickly obtained answers through such tools. The application of such products in the crypto industry can indeed significantly improve efficiency. More than two years ago, we also developed similar tools, and the test results showed that they could indeed enhance work efficiency. However, when we attempted to commercialize, the core issue we encountered was the lack of strong willingness to pay from users. Although the tool could improve efficiency, it did not address a core pain point, which made users lack a strong purchasing motivation.

Moreover, due to the high calling costs of such tools (each call to the GPT interface incurs a fee), the product's gross margin is relatively low. Therefore, although these tools have certain significance, their commercialization faces considerable challenges. Many calling behaviors are more for activation purposes, and there are limited scenarios that actually generate revenue, which all become challenges that need to be overcome. In summary, while this direction has great potential, more optimization and breakthroughs are still needed in practical implementation.

The Role of Data Accuracy and KOL Mapping in Marketing

When discussing these tools, a core question arises: how do they achieve revenue? If relying solely on a VIP model that allows users to call the API unlimited times, such products are unlikely to have significant profit margins, but their existence is meaningful. They can directly utilize Kaito's logic to read Twitter data for generating and distributing self-media content, such as "Wu Talks" or other forms of community news. These tools not only enhance efficiency but also help project parties distribute content across multiple platforms, such as generating short videos with AI for TikTok or posting directly on Twitter.

I believe this product direction is not something only Kaito or Galxe can attempt; projects like Mask are also very suitable for this. Strangely, Mask currently seems not to have deeply ventured into this direction. If any team members from Mask hear these suggestions, I hope you can consider trying it out.

For Kaito, its current product direction indicates that they wish to move towards a larger market value rather than continue along the Alpha tool route. While Alpha tools can be profitable, they lack productization potential. If focused solely on this, it will ultimately be limited to internal use and cannot form a product aimed at a larger market. By shifting towards KOL mapping, Kaito is clearly aiming to break through this bottleneck.

The early users interested in Kaito's products were almost identical to the user group that was paying attention to our tools at that time. Our tools were also suggested to be sold to some trading companies or secondary funds in the early stages. Although these trading companies were more concerned with profitability, this direction would fall into a cycle of "whether to be profitable." In contrast, KOL mapping provides precise support for marketing placements, enhancing the effectiveness of placements through data accuracy, thereby increasing the marketing value for project parties.

Data accuracy is key. While many companies can collect Twitter data, whether the data is accurate is another matter. In the open market, Kaito and our early tools are among the few that can achieve accuracy. The core of data accuracy lies in "data cleaning," which is the most difficult and critical step. Collecting data is relatively simple, but weighting and cleaning the data requires a lot of repeated testing and logical adjustments, which often need a combination of experience and intuition.

For example, the Chinese community's Crypto Twitter (CT) often has a lot of noise, and its weight needs to be reduced. This noise causes Chinese CT to typically lag behind English CT by 24 to 48 hours. How to effectively clean and adjust the data is a "core competency" and also the company's core competitive advantage.

Through precise KOL mapping, Kaito can help project parties optimize their placement strategies and improve placement accuracy. This product can not only assist project parties in achieving more efficient marketing but also generate marketing fees from it, forming a sustainable business model. Choosing this direction is a smart strategy that Kaito has demonstrated in market competition.

The Strategic Logic and Flywheel Effect Behind the Yap Activity

In the entire Crypto field, advertising has always been a relatively vague and inefficient behavior. Current marketing agencies essentially function more like simple tools for maintaining contact lists, with relatively singular means. In this context, the tools provided by Kaito can help project parties determine which KOLs are worth advertising and which are not, providing evidence-based references through data analysis. This precision greatly enhances the efficiency of advertising.

Kaito optimizes KOL placements through two key metrics: correctness and core circle. Correctness refers to whether the KOL's judgment is accurate, such as whether they discussed a project before its price increase rather than participating after the price has risen. Each time a KOL shares or promotes, their judgment's correctness is recorded and weighted, affecting their weight score. All of this can be repeatedly verified through timestamps and data analysis tools.

The core circle (referred to as "smart followers" in Kaito) measures the depth of a KOL's influence. If an account has more smart accounts interacting with it, its weight score will be higher. This helps project parties filter out truly influential KOLs rather than just accounts with a large number of followers.

Kaito's Yap activity showcases its successful strategic transformation. This activity significantly reduced marketing costs by leveraging free KOLs. Traditional marketing requires contacting KOLs one by one and paying high fees, while Kaito directly opened a page that allocates rewards to KOLs through a weight algorithm. This method simplifies the process and enhances credibility through data transparency. This model encourages many KOLs to voluntarily participate in promotion, helping projects spread rapidly.

At the same time, the Yap activity also addresses potential risk issues. Considering that if Twitter changes its API rules in the future, Kaito allows all CT users to bind their accounts to its backend through TGE, actively authorizing data usage. This approach enables Kaito to gradually detach from its reliance on Twitter API and begin to master its own data assets. This not only gives Kaito greater independence but also creates a positive cycle between supply and demand: as more CT users bind their accounts, project parties' interest increases, forming a flywheel effect of data matching.

Ultimately, Kaito has created a successful marketing ecosystem platform in the crypto industry through this model, akin to Alibaba's Alimama or ByteDance's 巨量引擎. Currently, this strategy has been executed quite successfully.

Entrepreneurial Reflection: How Non-Typical Elite Background Practitioners Break Through

If all CT (Crypto Twitter) users bind their accounts to Kaito's backend, then in the future, when entering the secondary market, Kaito can clearly tell the outside world: "This data is mine." Whether it is project parties or CT users, this binding behavior can form data consensus and trends. This is the core logic behind the Yap activity.

Before concluding the topic of Kaito, I want to share a small story about ourselves. Before Kaito's financing, we also developed similar products, and it could even be said that we were doing it concurrently. More than two years ago, we simultaneously tried directions for Alpha tools and GPT-type tools. At that time, during the industry's low point, our team was not very good at socializing, and we knew very few people in the industry. Although our products were interesting and had potential, there were very few friends to introduce us to VCs.

At that time, we approached four VCs, one of whom was willing to co-invest but needed us to find a lead investor. The other three directly ignored us, one reason being that our background did not fit the typical elite entrepreneur image. They did not delve into the logic behind our products, nor did they try to imagine its potential value; they simply voted us down.

It wasn't until later that we gradually gained attention from more industry professionals through platforms like YouTube. Most of these viewers were institutions and practitioners in the industry. Even so, I still did not mention the past to those VCs who had previously contacted us, as it felt somewhat awkward. Interestingly, I later saw employees from those VCs praising Kaito, which left me quite emotional.

We ultimately chose to pursue the Alpha tool route, a decision related to our limited social circle at the time. We believed that without external help, it would be difficult to successfully commercialize a ToB product. We hoped to gain market expansion through recognition from well-known VCs, rather than relying solely on our own difficult journey.

For those entrepreneurs with non-typical elite backgrounds, I have some advice. VCs are more focused on connections and relationship networks rather than the product itself. However, I firmly believe that a good product can speak for itself. If your product is truly good, do not be afraid to showcase it. Nowadays, I also realize the importance of building social influence. Through social networks, you can not only meet more people but also accumulate a certain level of recognition and trust for future entrepreneurship.

For friends who watch my videos or browse my Twitter, I hope to convey the belief that regardless of whether you have an elite background, as long as your product is excellent enough, I am willing to help you. Good products and ideas are more important than glamorous resumes. As long as what you present can earn my recognition, I will do my best to help you find resources.