This article is from WeChat official account:GGV Jiyuan Capital (ID: GGVCapital) , author: business services group, oral: Glenn Solomon, Aghi Marietti, editor: Rita, Yvonne, thematic map provided
Public cloud, containerization and microservices (micro service) are gradually replacing the original IT architecture, which makes the API in the company’s information There is an exponential growth in the system. Faced with such growth, companies will need an efficient management tool.
The open source API unicorn Kong is such a tool. Its main product is an API open source gateway, through which users can analyze traffic and manage developers, consumers, partners, and customers related to different APIs.
In our communication with Chinese open source entrepreneurs, we often discuss the following questions: What kind of project should be open source? If commercialization is required, what kind of functions should be open sourced, and what kind of functions should be charged? How does the community need to operate? How to balance the open source product team and the commercialization team? Can application software be successful through open source? As a leader in open source commercialization, how did Kong move from open source to commercialization? In the sales model, what ingenious arrangements have they made? How did they deal with competition and challenges? …
Today, the GGV corporate service series brings you a conversation with Aghi Marietti, the co-founder and CEO of Kong, in 2019.
Glenn Solomon: AgHi, can you briefly talk about yourself and the development history of Kong?
Aghi Marietti: The company we started with was called Mashape, which focused on building the API market. Similar to the GitHub platform that provides cloud services, people publish and purchase APIs here. When 20,000 APIs were gathered on the platform, including niche APIs and 300,000 software developers, we desperately needed a tool to manage this complex and huge system, but there was no such product on the market at the time, and we had to do it ourselves. Make one. We have been working on this platform for about two or three years, and user and community activity has been increasing, but commercialization has not made much progress. In desperation, we decided to open source a series of tools we made to see if there were any commercialization opportunities. Among these tools, only Kong succeeded. Six months later, the Obama Health Insurance Center called and said: Hey, we are using your product. In this way we got the first check for 180,000 U.S. dollars.
In August 2017, the entire company officially changed to Kong. I am the co-founder and CEO of Kong, and Marco Palladino is my co-founder and CTO. We have been working together for 10 years, but we have been doing the API market in Mashape for the first 7 years, and only started to do the open source business of this new software in the last 3 years.
Glenn Solomon: Is the current market environment more conducive to the commercialization of open source projects?
Aghi Marietti: In the early days of the cloud era, many functions were provided for free. Later, we reached the stage of Open Core, where people sold closed source products based on open source flywheels, using open source products as market entry points, similar to a free SaaS trial service. Basically, people get much less data from this kind of free SaaS than in the SaaS services we provide, but you can see this kind of free SaaS trial products similar to Trojan horses everywhere. On this basis, you can Realize monetization.
Among open source companies, only Elastic has successfully gone public, and its stock has performed well. There is also a group of new open source product companies on their way to the market. This is a turning point. The migration to the cloud has initiated a major change in the market. Compared with ten years ago, there are more software developers on the market, as many as 30 or 40 million, so they need more skills and use cases. Every company is becoming digital, so software is needed. All these factors combine inTogether they create more space for the infrastructure.
Today, infrastructure products are sold either through open source or through APIs in SaaS, such as AWS. So this is an all-or-nothing question. But this kind of closed-source product, as a distribution model, has already withdrawn from the market.
Glenn Solomon: How do you decide which products are open source and which are not open source?
Aghi Marietti: We mainly consider three points: First, if it’s about governance and security products, then launch a paid version to companies, because start-ups or those at the bottom Of small customers have not yet developed to the stage where they need to start paying attention to governance. For example, we are a start-up, and we have not yet begun to pay attention to governance. These products are only available in paid versions for those products that are valuable to large companies. And provide open source versions for products that small tail customers care about, because they need these open source products, and they have to be treated as customers.
The second point is the team. If there are many teams, not just one or two software developers, using your product, suddenly the number of customers grows from 20 to 30, 40, 50. When it comes to functions used by the team, we will list them as corporate charging functions.
The third point is that if you have different data centers on different multi-clouds, and this kind of complex architecture is not available to start-ups, then this type of function will be classified as a corporate charging function. Apparently they are surrounded by support functions. When we classify functions, we will consider these three aspects. Since then, the situation has been ideal.
The functions that meet these three points are classified as corporate fees. Functions that do not belong to these three types of characteristics, we classify them as open source. But every year we have about 10% of our products that are difficult to decide how to classify. At this time, we will let the CTO decide.
Before he makes a decision, we will list it as a paid product, because it is always easier to change from charging to open source. This year there are 2~3 products that operate like this. For example, Caching was originally a fee-based product for enterprises, but we open sourced it. On the contrary, if we don’t know how to define a product, open it first, then stop it and charge for it, it will be difficult to do so, so we decided to never do such a thing.
So, If we are unable to decide on a product, we first list it as an enterprise charging product. After 6 to 9 months, if enterprise customers do not need this product very much, but community users need this product very much, we It can always be open sourced again. It’s always easier to change from a corporate fee-based product to an open source product, and this is how we deal with this problem.
Glenn Solomon: What do you think of Kong’s open source community?
Aghi Marietti: When Elastic went public, Wall Street issued a report stating that companies like Elastic, in addition to the SaaS subscription business, have an open source business as a “moat”. It has more commercial value than a simple subscription service like Zendesk, and its valuation will be much higher. For Kong, we must also have such an open source “moat”. A common pain point of open source business is that the customers you get through marketing are likely to be lost during the sales process. We should think about community building in the overall framework of marketing and sales.
Users are not equal to communities. A company’s technology has 1 million users, which does not mean that they have 1 million community members. A good example is Ngnex. There are 1 million people using Ngnex, but there is no community. This makes us realize: Users are not equal to communities. This is the first point.
The question after this is how to turn these users into “fans”, they will help you design a roadmap, help you modify the code and notify you that your R&D costs can be reduced; help you provide support and Answering questions in the forum will reduce the cost of community support; spreading in the community depends on word of mouth, community members will speak for you, organize meetings by themselves, and save marketing expenses.
If the company has a thousand fanatical “fans”, no one can destroy the company. About 17,000 corporate users are using Kong. This is a huge number of users, right? But we do not have 17,000 product ambassadors, only 10 to 20 users are actively promoting our products. Therefore, the question is how do we turn these people into communities that we can empower, so that they can be empowered to become true product ambassadors. If you can turn 1,000 people into product ambassadors, it means that 1,000 people will speak up for your product, saying that this is a technology that people should follow. But to achieve this goal, it must be true, credible, and not tricky. So, this is a very delicate balance that must be maintained.
How to turn a group of users into a community? This is not as simple as hiring a community leader and organizing several meetings. The actual situation is much more complicated than this. In fact, we still don’t have a way to convert users into community members in batches. We can only convert one if we have one. HashiCorp has done a good job in this regard. The founders themselves are the ambassadors of the product, so they have a lot of “fans”. HashiCorp has great influence. I don’t follow too many people on Twitter, but Mitchell (the co-founder of HashiCorp) immediately tweeted 50,000 reposts and 40,000 likes. It is indeed a very influential practice for the founder to become the company’s product ambassador.
Glenn Solomon: When you are thinking about which products to launch as enterprise products and whether to open source a certain product, do you listen to the voice of the community?
Aghi Marietti: Our core management team often discusses the product’s potential market together. Every quarter, we have to think about the latest situation. If we need to change a certain technology from an enterprise fee-based product to an open source product, the sales team will go crazy. Once we turn these functions into open source, the network will suddenly get stuck until 3 o’clock in the morning. There is always a balance problem here. If you lose the community, you lose important channels. Of course, you don’t want to see your income drop by three-quarters, so you always have to look forward. You have to balance these two aspects. Then let us look back at the governance products, which are still very popular fee-based services.
Glenn Solomon: I guess you have got a lot of potential users from open source users. Will you try to convert them into paying users?
Aghi Marietti: When we first released Kong’s source code, our process was very simple: a landing page “getkong.org” with an “application for demo” (Request Demo) button. Those in need fill out a form and call us. At this time, our CTO will answer the phone and explain to him like a sales engineer. If they ask for the price, I will get involved. We have no idea if they will buy our services. Marco assumedThe role of a sales engineer, and I am like a sales representative. These conversations are very interesting. We talk to each other in all kinds of things, but we are not good at sales. We got our first 10 customers in this way.
Later, we formed a sales team and recruited 3 or 4 salesmen. We still kept the same “Apply for Demo” button, and then we started to divide it by region. This model has the “Apply Demo” button on one side and the sales representative on the other side. Orders can usually be closed within 3 to 5 months. We have used this model to achieve annual sales of 10 million U.S. dollars. There are almost 900 potential customers coming in this way every month, which is how most of our business comes.
This may seem like a good thing. We don’t need to do active sales, but this method alone is difficult to bring our sales to a higher level. We must open up new sales channels. We have also tried to set up a sales growth team (SDR-Sales Development Representative). But when we set up this team, the speed of the business slowed down, and then we decided to abandon this practice. We launched a free trial version, collected potential customers, and conducted qualification review based on some criteria to see if the counterparty has a budget, needs, and whether the time is ripe, etc. Then transfer this potential customer to the account manager (AE) as soon as possible. As a result of this, our sales conversion rate began to recover. We are also constantly expanding the way of sales, in addition to the existing sales model, adding a sales model.
Glenn Solomon: How did you build a sales model within the (Inbound) sales model that you have always depended on for survival The level of marketing model?
Aghi Marietti: Our past practice is that 50% of our business comes from inbound customers(Inbound) model, the other 50% we intend to let the team of business opportunity development specialists realize through the sales model. But this situation has changed in the past six months化. We are still in the business of inbound (Inbound), but the sales business is managed by the account manager(AE) to complete. Opportunity development specialists can help in promoting the business, but the account manager (AE) must play a major role. The main task of marketing is to increase marketing qualified potential customers as soon as possible (MQL). The business opportunity development specialist is concerned with finding leads and turning these potential customers into qualified sales leads (SQL), or sales. In the past, we only had three people in charge of sales. As our business continues to differentiate, more and more sales staff are joining in. By the end of this month, we have 25 sales staff.
Only relying on the business obtained by inbound customers (Inbound) cannot feed everyone. So we have to start organizing the sales channel ledger from Monday, every Monday. This is a change we made in the company. Unless you have the Datadog model and an unlimited credit card, you can use the simple inbound (Inbound) model to increase your business for a longer time, similar There is also Slack or Dropbox, so that your inbound (Inbound) business can grow longer. But if your company is an open source infrastructure company, a simple inbound customer (Inbound) business cannot grow to such a high level.
Glenn Solomon: Do you use sales engineers (SE) in the sales process? p>
Aghi Marietti: This is actually a sales process driven by sales engineers. The sales engineers in our best-performing sales team are also the best. When meeting with a potential customer for the first time, the sales engineer is the one who can win the trust of the customer. Then sales can push the contract for the next 2-3 months. People may think that the sales engineer is just the assistant of the sales representative, but in our company, the sales representative is the assistant of the sales engineer.
Audience Q&A
Audience 1: Every open source company you mentioned so far, as far as I know, is an infrastructure company, such as Elastic, Kong, and others. I want to know, what do you think of open source applications? And do you think they will have a place?
Aghi Marietti: Take GitLab as an example. It can be used for personal use, but it is really a collaborative software. There is no open source version of this product, and so is Mattermost. I think this is a new trend that we will see in the future, that is, not only does open source gradually occupy the infrastructure market, but the application level has also begun to become open source. This is a new trend. The key question is, how do you monetize the cloud version, I think this is very important. Only through the cloud can these products be realized. In terms of infrastructure, the current practice is still to monetize through application downloads and run the application on AWS. This is the case for HarshiCorp, Kafka, Elastic, and Kong. At the application level, it may be another situation. In 99% of the cases, services are sold in the form of SaaS. But this looks more like a closed-source model, so the influence of the open source model is still undetermined, but it is definitely not limited to infrastructure.
Glenn Solomon: I agree with Aghi’s point of view. Open source is a very powerful model, but it may be thought of in a slightly different way. If you start a company, how can you enter the market in a very disruptive way? Because if you want to enter an application market, there may already be such applications in the market, even if you want to enter an undeveloped field, I think it is difficult to open source in the undeveloped field. Because, in this field, people still don’t know what they need, and they don’t know that someone is already doing it in this market. You have to think clearly likeHow to raise public awareness on this point. But if you are thinking, how can I subvert the existing products that already exist on the market and enter the market in an effective way? This seems to me to be a starting point.
If you are making an application, in many cases, the correct answer is that it looks like an open source application, but it may not be open source. Because in the end you have to monetize it through cloud distribution, as Aghi said. It is possible to achieve this in a free trial mode. I was an investor in Zendesk before, and they did a good job in this regard. Maybe they will let you try it for free until you reach a certain level of magnitude, such as what Slack does. Slack is another company we invest in. These are some non-open source models that are very disruptive to the existing market, and it is difficult for you to compete with them. In these two markets, there are indeed some open source companies, and their emergence does not necessarily affect the development speed of these non-open source companies.
I would say that after Elastic, the next batch of companies that are about to go public will still be infrastructure companies, such as Databricks, HashiCorp, and Confluent. They are all infrastructure companies. I think that application companies are still behind them. Of course, application companies will also develop to this point, but the top five companies behind Elastic are indeed infrastructure companies
Listener 2: How do you convince the CTO to embed Kong in their infrastructure?
Aghi Marietti: Our view is that you don’t need to persuade the CTO. You need to persuade those who report to the CTO. These talents are the ones who really make an impact. When making a decision in a large company, the decision-making power will be delegated to the level of the vice president of technology, and the vice president of technology will play the role of CTO or chief architect. These talents are valuable roles for you. Only if they become your supporters, they will lobby the CIO when necessary. If the purchase price reaches the level that the CIO must participate in decision-making, usually when the total annual price reaches seven figures, it needs to be reported to the CIO, but if the price is only 100,000 yuan, usually the chief architect or vice president, or even the level of director It can be decided. So the key is that you have to convince the technical director, or digital technology director, or chief architect to get your 100,000 yuan order. Once you are successful, you start to meet with higher-level people, and eventually, after three years, you will sign a larger order. If you negotiate a $5 million order with Coca-Cola from the beginning, you may spend more than two years. But if you sign with the digital director of Coca-ColaAn order of $100,000 is a huge victory and the beginning of your success.
Many times, we have a misunderstanding that the CIO is the decision maker, but in fact the CIO is just the person who makes the purchase decision. They are only responsible for signing the documents. In fact, it is easier to convince the CTO than the CIO, because the CIO trusts the CTO, and the CIO is not an expert. When they encounter problems, they will ask the CTO, what do you think of this technology? CIOs will not make this decision on their own because they do not understand these technologies. You may find that many CIOs of listed companies come from the background of CFOs, so they are actually purchasing personnel. Only in rare cases, they will be engineers. You will find the presence of the engineer in the VP. Under the new sales model, I think the vice president of technology or senior vice president of technology will play a decisive role in the next 10 years. They will one day become people with the power to make large purchases. The CIO’s authority will also change.
Quick question and answer
Glenn Solomon: What is the book you most want to recommend to other entrepreneurs?
Aghi Marietti: “Primed to Perform”. This book tells how to maintain the company’s culture as the company grows. From this point, we talked about sales, and then we talked about personnel management. So for a growing company, this is really a great book.
Glenn Solomon: What advice do you most want to give other founders?
Aghi Marietti: We tend to overestimate what we can do in the short term and underestimate what we can do in the long run. Don’t give up, don’t just look at the first three years.
Kong’s inspiration for domestic open source entrepreneurs
In the IT infrastructure track, open source has been a proven and replicable business model. But the commercialization of open source is no easy task. The founder and team need to have a clear understanding of this. At the same time, there are no ready-made answers on the road to open source commercialization. It requires an agile organizational structure and decision-making mechanism, and constantly trial and error to find the most suitable direction. Kong was not an open source company at first, but once it chooses to open source, it can continue to iterate and eventually become a unicorn company.
In the commercialization of open source, the most important barrier is an active community. An active community can help the company improve R&D efficiency and reduce marketing expenses, and it can also help the company increase its brand appeal. At any time, the community should be the focus of the company’s work list, and sufficient resources should be devoted to this. Kong has always attached great importance to the community, and can even sacrifice short-term commercialization for the community.
Successful open source software companies do not rely on a certain “explosive” high-star number project, but are able to build a series of tool chains around their core business, such as Kong. In addition to the core API gateway, there is also a Star open source projects around the series can provide integrated solutions for various customers. Therefore, open source startups need to have a clear product strategic plan, and continue to launch products that are popular with community users and commercial customers. Finally, as an enterprise service company, open source software companies also need to build a strong marketing and sales system, which includes sales staff, customer success, pre-sales engineers, after-sales engineers, etc.
At the same time, gather customers (Inbound) and promote (Outbound) needs to be closely integrated to maximize the marketing value brought by the open source community. Kong started with the pure set of customers’ (Inbound) model and successfully built a sales promotion system, which is also an important guarantee for Kong’s continuous and rapid growth in revenue.
This article is from WeChat official account:GGV Jiyuan Capital (ID: GGVCapital) , author: business services group