Sina: Peng, I think something I mentioned as I was introducing you was that, again, many people who live life or build their companies and don't necessarily see many patterns around them that influence their life. Some people see those patterns. Some people, I guess, go one level higher. They are able to choose these patterns and, you know, to their advantage. Some people make these patterns and create them. Now you kind of have gone through that entire evolution and you've done it in three very different industries. In the whole dating industry you have done it in, I guess, identity management, you've done it in content management systems.
Before we get into more specific questions, I'm just curious, how do you see the world?
Peng: I have a thing that I say that if you get to know me a little bit more, you hear it over and over again. Ridiculous states of the world. That's what you look for. A state of the world where you go, “What? You know, this makes no sense whatsoever.” Once you see something like that and you validate it, it's not just a mistaken view of it. You really look at it and it's really nuts. Then you know there's an opportunity.
Having to go through classified ads with a finger and get it black. This is long ago. Then go, oh, I should call this person. That's really nuts. There are these things called computers that can digitize all that and you just put in what you're looking for in the search and it shows up. So that's the start of electric classified scripts.
Having thousands of people try to produce stuff on one website and not take the whole website down. That's ridiculous. You need a system like a manufacturing process where you can put things together assembling the car or assembling a document and then it shows up on the website in this case. So that's in the world and so on. So, I guess when I was younger, I looked at ridiculous states of the world in very narrow areas. But as I look at the world over and over again, you start to see big opportunities like the foundation I started, it’s called Solve Education. To summarize it – about a third of kids alive today will not be literate. There's not one person that is looking at this and “Go, okay, yeah.” Most people when they realize this, go “This is nuts. Why are one-third of kids not literate?” And then you start to realize you can even solve that problem using technology because most of us are going to have cell phones and so you just put the teacher in the cell phone and you can start solving the problem. So yeah that's how I look at the world. So you got to be curious enough to look at the world to look at ridiculous things to try and observe ridiculous states. Frankly, if you look at it, it's kind of yeah it's interesting but it's not ridiculous then maybe there's not really that much of an opportunity.
Sina: Were these ridiculous at the time you were looking at them like up with other people look at them?
Peng: Yeah, the companies I started all came up from that perspective.
Sina: Thank you for sharing. I do want to make a segue into tech because there's so many questions, so specific that I'm thinking people may get mad at me if I ask philosophical questions.
So about R+K>1, it’s a mental model, I know that you wrote the blog post about two years ago. I found out about a couple of months ago through Ty. How did that whole thought process come about?
Peng: Well the genesis, I mean from the start of Monk’s Hill we were looking at this LTV CAC model that Google and Facebook has got us thinking about. Of course they'll get us thinking about it that way because they are the CAC, right? So we started building business models where we don't control the parameters of the business, right? So CAC is a function of Google, Facebook, Marketplace, now TikTok and so on but you don't control CAC. So, now you have a business model where you don't control the business model. We've actually lost companies that way because the CAC when we entered the investment CAC was like you know 10x, LTV CAC was like 10x right? By the time we exited it was less than one, and that was over a four or five year period. So we realized that was not the right way to do it, it took us a while to figure out “Ok, if that’s not it, how do you think about your company?” At that time I had just come out of a board role in YY. How many of you have heard of YY?
Sina: What is YY?
Peng: It’s the biggest social network no one has heard about. At one point, half of China was on their system. It was throwing off, at the peak, about 1.3B in free cash flow. So it was huge, the parts had been sold off just in time before the government clamped down on all this stuff. But you learned a lot about human behavior at scale. You change the color of a button and suddenly the number of click throughs go up by 100%. So you start to think about issues like retention, engagement, virality and so on. So you can see about 8 years ago, I had blogs on these things but 5 years ago I started thinking about retention and engagement and how those things behave for a business to make sense. This is not rocket science, in fact it’s addition. It’s this plus this, it’s very simple. You realize there’s no articulation of engagement in that equation. Engagement is important for a business but frankly if you look at it from a long term perspective, if you just have R and K functioning at greater than 1, you have a business that just keeps growing itself. You don’t have to do anything, it just keeps growing. And that’s what we realize, that’s the formula we’re looking for to model our companies.
Question 3: What sample size is required for this to be relevant?
Peng: You want a certain level of confidence that your experiment is giving you the right answers and therefore you are in the realm of science, basically right? Statistical significance, standard deviation from norm. Depending on the level of confidence you have, if you’re doing a science paper, they talk about the 0.5P. Take a statistical view, but if you ask a VC the larger the number the better, but minimum would probably be 1500 sample points.
Question 4: How do you think about the relevance of this mental model to other types of businesses other than consumer and social?
Peng: I’m actually a B2B guy given my background. Most of my business has been in enterprise sales. This model initially, we thought, doesn't work so well for enterprise sales. I still don’t know if it will but we have CEOs who are starting to think instead of B2B, are starting to think about B2E2E2B. E’s are the employees. It’s always people you deal with even if you’re B2B. So how do you engage the E’s within the B's in such a way, they are excited about the product in such a way that they learn things and their career development? Maybe you’re one of the reasons why they got the promotion but if you’re not there to tell the story, they might not tell the story. But if they tell the story, you’ll get viral. So, maybe have career development sites to have your accounting customer expand their sphere of influence to be recognized as an expert. So how do you build that kind of stuff so that your fanbase becomes the employees that use your product and they start getting you more business?
Question 5: Over what time period would you measure R+K>1 for it to be valid?
Peng: There’s no point in measuring this over 30 days, it should be more human scale. Minimum I would get is 6 months to 1 year, ideally longer. This makes it harder for start ups because you need to demonstrate R and K over a longer period of time. If you do it over 30 days there’s a chance that you would stretch it out another 30 days it would come down but there’s a chance it could go up but I prefer 6 to 12 months minimum.
Question 6: To increase these two it means that, especially early stage startups, you need to dedicate a lot of product and engineering processes and resources, come up with new hypotheses, test them, and so on. Do you have any advice for companies with limited funding and resources to figure out what are the levers that maximize those?
Peng: Yeah, there's a whole spectrum, right? If you're doing B2B, it's okay, because your LTV CAC is probably, you know, big enough, you can take your time to build your community, and frankly, you need to engage a lot of employees before you can start building the community. So you can be a bit more delayed. But if you're building something that requires the community to improve your economics or your business, then it needs to be built up front. The R&K are part of the business model. You can't just, you know, wait till you see that, you know, absolute product market fit, scale your business, and then start thinking about building it because your cost structure will be very different.
Sina: So the virality is built into it by design, in the case of the previous slide.
Peng: Yeah, that's a whole ton of stuff we can go into about R and a whole ton of stuff we can go into about K, but you've got to use all these tools as you are building your system, I think early on. I think the first piece is product market fit, right? You know that kills like the lack of product market fit kills about 60%, 80% of companies in early stages. So, you've got to get that or you've got to figure out ways to avoid that problem, right? That product market fit risk. But right after you have that, then you need to be able to grow the business, and that's where the R plus K kicks in. It's almost immediate, right after you start.
Sina: How do you bootstrap, and how do you earn the right to use that, let's say? How do you measure your success before you can use R plus K?
Peng: Back to basics, right? R is a great product, K is a great product. Great product people stay, great product people tell other people. So you build a great product and you'll be fine. What is incumbent on you is to figure out how to structure the product so that it's as retentive as possible and as viral as possible. You can have a very high R and high K tendency in your product because it's a great product, but you didn't structure it so people want to stay longer. Well, if it's a great product, they'll stay, but for example, you don't have lock-ins, which cause users to not be able to leave the product even, because they're dependent on your data ecosystem, for example. And then the virality is also, you know, if it takes 20 clicks to get some of your friends in the system, you just won't do it. But if it takes one click, maybe it happens.
Question 7: Could you discuss how to balance paid growth strategy versus virality and retention?
Peng: So in the end, when you have zero users, you can start bootstrapping them by inviting all your friends, right? But that could be a very very slow way to crank up the system. So you start paying money to get people on the system to actually see if your R and K are any good, right? So it's a process where you use a traditional R plus K method to sort of jumpstart your engine, and then you figure out how much to pump in depending on what the R and K is. If the R and K is like very much less than one, there's not much point pumping in more and more users, with the one exception that LTV CAC is a lot greater than one, which means you're making money, so you can keep doing that old traditional way of making money. But it doesn't help that you have a leaky bucket, so everything you dump in it, leaks out in the end. I would not spend too much time cranking up revenues if you can see that the cost of revenues is gonna keep going up and then you have a leaky sea. I’ll use that opportunity when you have positive unit economics on LTV CAC to ramp up your R+K. When you can profitably pump the system with users and then experiment with the R and the K, right? You should just assume at some point in the life cycle of your company, your LTV CAC might be less than one, in which case you don't have a business model on that end. So the race you're running is to get to R plus K greater than one, before LTV CAC falls less than one or actually less than two, because once it hits two, you don't have a business, typically.
Question 8: Are there any ethical considerations when optimizing for virality and retention?
Peng: Yeah, I mean, you can do not nice things to create virality, etc., but I encourage you to be a positive, high ethics entrepreneur when you do all these things, right? And get people to stay because they want to stay. Just make it easier for them to do it. Get people to tell their friends because it's easy to do it and they want to do it, right?
Audience question: So the way I understand the formula, working for B2B, clearly, everyone is talking about PLG, product-led growth. So the virality sits, I believe, on top.
Peng: It's actually part of the product.Yeah, the ideal architecture would be for that R&K to be just part of the product, how it works and how the customers see it, et cetera.
Audience member: And so, when I was thinking about it, I was struggling to first see how to apply it in a traditional segment where you're selling tech into manufacturers, but one thing which is, you know, hopefully can become an aha moment is leveraging community. Because, how can we build something and build a community for our users, and that can have virality, that will take time, and that's probably something at a later stage, but I'm just curious, have you seen other things?
Peng: So there's a lot of good reasons for you to build those communities early on in creating your product as you go to market, because it actually supports you in your success. So if you go down that path, then you can think of the community as just part of the product you're building. The product has all this functional stuff that when you plug in, does things. And then all these use case experts and scenarios and consultants, you know, people that will help you if you get into problems, right? People are just generally helpful, right? So if you put them in a community where someone has experience doing X and someone is struggling doing X, they will help, right? And frankly you get the solutions a lot quicker than just having your engineers go figure it out.
Audience member: So community is a good lever in this kind of a traditional sector. Anything else you've seen working?
Peng: So that's been it even in traditional B2B businesses with built communities, but we never thought of it as a B2B, B2E to E2B kind of way of looking at stuff. It's only now that with the R plus K model that we're starting to see if we R&K in that world. So the quick answer is no.
Sina: I have a question that comes to mind. When we think about community, some businesses may look like they can't ever have a community because there's SaaS or something specific. Are there any odd examples of businesses that you thought they couldn't have a community but they somehow did and it feels pretty natural.
Peng: I've seen a lot of very esoteric businesses, semiconductor, supply chain optimization, all that stuff. There are people that care deeply about some very deep esoteric topics and enough of them they form a community. I think people collect all kinds of stuff, right? Human beings are interested in all kinds of things and sometimes you look, why are you collecting stamps? And you know, it turns out that millions of people collect stamps, right? I don't know. We can always find commonality.
Question 10: Are there any indications for when your product is ready for virality?
Peng: Your retention is as high as you can make it, right? So before your retention is high, there’s not much point pushing the virality. Because if you get virality, it’s gonna drop again. So if your retention is low, you've got to really think about if you have product market fit, right? And if you have designed a product that people want. And again, there's a whole bunch of stuff that goes into retention, architecture, etc. We're actually thinking of creating maybe more discussions on R and K and all that stuff. But I would be hesitant to spend too much time on K before R is reasonable.
Audience Question: So, I'm really interested in the enterprise sales space when the customers are both the tech leads who will purchase the product, but also the end users are the people below. How do you approach building community and what mediums do you use? You mentioned learning pathways and you mentioned also potentially using LinkedIn as well, but like, what are the, and I'm doing this to increase my chances of success, which are the most beneficial mediums that you see the best return on success?
Peng: So I'm not an expert on sort of community building for enterprise communities, right? I've just seen them done a lot. And there's all kinds of tools you can use to get people together from Slack to, you know, yeah, all that stuff, right? But I'll give you one hack that is very useful in my experience as an enterprise person. And I'm surprised this is not in someone's standard toolkit that they use. And that is to imagine your FAQ as a set of videos from your customers. The answers to your FAQs, right? So over time, you want to have an FAQ that any question someone asks is answered by a bunch of your customers several times over. That way, you actually don't even need to have reference calls. After 50 customers say, this is great stuff, this is why it's great, this is details of why it's great etc. No one's going to go, can I talk to yet another customer? They're just going to go, “Okay, I trust you.” And this is a deliberate strategy that you need to start off early on. Customer asks for a discount, I'll give you a discount, but you need to answer, you need to let me film you answering a bunch of questions as a testimony, right? So if you do that, after one, two years, you know, your sales cycles just drop because of this. I've seen this multiple times already, so this works.
Question 12: As a VC, to what extent do you try to validate the company's R plus K when you're trying to look at their pitch and invest?
Peng: So this is not religion in Monk’s Hill. The partners can choose to evaluate a deal any way they want to evaluate the deal. So all our founders, all our partners are former CEOs, right? So we don't treat each other like kids, right? So this is the religion, everyone needs to do that. We don't do that. So when it comes to me, I'll tell you my perspective, right? Any company that is going after mass numbers of users, I'm looking at their retention, I'm looking at their virality. And if they're not there, or worse, the founders have no clue that those things are important, we don't invest. I don't invest. It's strange that in this day and age, there are founders that when you ask them about retention, you ask them about virality especially, they just have no idea how, you know, lift works. Just, you know, you've got to know.
Question 13: What are some practical measuring tools for R+K? We are all using Mixpan, Google Analytics. What's the best way for us to just track that kind of number?
Peng: There's lots of tools, so there's no right answer to that question. It depends on your business, where you're tracking. In fact, if you think about what an active user is, right, that's not a well-defined thing. You have to define it for your company, right? And then once you have that definition, you can talk about retention, you can talk about engagement, you can talk about virality. So if you're going to define a lot of these terms internally, then the tools can be a varied bunch of tools. But the point is, you've got an instrument for measurement, right? If you don't instrument, you're not going to measure. K is a lot harder to instrument for, by the way, if you think through it.