Juli 1, 2019

Episode 8: How Does GOJEK Design A Product That Actually “Fits”

Episode 8: How Does GOJEK Design A Product That Actually “Fits”

Intro: Welcome to GO FIGURE. My name is Nadiem Makarim, CEO and founder of GOJEK Southeast Asia's first Super App. GOJEK does ride hailing, food delivery, payments, even on demand massages. You name it. We do it. GO FIGURE is a podcast dedicated to expose the inner workings of ambitious tech companies in the emerging world. We like to talk about things we like and talk about things we don't like. There are a lot of myths out there that we want to dispel. So keeping it real is kind of our mantra. Hope you enjoy.

Nadiem: Hey guys, welcome to GO FIGURE.

Guests: All right, thank you.

Nadiem: Thanks for being here. Um, today we're going to talk about product market fit and how to scale products and some of the pitfalls in product execution. Right. So, um, one of the most important topics I think for us. But first before I'd like to, uh, introduce, we've got Niranjan here, our CTO of GOJEK core. Thanks for being here.

Niranjan: Thank you.

Nadiem: And we have Aakash over here. One of our heads of product, uh, who has led multiple products all the way from food to merchant promotions platform to a variety of different things. We've all been here from the beginning, since the dark days of GOJEK, when, when there were only three tiles, right? There is only ride ride on a bike. Send me something and who buy me something. No, not even food.

Niranjan: Food came later.

Nadiem: Yeah, food came later. It was just like buy me something. Right. So why does this, why does this topic even important? Why is product market fit report? I think that I'll, I'll kick this off and say that there are so many biases and pitfalls in what we think people need versus what they actually do. That you know, many people even with the greatest intuitions can make some of the most damaging mistakes when thinking about product. And so let's talk about that a little bit and let's talk about, um, Aakash, you had a framework in mind with which, how to think about, uh, growing products or, um, actually, uh, scaling up a product.

Aakash: Yeah. It's a very simple framework, almost you can think of it as a growth loops when you are starting a new product. Our product is nothing, but you are imagining a particular need of a customer being served by what you're building. Um, first we need to validate that that needed system that need exists, a large enough number of people, uh, to be relevant in the market? I'm finding that fit was called product market fit is absolutely critical and that has to be the first step of any product development because if you don't get that right, you're effectively trying to optimize a wrong solution. Right? So that's the first phase.

Nadiem: Get the problem defined as accurately as possible

Aakash: and also find a solution.

Nadiem: So I cannot tell you how many startups I have, I've coached, mentored or seen that jumps straight to the solution. Like that's basically the default mode, right?

Niranjan: We have done that in the past?

Nadiem: We have definitely done that in the past. And that's why I recognize it and it's like, oh, I have this great new solution that does this, this, this, this, this, and like slow down. What is the problem you're addressing, right? So what is it about founders or entrepreneurs or leaders or managers in tech companies that have that natural instinct to jump to the solution? What's causing that?

Aakash: Because that's the exciting part. I guess. That's where they have fun. Um, but yeah, I mean that's what people consider as running the business or doing any work is, is that you have come up with a solution and it's not just related to the tech company founders. I mean we all do that in our lives. We try to solve things which may not even exist. We try to prioritize things in our life which are in the wrong priority. I mean it's just human nature. It is reflected in the product managers of product company's space because it is so fuzzy You can always lie to yourself if you don't have data or large enough number of people to validate it. Like when you're working in a financial market, you can't lie about what's a profit or a loss. When you are working in a technology space, either your goal works or it doesn't. But if I'm seeing that, hey, have I found product market fit? Or is this the right solution? It's just my judgment.

Nadiem: And there in lays the complication of having debates about product market fit without having data and without having something out there with which to triangulate your data

New Speaker: and extend your point, right? This problem is so core to a lot of people. That's one of the reasons our first company value, right? It's not about you,

Nadiem: it's not about you. Because people that you know, there, there are so many biases that you have specifically confirmation bias about your preexisting beliefs that you want it to be true, so bad that you will ignore counter, um, intelligence basically, right? Data that reveals the opposite of what you're saying. And I have a perfect example of this. Like in the beginning when we first launched, we didn't have, uh, we didn't have food yet, right? We only had ride, courier and a concierge service shop, uh, at the beginning and only on motorcycles. And basically I remember this meeting with Sequoia, um, uh, who were our early investors, uh, but this is pre investment and the meeting very quickly turned out to be a debate. And basically the debate was I was so convinced that GOJEK was going to be a primarily a courier and logistics arm, right. And that was going to be the biggest thing and that we would be serving businesses literally. That was the initial, because you know why? Because before the APP launch, that was the core of our business. That's the people who saw value in a pre app GOJEK where businesses that needed to send stuff around. Right. So I thought we were a courier business. Um, well, very quickly down that debate, you know, I think what Sequoia helped mentor me on that is that, you know, look into the data and find out, you know, what are actually bigger problems of the user that you're addressing. And it very quickly came out that when we looked at the shopping data or the personal concierge data, they were about 80% food. And that was the moment I lost the argument. Right. I couldn't dispute that data and when I realized that, I'm like, okay, okay, we're a transport and also a food delivery company first and foremost.

Aakash: That's such a beautiful thing to happen, I really love GO SHOP for that one reason. Um, apart from a lot of others, but this one particular reason. I see that as a petri dish of products, right. It just says that, hey, this has a slightly higher, um, user experience friction to use because you know...

Nadiem: Wait for the users that don't know GO SHOP is our personal concierge service. That's just a fancy way of saying I can tell a driver to go buy me something anywhere.

Aakash: No, I basically tell a driver that go to this particular location, I don't specify whether it's a shop, it's a mall or whatever it is, and I have an text box and I'd say that I want you to do this and I think it's going to cost this much amount of money, which I will pay you and that's it.

Nadiem: That's it. It's just a task.

Aakash: That's it. It's just open ended. Right. So open ended.

Nadiem: And it became the fountain of ideas and confirmations because if you create a generic service like that, then really interesting things come out.

Niranjan: Yep.

Aakash: Going back to the product market fit part, uh, not every business can afford to have something like this or have the ecosystem but for GOJEK this work very well because now you can actually look at the GO SHOP's data to find product market fit for new businesses. It's so beautiful.

Nadiem: That's right. And go food is now really depending on GO SHOP to feel new restaurants that perhaps weren't listed, but reflects the demand there or perhaps a restaurant's who's visibility was hard to find for that user. And we add that to the ...

New Speaker: We actually got that feedback. So we started looking at the GO SHOP data and found a certain set of restaurants being constantly ordered from there. And then we discovered that somehow our search did not really cater to those restaurants well, because there are mistakes in the set up. Uh, the search was not working properly. The ranking was somehow screwed up and we just kept looking at that data and correcting, improving the GO FOOD experience. It's just, it's just a signal. Right. And you need to look out for the signal whether you have something like GO SHOP or not. Not every company can afford to have or is lucky to have that particular set of signals, but then you look for a signal somewhere else. You look for signals in competitors. You look for signals in, uh, internet in any different tools that people are using. You look for signals in real life, what people are doing, right? What kind of hardships they go through. That's where you find the need. That's where you find the problem and you need to validate that, that is the real problem that you're trying to address. And it's not some hypothetical that maybe you and five of your friends care about and nobody else cares about. Right?

Nadiem: And so there's kind of two ways of going about starting a product fit assessment. On one end of the spectrum is you go and spend months doing research even before ever launching anything, validate research, et cetera like that. Then there is the other side of the spectrum, which is let's get something out with no research, and see what happens and then start collecting data. Now is either way right or wrong.

Aakash: Well in product development there is no right and wrong..

Nadiem: That doesn't help me Aakash. it's the nuance.

Aakash: What you're optimizing for is to find product market fit as early as you can, as economically as you can and with the high degree of confidence, right? You are optimizing for time, cost and your confidence that product market fit really exist. That's what you're optimizing for. Now, uh, what you need is you need early adopters for finding product market fit because someone has to use it. Um, you need to identify the right problem, come up with the solution and see that without throwing money to get the growth, it's exploding. That's what happened with coaching.

Nadiem: But before that, are you, are you saying that you mentioned get something out there. So is there, do you think there's any value in doing research without having something out there to bounce your assumptions from? Meaning some kind of an MVP, so, right. Some kind of a, a minimum viable product.

Aakash: So it completely depends on what is the cost of building that something is. Right, right. Building something like GO SHOP. It's very cheap but let's say...

Nadiem: Because it's simple. Cause it's a simple service.

Niranjan: Yeah. But let's say you're building a phone. Now building phone is you have to put in a lot of research, lot of time, effort and optimizations before you can get a product out, a prototype out. So you're optimizing for cost. You have to constantly looking for what is the lowest fertility tool. Which I can essentially take this thing to market and start talking to customers because you are looking for data, right. And hence, uh, even the kind of research you do has different steps, right? There is, there are qualitative service. Uh, after that you can essentially do a quick design and just show them paper mockups thing that hey, look at these paper mocks and tell me whether it makes sense for you or not. Or you can build a dummy prototype which partially works, uh, or you can build a full fledged product. Now on one side there is cost of building on other side, the number of customers you can approach also keeps increasing. Like if you have a working product and prototype, the number of customers you can reach is going to be far larger. And hence statistical significance of your data is going to be far more meaningful, giving you a larger, higher confidence. Versus if you want to do group studies, you can do it with 10 people, a hundred people.

Nadiem: So am I correct to assume that if the problem you're trying to address requires a simpler solution? Generally speaking, the better way would be to experiment quickly and fast and something out there. But if the problem that you are targeting has a far more complex and high investment solution, that is where you put some more upfront non-expensive uh, research, uh, in the beginning.

Aakash: All right. I want to add one more nuance there. Um, it's also depends on your ability to acquire customers. Just because you build something doesn't mean that you will be able to get customers on board. So, uh, in B2C space, that doesn't matter as much, but in B2B space, that's a really, really critical aspect. You are not going to build something and just hope random people will use it and give you feedback. You still need to see that particular ecosystem. The second part is in case of GOJEK, whether it's GO SHOP or GO RIDE it's a, it's a very common human need when you are the user. There are very few products where any, any person can get up and say that, I am the user, that's right.

Nadiem: And everybody uses transportation.

Aakash: Everybody uses transportation, right, you have to travel. That's, that's not optional.

Nadiem: So in those situations, you can leverage intuition. Much faster

Aakash: Exactly, at that point investing heavily into research is just stupid, right? It just a waste ...

Nadiem: Why do I need to prove that people need to go from A to B really quickly, right?

Aakash: Now going to the other extreme. Even if building something is really cheap, but you are not the domaine expert, you are not the most common user. You are not surrounded in that ecosystem. Very likely you are going to be wrong in whatever solution you come up with. So even if prototyping or even if you're coming up with that one solution was easy, there are going to be hundred hypothesis that you will come up with out of which one is going to be right, right. If you try to prove those by building software, it's just too expensive. You actually might want to go there, do research, validates some basic hypothesis, understand the problem space better, and then build those cheapest solutions. So where you just release maybe five iterations and find it in five terations, right? So, so it's a complex ecosystem.

Nadiem: You've experienced this yourself right? When you're building the merchant promotions platform. For the audience that doesn't know, GOJEK has a really large, uh, a set of tools for merchants to grow on our platform. Merchants define what we mean by merchants is restaurants, for GO FOOD and our food delivery platform. And so you've never run a restaurant?

Aakash: Nope. Never.

Nadiem: Right. You've never owned one. You don't know what it's like to operate,

Aakash: I do not empathize with that customer segment at all.

Nadiem: Right. So how did you overcome that problem? Did you just guess in the dark.

Aakash: So I think it was three critical components, one; is you can look at a parallel ecosystem. Something like merchants to merchants platform did not exist in Indonesia. So I can't look at for the local competition to validate whether the problem is basic or not, but I can actually look at places like China where the overall food delivery ecosystem or the food ecosystem is much more mature.

Nadiem: So you're looking at comparables elsewhere,

Aakash: comparable universes elsewhere where the similar dynamics of profitable business, food delivery, customers, data being available as long as these underlying components exist that are not too many variables in how they interact with each other. So you can actually study that and brought parallels, that is one. The second thing is; once you have that and you say, hmm, this actually makes sense and you have some value proposition that you can go and pitch to local merchants, you go on and start talking to different, different merchants. So here it's important that you are able to segment your merchants by their own incentives. And then in the surrounding, that's the restaurant business. You have the large, uh, McDonald's, KFC, Domino's, these players, uh, you have boutique restaurants which are not very heavy on delivery. They actually want higher food fall, uh, because these are fine dining restaurants. Then you have these smaller stores which really, really want as high traffic as possible because their cost of producing food is very low right. Now these are different segments that are far more actually. But you pick up the segments which are relevant, where their need for growing business and the way they look at the problem is going to change. You go and talk to them and you try to run your idea passing that hey, I will make this available to you for free. Would you use it? What's your problem? And that's how you start finding more data. That's what we did once we hired a decent understanding of the ecosystem, our customers, we developed something and then launch it to those 10 people. Uh, that's where we also used our research as a way of doing legion. So your research is not necessarily just research specifically in B2B space. Your research almost always will convert into your Alpha customers.

Nadiem: Okay. So I have this intuitively in my experience, having conversations with people always inevitably leads to more important insights about the problems themselves. And maybe it's because we're fundamentally dealing with human problems. And therefore language is the best way to actually unearth nuance and to find out the why. Right. You had a point before about the difference between qualitative and quantitative data. Um, how do you see the balance between the, because there's some die hard people who believe that man, uh, you know, having conversations with people you, that's, it's definitely statistically insignificant. Right. It's very hard to justify how having conversations with the people who are statistically significant. They are knee deep in confirmation bias. You could ask the leading questions during that interview process, but at the same time, for me, the first ever conversation, the first ever data point that I got to inspire myself to start GOJEK was pretty much 10 to 15 conversations with normal motorcycle taxi drivers that I took home from work. And based on those 10 to 15 conversations, I made the decision with the other co founders to let's do this. Let's start. GOJEK. Right. So how, how do I reconcile that?

Aakash: That's called confirmation bias.

Niranjan: Partly confirmation bias but more importantly, it's what you said earlier, which is you are the user as well. Because finally what your objective is to have that confidence that what I am thinking is right and hence I'm going to double down on it, invest in it and take it to the larger market. If that confidence can be built by talking to 10 people, awesome. If cannot then you need to talk to 100 people.

Aakash: So to take that to extreme, right? Um, when you look at the human needs and let's sort of stretch the Maslow's hierarchy further down. Survival is a very basic need. Let's assume that you are living in a war zone. You don't have to worry, do customer research. Whether you need to sell arms.

Nadiem: It's obvious.

Aakash: It's obvious. Uh, food, whether there is a need to sell. Make food available or not. That's really obvious. You go up this particular hierarchy. You don't need to do an extremely high quality qualitative research to find these things. But as those needs become sparse and sparse and sparse for example, if you're figuring out whether you are going to come up with an application, which helps people learn philosophy, you better do a lot of customers research to understand whether that need actually exist or not.

Nadiem: So where are you are in the Maslow hierarchy of needs actually determines the amount of research. But I also, I also have, I mean if you look at most founders stories, they're littered with stories like mine. Whereby something happened to them. They got curious. Talk to a few people and then say, wow, this is a real problem. Let's do it. And I want to offer a bit of a, of an explanation for why low number of samples maybe as powerful, if not in some cases more powerful of an indicator of a problem. Okay. So I mean highlight white. So for example, uh, remember when we used to have these, uh, I used to have WhatsApp groups with drivers and some merchants, etc. And, and, uh, I would also have, you know, because GOJEK, everyone in my family, my friends and everyone uses GOJEK, they all complained to me whenever something goes wrong, right. So if I get two or three complaints about a particular bad experience. That is related to the system. Yes. Like some kind of a bug or system breakdown. If I get two or three separate complaints on that day by my immediate friends, yes, I can already roughly calculate like within a very large range that error rate of how many users are experiencing this error. Yes. Okay. So there's also this interesting thing of that if the probability of you encountering that issue with just a few people is you're hitting it again and again and again. It's a big problem.

Aakash: Yes. So, right. Okay. Let's define this space to take away from the qualitative versus quantitative aspects, right? Uh, both type of study is critical because the quantitative aspect will always tell you what's happening and in what proportion it's happening. It can never you why it does happen.

Nadiem: Yes. The 'why' is never there.

Aakash: The 'why' never, it'll never be there. You in qualitative aspect, you'll never get the statistical rigor so you don't really know in what proportion this is happening and you are getting the opinions of people as to why they think they do what they do. The reason I'm making such a vague statement of why they think what they do is in B2B space, the businessmen, really think about their business, the quality of how much thought they put into it might differ but certainly far more than how people think of the quality of commute can be improved, right? So but you have the qualitative and the quantitative. So the way you want to use both of these is let's assume that you hear these 3 complaints about a particular bug or particular issue on the system. The best way to deal with that is go back to the quantitative analysis. See, hey, this is the particular problem. What is the signal that I can find in my data to understand what proportion of the overall user business facing this particular problem experiences as a problem. So do use this as a signal, but you cannot really have a final confutation of how big a problem this is. Treat that as a signal. And that is all it is, don't ignore it.

Niranjan: Specifically when it comes to friends and family. Right. It becomes really interesting because they are going to be most critical. So the data that you get from them, obviously they're going to be like, you know, even the smallest thing they're going to raise. Some other customer might not even talk about it. So that, that information definitely has high value.

Nadiem: Yeah. Because they are within your proximity. Yes. They're more open to share it. And you know, they're just, they love to annoy you.

Niranjan: Yeah. Right. Exactly.

Nadiem: And you know, there is this kind of epic issue in all kinds of product fit analysis of correlation versus causation. This is not restricted to product market fit assessments. This is true with all scientific findings in the modern world. There's a huge debate in medicine around what we thought was causative but turned out to be correlative. So this is an epic scientific debate of all time. Right. And what I've discovered personally as a founder is that you absolutely need qualitative and quantitative, but qualitative research has far more insight into causation. But it's very difficult to quantify the size of that issue or problem. Quantitative data defined in product market fit as getting your product out there. And then seeing how users interact with it or large kind of rigid surveys, et Cetera for me it's the opposite. They are extremely statistically significant. You can, you can identify the magnitude of the problem, but it tells you very little about the causation.

Niranjan: Yes. And the impact.

Nadiem: You just know it happened. You don't know what happened after that. You don't know what caused it. But, but qualitative goes much closer to the root cause of a problem. So as your product scales up, you know, the importance of qualitative in the beginning is an extreme high. And then as it scales up, quantitative data catches up and then you have to just maintain the balance of each. So what happens now in our product teams, I mean, just to illustrate what audience go, what we do is we usually identify issues through the quantitative data. Right. We see there's something here. This is really interesting. I don't know why it's happening now the instinct and these are smart people instincts. Okay. The instinct of smart people is to immediately associate, oh, obviously that's why. I have been the worst actor in this, in the early history of GOJEK. Because I thought, you know, because in may ways my intuition was correct. So I just kept thinking I would get it right again and again and again and again. Right. If you get a few things right, you just assume you're going to be right forever. Right. Um, and I began to notice that as our product scaled and as complexity increased, my rate of intuition, accuracy went down.

Niranjan: Yes. So I remember the early days of GOJEK right, after GO CAR launched. It's like,GO CAR orders are going up and GO RIDE orders are going down. It's raining. And we were right. More often than not.

Nadiem: That's right. I knew, I knew exactly what it was. I knew when it was hard rain or light rain. Oh, the curved shape is like this. Therefore it's raining only in some parts, right? Oh, and then, oh look, look how area drop. Definitely flooding. We have flooding guys I can tell, you know, and it was impressive for awhile until it stopped, right. And that's the instinct, uh of, I'm not talking about dumb people but of a very competent people all over the organization. It's a natural human instinct to jump to that solution.

Aakash: But it really boils down to discipline. I mean ...

Nadiem: or process. If we relied on discipline, personal discipline of everyone in GOJEK, we'd be screwed by now because how much, how much can you depend on just personal discipline as opposed to process or peer review culture.

Aakash: What I mean by discipline was that whenever you have the first instinct, uh, don't jump to the conclusion, follow a certain process, be patient, look for more signals, look for more markers. And then the discipline of following the process,

Nadiem: the pause, I call it the pause. Right. It's uh, it's the same pause that you have to do, um, just when you're about to like fight with someone, maybe a family member or your spouse, right. There was this already this almost like your lizard brain about to jump and like yell or say something you'll regret later on. And the key is just to pause. Notice that it's happening and then reassess. Right. That's the most important thing. And when you see some data and it's like so exciting, right. And you don't have that moment of pause and say, all right, I think it's this, there's a hypothesis. That's okay. It's my hypothesis. Now I need to go and ask some people and do the qualitative stuff. Right. Or launch my research team to actually validate whether that hypothesis was true.

Aakash: So what you actually mentioned about the qualitative versus quantitative analysis. I think they both have the same pitfalls. Uh, even in qualitative analysis, just because you are getting the 'why' immediately answered by people doesn't mean that they are right. It's a very common thing apart from the leading service, like leading questions, this and that. People have a tendency to please the interviewer is people have a tendency to appear smart in the interviews. People have tendency to just make things up. Uh, I mean a lot of literature everywhere mentions about this but even around uh, police interrogations, but police interviews off, um, eyewitnesses, right talk that if you are friendly to the eyewitness you tend to get a lot more incorrect information because they just want to ensure that you get some things so they will make things up. Whereas if you are very professional and you're a little stern, you will get little information but there will be accurate about the information. That is such a subtle things to get right and honestly I'm not that good with people so I tend to depend on the data side. But these are the problems of the qualitative aspect. So over there, although you are starting with "why", it can still be wrong. Similarly when you are starting on the quantitative analysis, sure you don't have you, there is a high probability of you getting the correlations at all, but at least you can then start looking for different, different, different, different signals to support your hypothesis. The best way is always to have both quantitative and qualitative analysis that agree with each other. The most accurate decisions you can ever make is to have a quantitative analysis. I mean you can start from either way but both the analysis need to agree on what is happening in what proportion, what is the impact of it and why it might be happening because even the "why" part you can find in data. When someone says that, oh if I don't get the ride in 30 seconds, you know what I just go to something else. You can find the correlation in the data, right? So try to marry, the quantitative and the qualitative part. That is when you know that you are not fooling yourself.

Niranjan: So data is the skin of it right. You keep saying data comes with two statisticians. They would come up with a completely two different conclusions and argue about it. And there'll be confident because they can say that data proves this. This is not me. This is data.

Nadiem: It's like a, it's like a white wash, right? It's, it's like a blanket statement. "This is data." I cannot tell you or the audience enough how many times data has lied to me. Like it happens even more so than humans. Right? Then that's why the cleanliness of the data and the sanity and the scientific nature of by how you approach that, it's so important. And that's why one of our core values is be a scientist, right? Be a scientist. What does that mean? That basically means just be a skeptic, right? Just be an intense skeptic. Just be methodical. Just be skeptic. Be informed to be informed and keep iterating and validating what you know. Now I want to introduce this other concept of, of when, when a lot of product marketers or product road specialists, uh, our product managers are in the growth mode, right? Let's say you've discovered product fit and or, or you think you discovered product fit, but from day zero you're already subsidizing that product with promos, discounts, all kinds of incentives. Into that program. Like this is, you know, and it's very hard because in high growth companies, et Cetera, your natural instinct is always to promote things. Especially if you've raised funding, right? What are you going to do?

Aakash: You're optimizing for growth? You're not optimizing for product market fit instinctively.

Nadiem: Yeah, you've got to spend the money. You just raised a bunch of money. Where are you going to spend it on you're spend it on to grow your product, right.

Niranjan: Nobody ever sees that growth is bad.

Nadiem: Yea, growth can never be bad. Right? Growth is the end goal. The end being of of, of what we're all doing. But is it, so the problem is when you bring exogenous incentives into the equation, that brings noise and noise to the extent that will create overexcitement potentially almost all the time. Like over excitement. When it's noise, it's the money that's causing that product to grow instead of inherent product market fit. So how do you balance that? How do you balance the fact that, you know, we need to grow our products, but at the same time running experiments in a subsidy or promotions rich environment almost negates your data in, in a very convoluted way.

Aakash: I mean, if, if I can be very simplistic, I would say that do not use subsidy when you are trying to find product market fit, right? Because if you aren't using subsidy to find product, market fit it's equivalent to saying that, hey, I have this particular coffee, I want you to drink it and I'm going to pay you to drink it. I cannot really say it's a good coffee because many people are drinking. So I mean, that's really bad. Like you can never, you shouldn't have already to do that. You use money for growth to get new customers, to try the product, uh, to actually deal with some shortcomings of the product. Because any product in the MVP stage, when you're just prude your product market fit, of course it's not a polished product. So there are friction points. You pay people to overcome those friction points and still adopt the product in a way, but not really for product market fit.

Nadiem: So if you want to determine the product market fit of this cup of coffee first, you can't give the coffee for free. You gotta charge the same amount as an average of any other coffee? And say, hey, try my coffee. Hopefully if you get someone to try it first of all, then they try it. Oh, okay. Give me another. Okay. That's product market fit. That's thing repeated again and again. It's product market fit.

Aakash: Give me more. They're calling their friends there's this amazing coffee. I mean we started a new coffee shop on 5th floor over here. I just learned about it day before yesterday and we have something called Cocochino, which is like a coffee with coconut milk. Uh, Ditto pull me over there and said hey, why don't you try this coffee right after that, I sit over there, why don't you try this coffee behind this guy coming in for interview? I'm just spend more for lunch or before he came here I'm like, Hey, you like coffee, why don't you go and try that one? That's product market fit, right?

Nadiem: That's right. And that's why NPS scores are a very powerful signal. Uh, or Net Promoter Score for the audience that doesn't know what net promoter score, it's a score that determines your propensity to, uh, speak highly of a product to your friend or family.

Aakash: Out of all the users you have, how many are going to promote it versus how many are going to detract.

Niranjan: So the extreme case of this, right? This, Oh, I'm running out of coffee beans, so I'm going to increase the price of this coffee and still people are coming. And we had that in GOJEK in the early days.

Nadiem: That's right. When we hit real product fit that we were, we couldn't handle demand so much that we increased the price, to try to taper the demand down. And yet we were still under a full utilization. Right. And that's when you know you've hit great product market fit.

Niranjan: Even beyond that, like I remember seeing this where we essentially increase the price in the morning rush hour and people started going to office early.

Nadiem: Yes. And we identified that you know it initially the early adopters were using our motorcycle taxi service for for two main reasons. One, the fact that it eliminated the problem of anxiety of getting a ride before they didn't know if they were going to get one. The second part was reducing the hassle of negotiating the fee. And that was such a big market fit that in a period of rising price or even time based pricing we didn't have surge pricing. Then we just, we have like for this time make the price higher. Right. Um, that people would change their behavior and leave earlier just to catch that promo. So that's what happens when you have powerful product market fit. There is some signalling of price insensitivity. There is no initial product. There's a high MPS, the virality. A factor is high and that's when you know that you are on the brink of market creation. Right? There's, there's one thing in trying to substitute some behavior. Uh, so for example, before we thought the only people that would take GOJEK GO RIDE - the motorcycle ride hailing, is people who already took motorcycle taxis before. They just did it manually without an APP. And without professionalized drivers. Turns out today, you know, the fast majority of people who are taking, uh, GOJEK had never taken a motorcycle taxis before, right? And that's where you and the vast - by far, the vast majority of people who use GO FOOD are people that never use a food delivery service before because the market was just so much more. That's what happens. That's what, that's the action or the mechanism that is taking shape when you will then expect a five to six, six X increase in the market itself because of your product, which happened,

Aakash: This particular aspect is so beautiful. Uh, specifically in today's trend. This is not so much about product or GOJEK as such. The world itself is going through changes where constantly people are creating new markets, creating new behaviors and your analysis of whether that particular behavior will stick or not or whether this particular product will like what is the time of this particular market cannot be studied based on the current showcased behavior. You now have to actually go one level deeper and look at what are the motivating factors, what are the market forces which are driving something and then extrapolate that if you reduce this particular friction, what were the time be? Right. So with the constantly changing ecosystems and technology making all your impossible possible, your analysis of time is fundamentally changing. It's no longer looking at competitor sensing that, oh, all these people together. This is what happens. Like now when you look at uh, ride hailing markets like the way, I think about it is the population that is there, the condition levels are out there, the GDP that is there. You don't look at what the current mode of transportation is because you're like, hey, this is so amazing that I know for a fact that when I put the service over there it has to work. As long as these supporting factors exist and those supporting factors are not competition.

Nadiem: Yes. And I guess that's the most powerful signal. but kind of going back by the point about money and about subsidy and spending. How does one actually reconcile when it's the right time to spend for something? So let's go back to the coffee analogy.

Aakash: My time to spend to find product market fit.

Nadiem: Yes. So let's take the coffee example. Okay. If I sip it, I bought it at normal price. Raise the price I buy. I've discovered a product market fare. Now let's say it's not coffee anymore. Let's say. It is a new cognitive improvement power drink. Okay? It is a power drink that improves your memory and you have to drink it every day for it to have some affect. Okay. I'm just making this up. So what's different about this product? If I just charge the same amount as a cup of coffee for that product? Yes. And say you want to buy it? I mean if you're asking me, I would say like screw you. Like that's the most ridiculous assertion. I don't believe you, first of all. That it is, I don't know what's in it. I feel unsafe about it. I've never had this kind of drink before. No way. Now when your product requires a significant amount of change in consumer behavior. If that decision point and hurdle is so much bigger instead of something mundane that you do, which is having coffee every day. That's where promotional incentives are a lot more important.

Aakash: The amount of behavior you want to change in your consumer is always in proportion to the incentives you are to give. And the insentives are not always monetary. So which is why just call it incentive other than money or promotions. Uh, the incentive can be social pressure. The incentive can be, uh, some sort of a status. The incentive can be how good they feel about themselves. You can use any incentives that you want. Money of course, is the most explicit that you can use it in, which is why it's easy.

Nadiem: It's the fastest.

Aakash: It's the fastest, right?

Nadiem: It's a hack

Aakash: It's wrong. Uh, but yes, the amount of behavior change you want in people is going to be proportional to the incentives that you have to provide. So for example, when you're doing ride hailing, people are still travelling, it's not that they choose to travel or not to travel depending on your substance, right? So, uh, which is why in the early days of GOJEK, even if you raise prices, they were still using it. But take GO PAY for example. Now we are trying to convert the cash rich economy, a cash heavy economy into a digital economy. We are asking people to change the way they, behave on a daily basis and build up habits. It's one thing to ask them to change it for one day. It's another to ask them to build habits, right? Like study shows that building habit takes 18 months of constant activity. That's when sort of the habit goes some where in the deeper part of your brain where it's just your second nature. Uh, to ask someone to change how the live for 18 months of their life is not going to be cheap. That's not going to be easy. Uh, although creates another question at that point that if you're going to pay me to do something, how do you know if you're going to get product market fit.

Nadiem: And different behaviors have different, what we call it, the threshold or the A-ha moment. Um, for example, food could be like X amount of times in a month builds habit for X number of months. For GO CAR. It could be a X number of trips or X percentage of trips in a month that you take this, then it develops habit. Right. This is how you, every single product has different A-ha moments by which habit is built and then it becomes a learned behavior. And then it dependency. And then some form of ways it could go to the extreme side, which is a bit negative, which is addiction. Right. That's something that we need to call out explicitly. Right. And so there is, uh, the best product managers are constantly trying to triangulate or isolate those A-ha moments. One point, how far do I need to incentivise person before I can turn off the gas? And lock that user in. Right. You know, one of the most beautiful things we've noticed is that a GOJEK as a Super App has this A-ha moment called the Golden Triangle. Whereby as soon as people, a user uses three different services in a month, uh, and they're on GO PAY, they're pretty much our user for life. They don't go anywhere. It becomes their home base. Right. And this is a really interesting phenomenon that can, that is revealed through a product analysis revealed through user behavior, revealed through data, but then informs the entire spending strategy of the organization. That you're in. So product market fit is not just the principles and frameworks and skepticism that you need to analyze the data are not just important for product managers, product marketers, et cetera. It's extremely important for the entire identity and goal of the company itself.

Aakash: Your OKR should really reflect which state yourproduct is in. If not, you're doing something really wrong. Going back to the product market fit plus subsidy angle, right. One way of identifying whether you have found a product market fit in a subsidy rich environment is to actually reduce the subsidies. You may not want to do that for the whole market and that's where the experimentation comes in, so select your segment very carefully. But if you're able to reduce subsidies over there and see what the change in behavior is. Is the change in behavior proportional to the reduction in subsidies is a very good signal to tell you if their habits are being built or not. Although you can afford to fully take it down in the face of competition probably, right. But yeah you still need to make sure if you don't do that you are in a world of trouble\.

Nadiem: But you know, there are so many cases that I've heard of people and sometimes in GOJEK, but not as often but in a lot of other companies whereby there is such a huge psychological fear of taking down subsidies to see and to assess that people just continue the lie. They keep continuing. Like there's like, I don't want to go out of this dream.

Aakash: That is equivalent to me not wanting to go to a doctor for a checkup because I'm not sure what my blood pressure is going to be.

Nadiem: That's such a perfect analogy.

Niranjan: I rest my case.

Nadiem: They don't want to know the truth. Right. And the less secure product managers will consistently create reasons and excuses.

Aakash: What is important? Do you want growth or do you want truth.

Nadiem: That's hard. That's about becoming a scientist, but a high performing company culture, should encourage those kinds of breaks on spend to be able to crosscheck is this spend retentive? Are these users actually using it again, right. Without a financial incentive and without an ingrained culture that doesn't reprimand the product teams for the drop in growth but already expecting it and celebrate it as a good thing. Okay, great. Now we know now we know what's real and what's not and in this crazy roller coaster of raise money, burn money, a game that is played in high growth tech companies today in the emerging world and in the developed world it's still the same. It takes a lot of courage to kind of institutionalise that experimenting with subsidy, right? And, and telling everyone it's okay, we know your numbers are going to drop. Like how many times have we actually comfort a teams and saying, hey, it's okay if it crashes both from a spend or even from a reliability perspective. It's okay if the system goes down, we have backups, et cetera. It's okay. We need to know right. it's the same thing with spending.

Aakash: You just touched on something very interesting, but I don't think a lot of people realize that. It took me awhile to realize that, uh, when you add subsidies to a mix. Sure. The companies are very different beast and your expenses are very large and then because of that you need to be a lot more a giant in how you manage things. That's all that is fine. But really the impact that it has in the way product manager thinks of his product fundamentally changes in compared to a non subsidy product.

Nadiem: Yes. How?

Aakash: It is not just one more dimension that being added. It skews everything because it's noise. It's not adding a dimension. It's adding noise. We have been talking about actually that I just want to call out explicitly that dealing with a subsidy, uh, products, subsidized products versus non subsidized products. The product manager's job is literally order of magnitude more difficult because of how difficult it is to find out the truth. If I'm not paying my customers, everything that my customers do is much more 'honest', in a way. I'm going to use that terminology. The market feedback to me is much more organic because I'm literally executing the business as if it's a sustainable business from day one. Whereas if I'm not doing that, if I'm actually spending subsidies, everything has a lot of noise, the market signals are skewed. Uh, we not only have to get to neutral, where to figure out how to be profitable. Many a times the product that you are subsidizing is not really the one that is going to make your money. You are actually going to build on something else. It's a very complex beast to build. It is not equal to any other product.

Niranjan: Because longterm reality is those subsidies are going to go away. It has to be a sustainable business.

Aakash: So how do you keep optimizing or building your business for three, four years without removing subsidies? At the same time, you need to know that you're course correcting in a direction of sustainability.

Nadiem: What's ironic is that the further away you are from the truth, the easier it is to get all the glory, right?

Niranjan: So many startups we know have failed on this fundamental thing. Right? Like they were able to show growth. They were able to raise money, but as money dried out, they had nothing to show.

Aakash: To call out nuance. It is similar to how a lot of startups spend for or customer acquisition in the early days, uh, using just to spending on marketing. I mean spending subsidy for per transaction is the most extreme form of marketing spent for acquisition. It's extreme, it's like orders of magnitude more extreme but yeah, it's the same problem.

Nadiem: Let's go, let's go back to that framework, right? So we have to identify product market fit. Then grow the product. When you've established this, go spend some money.

Aakash: So let's assume that we have found a product market fit with the people are like adequate to your product.

Nadiem: NPS is high and you spend a bunch of money to let people know about your product. Okay and convert people into your product.

Aakash: But like then the key part is like how do you really spend, because many a times you are going to find product market fit on a smaller segment of the market. Let's assume that you get a 10,000 users to really be loyal customers of GOJEK and those people come from different, different backgrounds, right? Different economic struggles. I think that's really good to know that that is product market fit. But now you want to expand how you want to capture rest 99% of the business, our market. And that is what have you call the growth phase.

Nadiem: That's right. And this is the tricky part, like in GO FOOD. An example of this isn't in GO FOOD. We were growing super fast and we thought we were addressing, uh, we were capturing the entire food spend of everyone in Indonesia. Right? We thought that was the pie. And then the data keeps growing, growing, growing, growing. And then we realize you, you hit certain plateaus whereby what we then realize that, oh, wait a minute, we are in a particularly middle high economic segment. Yes. That was the market that we were actually, market problem that we were addressing. Oh, but that's only a subset of that market. Yeah. The larger market must encapsulate the lower levels of economy. And thus it was that realization in order to capture the bigger side of the market, we realized that we had to go down market in terms of cheaper, uh, cheaper SKUs, right? Things that are more budget sensitives to be able to..

Aakash: Delivery fee needs to be lowered.

Nadiem: Exactly. So at that point of stagnation, we unleashed it more when we realize that, oh, actually because everything is black and dark around here, right? It's like a Warcraft when you're first starting out.

Aakash: Well, for GOJEK or GO FOOD the segmentation is always much more difficult. So whenever you have your whole population as a market segmentation is an incredibly difficult problem, right? Uh, but the best way to think about this is if you have a larger picture, which is split into multiple squares, let's say each square representing one segment your product will almost always start finding product market fit for one segment. Then you get into growth phase, you fill that particular square fully and then you're like, oh, I'm not seeing growth. And then you find the next square, but your current solution does not really work. And the solution is not necessarily features. It is pricing incentive. It is positioning. It can be anything.

Nadiem: If you don't do anything, it won't move out of that one square.

Aakash: Exactly, it will just get stuck.

Nadiem: Right. It will get stuck. And this data will tell you that the cost of spend, the return on incremental transactions will go up. Yeah. Right. That's a very important metric that people need to try.

Aakash: Your ROI on your acquisition will keep topping constantly.

Nadiem: So your ROI goes up, your cost per incremental transaction goes up, right? Your ROI goes down. And that's when you know, okay, you need to hit that next.

Aakash: You need to get to the new segment. Like that can be different geographies. That can be different...

Nadiem: But a segment does not necessarily mean like the example in GO FOOD that I highlighted. That's a different economic segment. For example, a segment could be within the same economic class, but a different problem set. Like pick up food, I'm just hypothetical now. Not the people that need to get food at their location brought to them, but they just want to pick it up. That's an example of an expansion.

Aakash: There's a more nuance one. So either you can increase by geography. You're going to expand, right? So we were talking about how do you get growth, right. So the way you get growth is by moving into different geographies or you get growth by addressing a different customer segment altogether. Or like which are different people in the same geography, let's say. Or for the same customer, you increase the number of use cases that let's say people who are using it for dinner, what do we need to do to make sure that they use it for lunch as well? Because the constraint for lunch delivery versus a dinner delivery is so different. A lot of people who use GO FOOD for lunch, they are actually in offices. They need to know that the food is going to be delivered exactly in this particular time. They are generally doing orders for an individual, whereas the dinner orders tends to be for groups. So for the same person who is a loyal dinner GO FOOD user might actually never use GO FOOD for lunch and always prefers to walk down.

Nadiem: And how did we figure that out? Because we hit the limits GO FOOD was growing but only for dinner. And we a disproportion amount of growth for lunch and that's when we pivoted towards that lunchtime and try to build that behavior through shorter delivery time. Right. You need your food fast, it needs to be more affordable, et cetera. So it's faster turnover and so on. And then it could unleash the next phase of growth.

Aakash: And in a way we didn't change any of the features, right. What we changed was the pricing. What we actually changed was the selection of restaurants that you see in the morning versus the evening. Uh, those are the things that we selected. It wasn't really any features that we added and that allowed us to crack one more customer segment. One more need for the same customer segment. So yeah, the growth comes from all aspects and you need to make sure that you are not blind to one of the damages because in a way moving into a separate geography or acquiring a different customer segment might be far more expensive than deepening your engagement into the existing customer segment.

Nadiem: I guess as the last part of this podcast, I really want it to go down to the PM itself and what you guys have seen to be, uh, the worst characteristics of like say poor PM's and the best characteristics of the best PMs. PMs are product managers and a pivotal part of every tech company in the world. They're the ones who actually, uh, develop the product in conjunction with engineering teams. They do a little bit of project management, they do a little bit of research, they do a little bit of marketing, they do a little bit of optimization as well and analytics, right? So they are that owner of that product. And a product can be something as big as GO FOOD, which is a master product, right? It can be something as small as the buy button or best sellers within GO FOOD.

Aakash: So I think of product managers as someone who own a problem. Whether it's a GO FOOD, whether it's just a search in GO FOOD, whether it's just a buy button over there, or there is a shuffling card over there. It makes no difference. You are owning a problems space. It needs to be well defined problem space. Sure as you grow, you take up larger and larger and larger problems basis, uh, the good and bad characters on that.

Nadiem: Maybe start, maybe it's easier just to focus on the best product managers, you know, do this. What is this? I think bad is too easy to define.

Aakash: I think the best product managers actually get the teams to solve the problem. And I don't mean this in a, you know...

Nadiem: instead of themselves.

Aakash: So if you are saying that you're a problem solver, right. I consider a design engineering, UX, QA, research, finance, like depending on the domain expertise that are there. All of these are domain expertise, which are just part of the solution. So if you have a particular problem, let's say that your problem is people are not able to find a feature, people are not using a particular feature. You can solve that using design. You can solve that using engineering, you can solve that using incentives. You can actually solve that by using user research to understand why people are not using it. Is it the problem or discovery or is it the problem of a new need being there? Now, if as a product manager, I try to solve these problems, the probability of me solving that problem correctly, consistently in a more and more complex space is nil. Right? I can do that in the intuitive ways probably. But the job of a product manager is to ensure that all the relevant information is available to all these people. All the people are coming to the table with the right intention to solve that particular problem. Have no egos coming in between are able to exchange ideas and collaborate with each other.

Nadiem: But why does that sound like a CEO's job? It sounds like you just described as CEO job.

Aakash: It is, but it's much more focused in a way, like they are not focused like, so for a product manager the aspect of our building is much lower in compared to what it is for a CEO. They're not very worried about my operations, they're not worried about sales. They're not worried about HR. They're not worried about finance. They're not worried about budget projections. They're not worried about fundraising.

Nadiem: Should they be?

Aakash: Uh, I think it's an overkill. I mean, sure. At a certain seniority, yes, they should be. Uh, but yeah, I mean, I would say at that point they're not really product managers. Like product managers manages a product. Not a company. CEO manages a company, but it's problem solving. So the best product managers bring different, different disciplines together and that group comes up with a solution which blows the product manager's mind.

Nadiem: Awesome. What in your mind is the most important characteristic of the star PM's?

Niranjan: So I would say one of the most important thing is product managers being aware that they are not solving for their need, they are solving for customers. And then having that ability to build empathy with that customer segment and figuring out what that customer really, really wants and then serving those needs, that's very important for that the product manager needs to be self critical. Like they need to be aware that oh my first instinct is going to be wrong. Yeah. And be okay with that.

Nadiem: Be okay with that and not insecure about it.

Aakash: Exactly. You're trying to empathize with a completely new to me and you are trying to do that as fast as possible, which means the, you are going to make an incredibly large amount of mistakes and all of your existing mental models bias as for example, when you look at incentives of drivers or incentives of merchants, they are more similar to a human biases or instincts of a product manager and an engineer, right? So trying to empathize with that fundamentally means you have should be able to say that the I think is not the way that person thinks and that's okay.

Niranjan: And the best product manager I would say. Is domain agnostic? It doesn't matter what domain it is, like obviously being familiar with the domain helps you form those ..., experience does matter, but if you take someone.

Nadiem: What do you mean? Like domain agnostic. What's a domain?

Niranjan: Like? Food is a domain.

Nadiem: Yeah. You can move that person. The best ones you can move them into PAY. You can move them into TRANSPORT and they would perform just as well.

Aakash: I mean they'll take time to build up things, but they are not, they don't say that I'm a PAY product manager. Like if someone says that, oh, I manage only payment, that fundamentally means they are giving too much credence to their understanding of the payment space in compared to the reality of the market. I think what happens with the market changes will they change? Will their opinions change when they move from one geography, the geography, they're trying to impose their understanding of payments space on a completely new country? So that, I think that's, yeah,

Niranjan: That's what I meant.

Nadiem: For me, I think the most important characteristic of the best product managers that I know are those that are willing to defend the customer to the bare end. Right? The user, whoever the user is, the user can be the end user. For us it can be merchants. It can be drivers, whoever that user is.

Aakash: He should be the user's voice at the table?

Nadiem: Yes. And willing and having the courage to fight for the customer to the bitter end. That's, that would be my criteria.

Niranjan: I would say that is applicable to every person in the product? From different aspect. Like designer has to do the same thing.

Nadiem: Yes. But I think that the product manager needs to be the one. They need to be the guardian of that. Shield the designers, the engineers, all of the business analysts, et Cetera. They need to shield them away from whatever whims the leadership team might have about, oh, we should go here, oh but the business needs to go here. No, the best product managers I know, shield their product from business needs. Defends the customer to the bitter end.

Niranjan: I will say this both ways because even engineers come up with like, oh, we need to build this for our customers. Yeah, right.

Nadiem: No, that's very true. Guys, thank you so much. It's been a very insightful discussion about product. A lot of controversial comments made. A lot of people I know will disagree with us, but as always, it's been exciting and an honor. Thanks a lot. See you until the batch.

Aakash: Yes. Thank you.

Nadiem: All right.

Outro: Hey guys, hope you enjoyed the podcast. If you liked it, please hit like, subscribe and follow us on social media. Thanks so much for tuning in.

See Full Video