[00:00:00] Aishwarya Goel: So we created these three assumptions. Then we created a very mathematical approach. If we get these many greens which means yes and then out of red, so that means there is a play out here, right? So I think that really helped us be very focused, be very mathematical about it and at the same time, really get right insights rather than just, like you know, fluffy conversations.

[00:00:25] VO: How do you build a habit-forming product? Do you need to invest… We saw these different… You don't just gun for revenue in the door.

[00:00:32] Andrew Michael: This is churn.fm, the podcast for subscription economy pros. Each week we hear how the world's fastest growing companies are tackling churn and using retention to fuel their growth.

[00:00:44] VO: How do you build a habit forming product? We crossed over that magic threshold to negative churn. You need to invest in customer success. It always comes down to retention and engagement. Completely bootstrapped, profitable and growing.

[00:00:57] Andrew Michael: Strategies, tactics and ideas brought together to help your business thrive in the subscription economy. I'm your host, Andrew Michael, and here's today's episode.

[00:01:08] Andrew Michael: Hey, Aishwarya, welcome to the show.

[00:01:11] Aishwarya Goel: Thank you for having me, Andrew.

[00:01:13]  Andrew Michael: It's great to have you. For the listeners. Aishwarya is the CEO and cofounder of Infirless, building the world's most reliable and fastest serverless GPU inference for AI/ML companies and recently backed by Sequoia. Aishwarya iterated on four ideas in 12 weeks before landing on Inferless and was previously the cofounder and CEO of Peak Performer, where she scaled the service to 900K ARR before deciding to pivot the business to Infirless after failing to reach product market fit. My first question for you today, Aishwarya, is if you were explaining Infirless to a five year old, how would you do it?

[00:01:51] Aishwarya Goel: Okay, quite interesting. So I think, in the most simplistic manner, think of Infirless as a place where you host your services once you have, let's say, trained your model. And what is the biggest value, prop, and where we come in from a solution perspective, is that if you are at a very early stage in your product building phase and you know that your usage is very less and you're... I mean you're growing gradually, that is when you can just leverage Inferless as a service to only pay when it is used, at the same time, provide the best experience to your customers and scale without the hassle of managing infra. So it's more like more scalable fashion and without any hassle of hiding team members or you know building your infrastructure yourself. You can just give it to us and we'll manage it for you.

[00:02:50] Andrew Michael: Nice. If I was a five year old, I'd probably have another hundred questions off the back of that. I think it's a good, simple way to explain it as well. So basically, for AI and ML companies, you give them a way to be able to host and leverage their models without having to pay continuously for expensive computers and machinery to run those models. So you give them a way to pay for what you use. Basically, pay as you go.

00:03:12  Aishwarya Goel: Correct. And even for bigger companies, because generally, I mean, as you grow big, right, and that is when you need to solve the problems around auto scaling. So that is also where the user part.

[00:03:25] Andrew Michael: And yeah. Nice. So today I'm going to be talking a little bit about this journey to getting to product market fit. Obviously, I mentioned at the start, having started Peak Performer scaled the business to 900k in ARR, it's almost a million in ARR Like it's sort of that point where people say you've reached some level of product market fit. If you can hit that size and scale, then to deciding it wasn't right and new to pivot, so maybe let's just get started with that. What was Peak Performer and why did you feel at that point in time that it was time to pivot and move on to something else?

[00:03:59] Aishwarya Goel: Quite an interesting question. So, in a nutshell, at Peak Performer, what we were building was so we were trying to democratize leadership coaching for early to mid managers. It started with my own personal pain points that I felt as an early manager in my life. So, again, I think we ran it for close to three years and the reason we decided to pivot even after, at a certain scale of ARR and with a decent product, 30 plus customers, I think one learning that I have personally experienced, and that happened, like when we were going through our surge accelerator, which is by Sequoia, now Peak XV Ventures, was, like you know, I mean, the journey of building a company with a weak PMF, so like really understanding the meaning of what high quality product market fit means, and that really helped us go very deep into understanding the right metrics that we should focus on.

[00:04:54] Aishwarya Goel: So, from a SaaS perspective, our product was... we used to offer these offerings to companies where they can provide this to their leadership and managers, and there were certain metrics that we were chasing to understand from a like a viable business perspective. And there were a couple of reasons which made us feel that maybe it's like something we need to figure out whether there is a strong product market fit that exists or not. So one was, of course, retention. That was the holy grail, and there were a few reasons where we couldn't achieve that. One was the type of business, so when we started working on this idea.

[00:05:33] Aishwarya Goel: So again, we were trying to build like an Asia version of better up or like any such players where we were democratizing coaching. And what we learned about the Asian market was that maybe the companies here are not ready or, let's say, it's not like a productized or a fixed budget that they actually invested. So it was a lot of hard selling rather than something which already, like there's a budget that exists versus how it operates in US. So I think we didn't do enough discovery properly at the beginning to realize that it's going to be more of like a lifestyle or, let's say, a one time investment, rather than like a fixed cost that they actually invest in.

[00:06:12] Aishwarya Goel: So as a nature of business, it was not like a pure SaaS which would kind of fit into that. The second, and which is more related to the solution that we came up with, that coaching as a concept has a natural turn of after 6 to 12 months. So let's say, once you start your leadership coaching journey or any therapy journey for that matter, you take some pauses.

[00:06:35] Andrew Michael: The graduation journey.

[00:06:37] Aishwarya Goel: Yeah. So which was kind of against the philosophy of the SaaS offering that we were building. We wanted people to do it for a longer time.  And people were like, okay, we love your product, we love the coaches, we had high MPS. All the matrix that we save, but at the end of the day they were like, okay, I want to now first go and implement this goal and then come back. So I think these were two major reasons which made us feel that we are trying to focus on a particular product and kind of trying to implement a playbook which could have been a success in the West but doesn't mean it's going to be implemented in the same manner here.

[00:07:15] Aishwarya Goel: And then the only way we could have then built that business was to make it a consumer business, which we as founders didn't want to get into because our background comes from enterprise and building engineering products. So hence we decided that okay, I mean, whatever we are left with is not something which we feel very excited about and may not be the best fit to even do that.

[00:07:37] Andrew Michael: As founders.

[00:07:38] Aishwarya Goel: Yeah, so I think that was like the high level.

[00:07:41] Andrew Michael: And just a little bit of context. Was this a bootstrap company? Had you raised any venture capital for it, or?

[00:07:48] Aishwarya Goel: Yeah, for the longest time, I was running it as a bootstrap, we just raised less than 100K as an angel, I think, after we raised our seed round from P15 is, after which we decided to pivot. So, yeah.

[00:08:00] Andrew Michael: It's interesting that it's only post that you decided to pivot and not before because I think typically maybe it's a lot easier when it's bootstrap for you just to say, okay, let's can this and let's go, it's this discussion. But it's cool to see that you sort of had investors as well, that sort of said go for it and figure it out.

[00:08:18] Aishwarya Goel: Yeah, we almost had a run of like four years.

[00:08:21] Andrew Michael: Yeah, so you could have just kept chugging away at it and I think that's... yeah. It doesn't... Four years is a long, long time and a big opportunity cost to lose on something that's you can see is not going the right direction. So you then iterated on four different ideas in 12 weeks before you landed on Inferless from there. So you started out, the business was doing decent, but you saw these big major flaws that got you to believe that it wasn't going to make it and was time now let's make a pivot. Starting with that discussion, how did that go? How did you discuss with the investors and say, hey, this is not working. We need to turn this around fast.

[00:08:54] Aishwarya Goel: Yeah, I think there are multiple parts to this question that you just asked, right? One was transitioning from what we were building to even thinking that you know. I mean, of course there is what's next, but before that, how do you really tap up something where you already have this much of revenue, right? So of course, we went through this phase of transitioning out and I think one thing which I personally feel that we did really well was around communication. So we were very transparent with our investors, with our team members. In fact, with our customers, coaches that you know. I mean, we tried and it's not working out. And we gave every step enough time rather than just rushing through it.

[00:09:33] Aishwarya Goel: First, we discussed with our team members, went to investors and so on. So I think it's always good to over communicate, then not communicate. I remember, when we were even thinking of different ideas, I used to buzz bunch of my investors and now when I look back, I actually think that they must think that I've gone crazy, because every day I had a new idea. So I think it's always good to let them know what's going on rather than come out as a surprise. So I think we gave it a good amount of time, which also then, when we actually started iterating on multiple ideas, gave us enough space to really focus without any baggage. So I think that is to answer your first question.

[00:10:11] Aishwarya Goel: Now, coming to understanding and iterating on multiple ideas, that was like a whole framework that we kind of build right, and I think everything started with one core thing, which we were super bullish and focused about that. This time, no matter whatever time it takes, we are going to do customer discovery really well. Because in hindsight, when I look back, I always feel that Peak Performer would have been a very different product if I really invested for three to six months, not building anything, not selling anything, just talking to customers, potential customers. So I think this was the core hypothesis of how we even thought of figuring out ideas.

[00:10:52] Aishwarya Goel: And the first step was to me and my cofounder, so we both come with very different kinds of backgrounds and experiences. And our first step was really writing the non-negotiables as a founder market fit, which is very important, because there is no point of wasting 6 weeks, 12 weeks figuring out an idea and then later on realizing that, okay, this is something that I may not be excited about. So why waste that energy? So we really wrote about what kind of problems you want to go after. What are the personas we are more excited about? What should be the GTM motion? What is the market that we want to go after? Like we were very serious about targeting US as a market. So I think these were certain things that we wrote on the day one that, okay, this should meet those criteria, right?

[00:11:40] Aishwarya Goel: And then we started writing down. We took some time to reflect that what are the problems that we have ourselves faced in the past, which we can relate with and deeply care about. So, since I come from sales background and my cofounder is a high core engineer, so we started reflecting on those pieces which we felt more close to and then we came up. So Inferless idea also happened, because when we were building Peak Performer, we were trying to also pivot into like an AI coach and my cofounder, Nilesh, tried multiple ways to deploy and we were like really frustrated that it felt like that we are living in the 90s when he was trying to deploy that model. So I think that was also somewhere like that experience helped us to figure out Inferless, but of course we didn't test that idea on day zero, it was somewhere in the back. So again, I think these were certain steps. I'm happy to go deeper, but I'll pause to be happy to take questions.

[00:12:36] Andrew Michael: Nice. Yeah, I really like parts of the framework we mentioned. First of all, like the founder fits, just sort of evaluating and taking what do we really want to do? What do we get excited about,  where do we want to go? I think it's super important to start there, because it's the worst place to get to, where you might even start seeing some traction, but you just don't feel it and you need to be feeling it for it to make it work and to stick it out when times are tough. So I like that approach. So you came together having these discussions sort of ironing out like what it is you want, what it is you don't want. You also said, like customer research, number one place to start this time really doing deep discovery. You came up with four different ideas like how did you go about then invalidating these ideas throughout before getting to Inferless?

[00:13:17] Aishwarya Goel: Yeah. So we had this entire two to three week framework where both of us we decided literally every single day what is the job that we have to do. So I think that really helped us be very focused and at the same time not lose confidence because again, as a founder who is just coming out of a pivot and doing this thing, again most of the times like very demotivating also. So I think that was writing down the process was really helpful. And then what the process was. So let's say, you know, I'll give you an example of an idea that we tried. In fact, I'll maybe pick Inferless because that will be more easy to relate with.

[00:13:55] Aishwarya Goel: So first step that we did was that to not have a solution offering or a solution mindset when you're thinking of a problem. I think it's very important because he has a behavior. There are certain things that we really need to keep in mind when we start a customer discovery process, and those behaviors are two things. One, to always think problem first from a customer perspective, because the moment you jump onto the solution you're always trying to find ways to fit that solution into a customer. But when you go, a problem first approach to a customer, you also figure out when they are willing to talk to you, that, okay, they actually relate to this problem. So I think this was one behavior that we made sure as part of the process of taking an idea.

[00:14:43] Aishwarya Goel: Second was always having the mindset to invalidate but not validate. Because the moment you start, you go on a call or you're having those hundreds of conversations, and you're trying to validate something, you get attached. You don't want to hear or know, but if you're always thinking that, okay, I want to make sure that this is not going to work and again, it's very hard to do that. When you go to a conversation and you have a mindset that, okay, I want to make sure that this is not going to work, you may get some kind of insight on it. I think these were two very like, I would say, behavioral approaches we took when we started doing these conversations.

[00:15:18] Aishwarya Goel: Now, coming back to the process. First, what we used to do was that we used to write assumptions. So let's take when we started working on Inferless, right, so we wrote that, okay, these are certain problems which we feel our users would be facing in this process, problems like not getting enough GPUs on time or, let's say, overpaying for the GPUs, getting bad performances. So now these are certain problems. Now we broke down those problems into assumptions, and assumptions are generally, in a way, facts. Or you try to craft certain simple facts that you believe in. If I talk about the problem that, okay, I end up overpaying for the GPUs, the assumption could be that machine learning engineers are only utilizing 50% of the GPUs. Now you have broken down that particular problem to an assumption, and then we used to then go and test those assumptions with the users by asking three very binary questions.

[00:16:20] Aishwarya Goel: First, do you relate with this assumption, do you have this unmet need or not, where you are only paying for 50% of the use? You are paying full, but your utilization is just 50%. So do you have this unmet need? The second thing we used to ask was that are you prevented from achieving something if you don't fix this problem today? So, for example, in your life you would always be complaining. Okay, I don't like to go to office on a daily basis, I just want to work from home. Now, this is a problem, but do I want to fix it? No, I'm happy. As a human, you can live with a bunch of problems and you just grim about and just move on. So I think it's very important to understand that. Are you prevented from achieving something that you really want to solve it?

[00:17:06] Aishwarya Goel: And third, which was very important to understand the urgency of that problem was to ask have you yourself tried solving for it or not? Because if they have successfully failed in solving something, that means that the problem was so intensive and urgent that they wanted to solve it themselves. So we created these three assumptions and then went to the person and used to ask nine objective questions for each is a three for each. Then we created a very mathematical approach that, okay, if we get these many greens, which means yes, and then out of red, so that means there is a play out here. So I think that really helped us be very focused, be very mathematical about it and, at the same time, really get right insights rather than just fluffy conversations.

[00:17:56] Andrew Michael: Yeah, no, I think it starts as well like from the very beginning, as you mentioned, just really focusing on that problem and not starting with solutions, because then you come to the conversation with way too many biases and I can't remember who the person is, but the quotes is also something along like obsess over the problem and not the solution and continue to reading there and I think I've been guilty of this in the past as well. We've obsessed more over the solution and not the problem itself, and then maybe even listening to customers' feedback too much, instead of going back to the original problem and say, okay, what do we try to solve? This problem hasn't changed, just our solution didn't meet those needs.

[00:18:35] Andrew Michael: But what you started to do is first of all validate. Okay, we believe that this is a problem, let's go and first validate now and see how big of a problem it is. We have a set of assumptions that we're making. We need to go now and validate those assumptions, and we have three very objective questions we can ask our users that are not based on their future opinions, but rather than their past behaviors and their experiences that they've had. So I love that approach.

[00:18:59] Andrew Michael: And then, as you mentioned, being able to then quantify that qualitative feedback to be able to say, okay, there is something here. So, with Inferless, you had your three assumptions you wanted to validate. You sort of got the yeses. You wanted to see at least a certain level. What was the next step then? So it was easy enough. Okay, like this is a bit of qualitative research. Now, how do you sort of sell this idea to your investors that this is the right way to go to the team, and what was the triggers that said, okay, yes, let's go for it.

[00:19:29] Aishwarya Goel: Yeah. So I think when we were doing this process, we actually narrowed down to two ideas. So there was one more which was very compelling and that was actually in the marketing, like marketing side of the thing. But the reason we picked this one was, again going back to the first question, which is founder market fit. So I think for us and again it's very important to convert your founder market fit non negotiables into certain measurable metrics. So, for example, for us there were certain things which we really wanted to make sure that is there part of the decision making.

[00:20:04] Aishwarya Goel: One was that we wanted to solve a really hard engineering problem. And the two ideas that we narrowed down Inferless felt is something that if we are able to crack it, of course there is a huge if at that time, but we will actually innovate at certain very fundamental level. So that really excited us and, of course, we had the right expertise also to solve it internally as a founding partner. The second part was also to look at personas. In between these two ideas, engineering was something which was more closer, because I again come from more sales rather than marketing. So that was another parameter that we saw.

[00:20:39] Aishwarya Goel: And third was how big it can become. So, eventually, of course, cloud computers, one of the largest times that you can think of. So, yeah, I think this is how we choose from the qualitative: first the qualitative assessment makes two ideas, reach to the final conclusion, and then we use these measurable matrices to then take a final decision. And we actually used to do this a lot. So, whenever we went to our investors, the question that you ask, right? I had this habit of which I kind of formed during this iteration period was to document really well. So every time we picked an idea, we used to have a big five to six page document that we used to convert about the entire process. Who were these calls like ICP, what is the product wedge, why it will fail, why it can succeed. And I think that was like used to be a summary that we used to send to our closest investors who are helping us in that process.

[00:21:35] Aishwarya Goel:  And so, yeah, I think when we decided Inferless, I guess they also were able to build that confidence in us, that how we evolved and how fast we were in terms of figuring out the right piece, so I think they were quite supportive in our decision making and that's how we used to basically convert the final thesis.

[00:21:52 ] Andrew Michael: Awesome. I think the key theme is what he has being transparent and over communicated to some degree as well is how you get the support you need. And, yeah, absolutely echo that as well, having previous experiences while working with investors and definitely... I think there's like one investor mentioned this one time. He was like the time when I know there's a problem with the company is normally when I don't hear from that company, and then the time they come to me is when they're in desperate need and by that time I don't really want to help that company anymore. But the companies that are open and transparent and communicate things are good and bad all along the way like those are the companies I want to continue to support because I know what's going on, I have a good picture and I can understand that the work and effort that's being put into to get to this stage. So you did your founder market fit. You've done your customer research. Now you figured out okay, this is a good match on both ends when you're at today, how Inferless is doing, what's the vision.

[00:22:52] Aishwarya Goel: Yeah, I think, in terms of vision, we are pretty focused on.... again, I think this is also one of the learnings that we very focused. So we are working on one, this single offering which is building world's most reliable serverless GPU Inferless offering, which is  think of us as like AWS Lambda, but for GPUs.

[00:23:09 ] Andrew Michael: That's exactly my comparison out of my mind, yeah.

[00:23:12] Aishwarya Goel: Yeah, so that's our core focus. And again, I think I even say that that process that we ran earlier was like part one. So we are still in like the part two of customer discovery, where we have again created a new process which we are betting on and I hope it works out. So, as of today, like we have been for last one year, we have been heads down building the product. We have been able to build some really good, I would say. We have been able to solve some really hard problems that we started with. And our benchmarks are like 10x better than anyone else in the world today. We are currently in private beta, so we have around five paying customers who are very closely associated.

[00:23:49] Aishwarya Goel:  They have transitioned from competitor products in the likes of the biggest cloud providers and we have like a waitlist of more than 200 users who have come to a platform through different content and recommendations. And again, so currently we are in a private beta, where we also have a different approach on how we are picking customers right, which is also very fresh and controversial at the same time that how we have created a private beta criteria. But yeah, I think let's see how it pans out, but so far it's quite exciting and fun.

[00:24:20] Andrew Michael: What is that fresh and controversial take on choosing the private beta? How are you selecting customers? And just, it sounds also very pretentious, how are you selecting customers, like I think it's a very good position to be in, but yeah.

[00:24:33] Aishwarya Goel: Yeah, it's very tempting. So, again, I think, coming back to the private beta. When we started working on this offering. So our customer discovery really helped us to narrow down our ICP because, again, we were speaking to hundreds of customers. Yeah, so I think, from a private beta perspective, the criteria that we have made sure is one,  for us to test out that if we have really built something which is quite good, so we make sure that. So we are falling like a rip and apart strategy, right? So all our customers are the ones who are frustrated with the current offerings just to get, just for the sake of acquisition we didn't want to acquire at the cost of acquiring them better.

[00:25:13] Aishwarya Goel: You know, having a better sales funnel. We have always made sure that all our current customers who are coming to us, they have had bad experiences on other platforms. Because that helps us validate that if today I can replace this XYZ offering and if I'm able to do repeatedly for five customers, I have the potential to do it for 500 more. So that is one. Second, it's very important to also test in your early days that there is a willingness to pay because, again, you may be able to capture customers by giving better discounts or just being more affordable and cheaper solution. Of course it's important to be affordable for your customers, but we make sure that they are at a certain scale and they are able to commit X amount to us before we unbolt them.

[00:25:57] Aishwarya Goel: And third, we are very, very centric about our Geography and companies that we are targeting, because the whole point of seed founder, let's say, is that you cannot do it forever and you have to make sure that, whatever the next goal that you're optimizing for, there is a sense of repeatability that you can build so most of my conversations we say no to potential customers. Even if we know, we can... they can benefit from this current offering, but we have to politely say them no from a business standpoint.

[00:26:26] Aishwarya Goel: Because you know, being a lean team, it's very easy to get distracted and have so many customers and then when they churn, then you have like all the excuses in the world, right.  So I think it's better to focus on less, give them the best experience, build that kind of repeatability so that when you transition from a founder led to a more founder assisted sales, you can scale better and have better NPS and retention. So, yeah, that's like the criteria we look at and test.

[00:26:54] Andrew Michael: Yeah, I think, because when it's found led sales, you can promise a lot more and deliver a lot more, and that's not something that you can do when it's found assisted or when it's sales focused. And, yeah, just in having like a really strict criteria as well, I think having people that have the pain point, they've tried to solve it already and they've come to you, is a strong place to start with the kind of feedback that you want to get and understand and how you iterate some product as well. I think Rahul Vohra from Superhuman had a really a good framework for this as well of like only counting feedback from people who mentioned their main value prop that they thought they were building for as well. And this is sort of like a step even further that you're going is like just restricting and making sure you're only getting those people to start with through the funnel so that you can iterate and craft a bit expense.

[00:27:38] Andrew Michael: Well, it sounds like a very, very interesting and exciting time now for you, Aishwarya. Is there... actually, before we wrap up today, because you're running short on time as well as one question I was quite a lot is what's one thing that you know today about general retention that you wish you knew when you got started with your career?

[00:27:56 ] Aishwarya Goel: About customer retention?

[00:27:58] Andrew Michael: Yes.

[00:27:59] Aishwarya Goel: I came with this one line, which I repeat a lot, and try to figure out that missing gap for the great brands that I really look up to and I like, I think people may choose your product for you and for the pitch that you're doing, but they would stay because of the product that you're building. And I have been someone who loved to sell and I can do anytime anywhere, but, yeah, this is one thing that I have learned the hard way, that you know retention is much more powerful and tough than acquisition. So we really make sure that you're building for the right reasons for the... you have a solid pieces from day zero. And eventually product will help you retain your customers, not just how good you are at selling. So I think, yeah, this is one thing I've learned the hard way, but yeah.

[00:28:49 ] Andrew Michael: Nice. Yeah. I think it's a good lesson. It always used to surprise me as well, like how some of the worst products in the world like growth is actually normal sizes but just having really good sales and I think you can get a long way with good sales but then to build great, long lasting businesses, like retention is where it needs there. The focus needs to come and very nice. Well, it's been an absolute pleasure chatting today and learning about your experience, so is there any other final thoughts you want to leave the listeners with before you wrap up today? How can they keep up to speed with your work?

[00:29:20 ] Aishwarya Goel: I think the final thought that I can only say is, again, a very strong learning that I have got that if you are someone who is listening to this and is an early stage founder, early PMF, please don't outsource product market fit to an employee or to someone outside. It's your job. You are the best person, even if you don't feel so, who can achieve it. I am someone who has never done something in cloud compute, but I do feel confident that, yes, I am the best person to do it. So, yeah, it's very important to make sure that you don't outsource your product market.

[00:29:53] Andrew Michael: But nice, yeah well, thank you so much for joining the show today. I really appreciate the time. It's been really enjoyable listening to your journey and I wish you best of luck now going forward as you iterate closer towards PMF.

[00:30:03] Aishwarya Goel: Thank you so much, it was quite fun.

[00:30:05] Andrew Michael: Thanks, cheers.

[00:30:14] Andrew Michael: And that's a wrap for the show today with me, Andrew Michael. I really hope you enjoyed it and you're able to pull out something valuable for your business. To keep up to date with churn.fm and be notified about new episodes, blog posts and more, subscribe to our mailing list by visiting churn.fm. Also, don't forget to subscribe to our show on iTunes, Google Play or wherever you listen to your podcasts. If you have any feedback, good or bad, I would love to hear from you and you can provide your blunt, direct feedback by sending it to Andrew@churn.fm. Lastly, but most importantly, if you enjoyed this episode, please share it and leave a review, as it really helps get the word out and grow the community. Thanks again for listening. See you again next week.