She bet on a consumer app when every VC wanted B2B—then grew to $10M ARR. | Anada Lakra, Founder of BoldVoice

Anada Lakra (00:00:00) :
If you're not embarrassed by the first version of what you launched, then you launched too late. So, it was very important for us to just get the app out there as quickly as possible. So that we could start iterating and learning from real user feedback. We had both gone through this and we had first hand experience, and a genuine passion about building a solution for this. And also genuine understanding of the pain point. And so, basically, the founder market fit was there even before we had product or product market fit. I don't think of product market fit as this on or off toggle. I think of it as a spectrum. For sure, if I had to pick a number I think the one million ARR felt like, OK, this is not just a novelty, or we found a few people who really like this. But this is something that clearly a mass of people are starting to use and I think it was not just in the revenue metric, it's also in the usage metrics.

Previous Guests (00:00:54) :
That's product market fit. Product market fit. Product market fit. I called it the product market fit question. Product market fit. Product market fit. Product market fit. Product market fit. I mean, the name of the show is Product Market Fit.

Pablo Srugo (00:01:06) :
Do you think the product market fit show, has product market fit? Because if you do, then there's something you just have to do. You have to take out your phone. You have to leave the show five stars. It lets us reach more founders, and it lets us get better guests, thank you. Anada, welcome to the show.

Anada Lakra (00:01:22) :
Thank you for having me.

Pablo Srugo (00:01:23) :
So you started an app called BoldVoice. Which is, it's a little bit like Duolingo, except it's focused strictly on the accent piece, right? The how you speak, not just learning the language but really getting the accent right. You also started in 2021, which was an interesting time to start. I mean, there was a lot of hype but then, it was painful afterwards and yet, things seem to be going up and to the right. You just raised a $21 million Series A, a couple of months ago. So we'll go through that whole story and kind of talk a lot about what it takes to launch a successful B2C. We don't see that many B2C, frankly. We see a lot more B2B, right? Direct to consumer app in the 2020s, right? Not the 2010s when there was like two apps, now there's a lot. So anyways with all that said, maybe just as a first question, tell me a little bit about that 2019, 2020, 2021 period. What were you doing before you kind of had the idea?

Anada Lakra (00:02:14) :
Yeah, so the story starts in probably 2020. I was in my dorm room at Harvard Business School. I had recently racked up a consulting internship that was not for me and I was thinking about actively what to do next. And I had already done this startup before that. I'd been through the founder journey, and I knew how hard it was. So I wasn't necessarily eager to jump right back into it, knowing what it takes. I was actively thinking, though, about different ideas that would make me compelled to do something like this again. When I was doing that, something that was very top of mind for me at the time was the fact that I was surrounded by a lot of international students back in business school. Who were coming from all types of diverse backgrounds from almost any country in the world, and I remember thinking how much those international classmates were struggling to raise their hands in class or how they were talking about how difficult the interviews were for them. Because they felt like they were not that confident. They felt like they couldn't express themselves as clearly and these were, very smart and very capable people that were being hindered by their English. Even though on paper, their English was great, right? It was just when it came to speaking, their accent was getting in the way and this is something that was very, very familiar to me. Obviously, you can hear I have an accent. I'm an immigrant myself. I've moved to the US at this point well over a decade ago for college, and I experienced this firsthand, right? You learn English for ten years, you feel like you passed all the tests and you're so confident. And then, you move here and you quickly have everyone asking you where you're from, and if you can repeat yourself. Just like universal immigrant experience that is almost a rite of passage and I remember being struck by, wow, it's been so long and people are still struggling with exactly the same thing I've gone through, and nothing has changed. And around the same time, again, this is pre-ChatGPT, this was pre the big AI revolution that we've been seeing over the last couple of years. But already AI and audio AI were becoming more powerful, and more interesting. So I started to think about, is there a way that we could potentially build something that can serve as an accent coach and a speech coach for all these people who have already done all that English learning on paper, but want to speak it more confidently in real life? Because that has real stakes. It's not, English is not like a Duolingo language of, I'm learning a little bit of this and I'm at a hobby level, and that's enough. English is the global language of business. You need to speak it at a professional and very confident level. And that nth degree does make a difference in your career outcomes. So I felt like it was a big market. It was a very overlooked problem and for the first time, there was this nascent AI capability that could actually allow us to do something about this.

Pablo Srugo (00:04:56) :
Were there any digital products at the time, any other apps that were trying to tackle this problem? I'm sure you, especially coming from consulting and hardware. Did a good, extensive market mapping, that would be my assumption.

Anada Lakra (00:05:08) :
Yeah, yeah, exactly. There were maybe one or two, but basically maybe one that was doing something around pronunciation training. But it still felt like this was very underdeveloped compared to where it could be and we also had this vision where just using AI and giving people feedback, without helping them actually fix their problem and improve it, was not going to be enough. And still, a big part of our product direction and vision has been combining the best of humans with the best of AI. So the app users also get video lessons from Hollywood accent coaches. Who are the best in the world at teaching people accents and confident communication. And we pair those personalized pre-recorded videos with the speaking practice that you do, that you get feedback from AI. Because AI in itself will tell you where you made the mistake, but then the coach comes in to actually help you fix the mistake that you made. So we had a specific product vision that even though there was one or two things in the market that were kind of trying to do that, we saw that there was a big gap between that and the vision that we had.

Pablo Srugo (00:06:14) :
And was there a big non-digital industry that helped? Obviously, learning languages is a huge industry but specifically pronunciation, accents. How big was that, on its own? How much did people already pay for those sort of things? Or was it more of, you live and learn, and most of the time you just, you know, you deal with it? What was kind of the status quo?

Anada Lakra (00:06:35) :
Status quo was accent coaching exists, there are professional accent coaches who you can hire for $200 an hour to help you improve your pronunciation. It's very expensive, and there are a lot of C-level executives that do this, and, you know, silence. Because there is, let's be real about it, there's also a little bit of a taboo about talking about accents and improving your accent. It's something that is very tied to people's identity and also it's something that people feel judged about and subconscious about. So they're not out there telling the world that they're doing it. But you better believe that a lot of your, the CEOs that are non-native speakers have speech coaches and accent coaches actively helping them through this, and it works. It is very expensive, and the average professional absolutely cannot afford something like that, nor can they make a habit out of a very expensive one hour a week class in the middle of their business schedule. So our vision from the beginning was we're not going to replace or disrupt this market. This market right now is too small because the price point is just inaccessible. We're going to create an alternative that is fully digital-based and therefore fully scalable that can reach not just the few people who can afford it, but the one billion non-native English speakers in the world.

Pablo Srugo (00:07:54) :
And so when you have this idea, you're at HBS, you're at Harvard Business School? Do you drop out to focus on this? Do you finish out the semester? How do you play it?

Anada Lakra (00:08:02) :
This was my second year of HBS where the curriculum was a little bit more flexible, so I used my classes actively as ways to narrow into this problem and I used this as my case for a couple of classes that I took. So, it didn't feel like I had to drop out. I wasn't, I was at that point maybe six months from graduating. So I figured, let's just keep going through this and actually, it was a great forcing mechanism. I knew I had six months to graduate and I set my goal for myself that after, I would not recruit, I would not accept any job offers, I would not think about that at all. I'd spend those six months just building and trying to raise our pre-seed round. If that worked, great, then I'm doing this full time. If it doesn't work, then only at that timeframe, I can start to consider other options and that. Again, it was a great forcing mechanism to drop everything else, focus on this and we got lucky that, we both got a term sheet from a pre-seed investor but we also got into YC. And I didn't have to think about alternatives.

Pablo Srugo (00:09:06) :
And I see you have a co-founder, when did you set that up? Was that at the beginning or was that kind of later on?

Anada Lakra (00:09:12) :
Yeah, my co-founder and I have known each other for ten years plus, first job out of college. So I was very lucky to already have someone that I knew and deeply respected, and had a very complementary skill set to me. He's technical, I'm more of a product business person. So I involved him early on with some of my exploration. He was gainfully employed, not interested in the startup lifestyle at the time, but definitely supportive. But also wanted to see more evidence and so again, I was very, very focused on de-risking this as much as possible. Which in the early days meant, can I get any funding for this? But before we even talk about that, I'm not the subject matter expert here. I'm not an accent coach. Can I recruit a stellar accent coach to join our efforts? And so I did that. I reached out and got a fantastic coach, Ron Carlos, to work with us. Who was a Hollywood accent coach to the stars, has trained Game of Thrones actors and more. And yeah, so I was working on de-risking the product and the concept a little bit. And then, my co-founder and I both applied to YC. And then when the decision of YC came through, there was no question. We both knew that that's what we would both do full time.

Pablo Srugo (00:10:26) :
Was the first version matching you to an accent coach or was there always an AI piece to it? Like, a fully digital piece and then the accent coach was the premium tier?

Anada Lakra (00:10:35) :
It was never a one on one accent coaching, because that was the very thing that we're trying to disrupt, right? We're not trying to do the very expensive thing. We're trying to do a digital thing, but the level of feedback that we offer now compared to what we offered back in the day was very different. Because again, we have been training our own models based on our data and based on expert ML engineers that we have on the team. Plus coaches who help us annotate and train the model to be as close as possible to how they hear accents, and all the component parts of them. So we had a very rudimentary first version of that in the early days that we've worked to improve. But yeah, the vision was always, how can we make this into something that could be as effective, if not more effective as a one on one coaching session, but at scale.

Pablo Srugo (00:11:20) :
Tell me a bit about, YC. So many founders are trying, thousands of founders are trying to get into YC. Most don't get in and a lot of the stuff I see is like B2B-focused, right? And B2C in apps, it's just generally harder to kind of have this, you know, massive market size opportunity. Obviously, there are companies that do get there, but I think in general, you know. What can you tell us about getting into YC and what you think was the thing that they really latched on to, or the thing that you did really well that gave them the confidence you could build something big?

Anada Lakra (00:11:49) :
I want to push back a little bit on B2C as an opportunity because, yes, I think a lot of VCs do go after B2B as the more repeatable and proven business model. But B2C I think has huge opportunity now with AI to really transform a lot of the way that we live, we learn, we entertain ourselves, we work, all of those things and I do think now VCs are starting to pay more attention to consumer. But in terms of us getting to YC, I think the reason that we did it was twofold. A, we understood the problem. This was not two people who were wanting to do a startup and, found an idea and said, this is what we're going to do randomly. We had lived through this. Again, I already shared my experience. My co-founder, he saw his immigrant parents struggle with this and struggle to advance in their careers. Because they never felt like their accent and their communication was good enough to lend them into management roles. So we had both gone through this and we had first-hand experience, and a genuine passion about building a solution for this. And also, genuine understanding of the pain point and why this is difficult. And so that was basically the founder market fit was there even before we had a product or product market fit. And the second thing is that, again, as I said, we had already de-risked a lot of things. Which showed that we can get the coaches involved that we need. Which showed when we applied to YC, we already had a pre-seed investor who had given us a term sheet. So we tell them you're serious about this, we've already gone outside investment. So you want to gather all the potential proof points possible in order to differentiate yourself from a napkin idea to a real business, and of course now things have changed a lot, right? Anyone can build apps with VibeCode these days, right? So the threshold and the barriers are only higher, and expectations are only higher. So I think my best advice for someone applying to YC right now is, do as much as possible before you apply. De-risk as much as possible, you need a product. You can VibeCode a product, so you have no excuse not to have a working product. For at least most things, I'm sure there are some things in hard tech and so forth that maybe that's not the case. But for a consumer product, you can certainly VibeCode something, and you should also be then focusing on how do you get users and have some real usage, and ideally also some revenue coming into the app. And again, that's become easier to build, which means that now it's the expectation.

Pablo Srugo (00:14:14) :
And then one of the issues with B2C, these outcome-based B2C apps. So take like, dating is maybe a weird example, but if you think about it. It's like, if the apps are really successful and you find someone, you churn out of the apps. And I don't know how much that drives, you know, we saw a lot of these companies like Bumble go up and then come down. I don't know how much that's driven by that versus just change, but it's obviously a retention-related issue. When I think about Duolingo, they almost solved it, if they have solved it, by turning it into a game. Where it's just, people do it because it's fun and there's a lot of people that'll tell you, you know, you don't really learn a language through Duolingo but it's kind of like edutainment, you stay there. How do you think about that? To an extent that somebody comes in, you know, you help them get their pronunciation to a certain level, fix their big accent issues, and then they feel like, OK, cool, I'm good, I can speak English at eighty percent, ninety percent. Whatever that threshold is, do they just churn out? How do you think about? Is that even a problem? How do you think about that retention piece?

Anada Lakra (00:15:08) :
Retention is always a focus, and they almost definitely seal with any consumer product, right? Especially ones that are in the education area. Best in class is not very good, and best in class being Duolingo's and the way we think about it is that there will always be some churn. And if it's positive churn, meaning these are users who had a great experience. They got some improvement, then the market here is large enough that they will now become ambassadors, and they will tell other immigrants they know that this helped them. And they will bring the next people in, right? So if the market is large enough, you can afford positive churn and we really very much focus on. Again, that learning experience working and if the person got what they needed, and they churned. And they will tell the next person, that's great. But then the second thing is that we're working on a problem that seems deceptively easy, but it's actually really hard. Because we're solving the last twenty percent of language learning that takes eighty percent of the time as anything else when it comes to mastery. You can actually learn grammar and vocabulary pretty quickly, but learning how to change the way that you speak and accent and pronunciation is something that takes a lot of developing of muscle memory. You're learning new ways to move your mouth to pronounce sounds. It's a little bit like going to the gym. You're like strengthening new muscles in your mouth and then you also have to keep it up. Because if you get lazy and you don't keep up the training, and the skills, then you'll atrophy. You won't keep those results anymore. It's a bit of an asymptotical curve where you kind of get close, and then different people again have different thresholds of what they're trying to reach. But for someone who constantly is getting results and sees an improvement. It becomes addictive, and they can always improve that little bit more. And then it's the product's job to always unlock that next level of improvement for them and to also be relevant to their daily life, and to their daily, for example, work meetings, right? You always might have a column calendar that's important. You might be able to record that with BoldVoice. We'll give you insights into, OK, I mispronounced this word. This one was good. I improved this one, but I still need to pay attention to this one. So it becomes more like not just an education product, but a continuous utility product that you continue to use. And that's the one kind of highlight category out of all consumers, utility, right? If you nail the Spotify business model, then you're not going to churn as much. But frankly, it's still a work in progress for us to kind of bridge the gap from just pure education, where you wrap up a course, to something that feels much more like lifelong utility.

Pablo Srugo (00:17:42) :
Was is? You mentioned Spotify, what is the business model? I assume it's freemium and there's some tier above which you pay a monthly fee. What does that look like?

Anada Lakra (00:17:49) :
Yeah, for Spotify, I was thinking less the freemium to paid. More the fact that, you know, you're not going to listen to all the music that you need to listen to for one month and then leave, right? You'll always want to benefit from it and there's that continuous use case. Where if you want to continue to listen to music without ads, you can't binge it and then be done with it. You have to continue to use it and pay for it. So similarly, what we focus a lot on is, again, not just giving you a static learning experience where you're one and done. But a hyper-personalized learning experience that is contextual and pulls in from your work meetings, and from the real life speaking that you're doing to always surface the next thing that you miss, maybe that interaction you can say better next time.

Pablo Srugo (00:18:33) :
So what is the business model then for BoldVoice?

Anada Lakra (00:18:36) :
For BoldVoice, our business model is a subscription and users pay $150 a year for unlimited access to our personalized curriculums with Hollywood accent videos and unlimited speaking practice with feedback.

Pablo Srugo (00:18:50) :
So going back, you get into YC and I know YC is all about, in that three-month period to Demo Day, build and ship as fast as possible. How do you work through that stage? What does V1 look like? When do you get it out? What happens? Walk me through that stage of the life.

Anada Lakra (00:19:05) :
YC was crazy. Again, it was three months. It was super compressed. We joined YC pre-product, pre-revenue.

Pablo Srugo (00:19:12) :
This was when at this point, by the way? Timeline wise.

Anada Lakra (00:19:15) :
This was summer 2021. So this was June, we do the first week of YC and the first goal, no question, is we need to get this app on the App Store. Because there is so much that we could build and we wanted to make it perfect. But I think there was this quote, I forget from whom, but if you're not embarrassed by the first version of what you launched, then you launched too late. So it was very important for us to just get the app out there as quickly as possible so that we could start iterating and learning from real user feedback. So I think it took us, we were already working on this for a couple of weeks, a few weeks before YC started. So about a week into YC, I believe, we launched the first version of the app on the App Store and it was definitely not something we were proud of but it was out there, and people could finally download it. And then I remember within that first week, we got our first purchaser who bought a monthly plan. Paid, I think it was $10 a month.

Pablo Srugo (00:20:13) :
How did it work by the way? Was it a seven day free trial or you had to pay right out of the gate? How did you structure that?

Anada Lakra (00:20:20) :
I don't think we had a free trial at the time. Right now we definitely do, and we could quickly implement that. But I think at the time it was an instant purchase, I believe. It wasn't about let's invent the exact, the perfect pricing structure and invent the wheel when it came to it. It was just, from the beginning, I knew that I wanted this to be a paid product. Because if you just get a bunch of free users, you're really getting a lot of mixed quality, mixed intent, and having people actually pay for it filters it down to those who actually have a need for this app, and therefore the quality of the feedback you get will be much higher. So it wasn't even about making those $10 in the early days or monetization. It was just about getting the best learnings, and also showing to ourselves that this is something that people care about enough to pay something for it.

Pablo Srugo (00:21:07) :
It's a big difference between what you're doing and what Duolingo is doing, and not to overly compare. But I mean, paid here makes more sense, I would think. Because you're going after somebody who's trying to, like you said, master English and usually you're mastering it for some sort of professional setting. You know, there's like a dollar attached to it, versus Duolingo. Maybe for a million different reasons you're doing it, but it could just be a hobby, and so if they were to say, "You have to pay $10 a month," they'd probably kill the business overnight. Whereas in your case, forcing people to pay means you're only getting the top ten, twenty, thirty percent who are serious about it.

Anada Lakra (00:21:41) :
Exactly, yeah. I think the comparison to Duolingo makes sense at the very high level when you think, OK, it's language learning. But then when you think about the user type that we bring in, their intent, the ROI they get from this kind of investment that makes a direct impact and how they show up in their next business meeting. How they interact with their client, whether they can get that job and that promotion, and all of that is totally different. So we don't even think of ourselves as language learning. The user has already learned the language by the time they come to us. We think of it as communication training. They want to be the best, most confident, most credible professional they can be and as a non-native speaker, that means perfecting your English communication skills. So yeah, it is a different business model. You know, this is something that is directly juxtaposed to, again, as we said earlier, you pay $200 an hour for one hour of accent coaching with a real coach. With BoldVoice, you get a year worth of unlimited practice for less than the price of one coaching session. So the investment and the ROI makes sense.

Pablo Srugo (00:22:44) :
Tell me a bit more about V1. What was in V1? How did you scope it out?

Anada Lakra (00:22:48) :
In V1, we started with working with our coach to produce video lessons. I remember we shipped him a camera, and he scripted these lessons, and shot them from his apartment. And we tried to make it, as high quality as possible but it was clearly, you know, kind of not fancy production. But the focus again was to have enough content that we could launch an app with it and we took these videos back, we had a video editor. I was, you know, manually looking at every script and changing every script, and then making sure every video was perfectly edited to get it into the app. And then we also worked, especially my co-founder, worked on building, again, this rudimentary feedback mechanism. Where people could record their voice and get feedback at a sound by sound level. Because otherwise, just a simple thumbs up, thumbs down will not work for the level where our users are coming to us from. In the early days, we were just focused on one language, and that was Hindi. In hindsight, I don't know why we picked that language because neither of us are from that background. But again, I think it's something that we learned very quickly. It's like, why are we limiting it to just this language? The theory there was that accents are so based on your native language, we can create a much, much better learning experience if we can create lessons that are personalized to the native language background. But then we were quickly able to think of the video lessons as not like each language needs its own unique, specific video lesson but more as, here are the lessons and we can use them as building blocks. And this language will not get this building block, but this one will get this extra one based on. So we were able to think about this in a way that allowed us to much more quickly add and cover basically every language out there. So the first version of the app was a bunch of videos and some practice materials. I think we had, in the first version, maybe enough materials for someone who used it for thirty days to run out and then we launched it, and then we worked on adding more videos and more materials. So by the time that person ran out, they had more stuff to work on.

Pablo Srugo (00:24:51) :
I mean, the video is one thing, the content. But the fact that you could speak and get feedback, you'd probably be at the cusp of what was possible. I mean, these days I think that would be taken for granted, but I would assume at that point that was the cusp of what was possible, right? I don't think that technology was all that old or would have worked maybe three to five years before when you started.

Anada Lakra (00:25:06) :
Yeah, it was. Again, as I said, it was not very precise and nowhere near the degree of the sophistication that it has right now. But we were able to pull together, using some open-source things and some vendors. Pieced together something that worked enough for us to, you know, get something that looked and smelled in the right direction. And then allowed us to, OK, there's a flywheel, people are coming in, they're recording their voices. Now we have our own real data to train models on. They're reporting where there are issues with the feedback that we're giving them, so we have all this qualitative feedback from the users. Then we have our coaches who can, now that they're done with this big content sprint, also focus on annotating this real user data to help us figure out what to train the model on. So we got something going, we kind of solved the chicken and egg problem with this initial solution that was good enough. And then, we knew that that was just a starting point and that would allow us to get the data so we could improve the model.

Pablo Srugo (00:26:09) :
Let's go deep on go to market, right? Because I think there's a lot of apps that get launched all the time, and most of them go nowhere. Get very little to no traffic. So YC starts, you put out V1 in the App Store. What do you do to get people to find it, discover it, download it, spread word of mouth, et cetera?

Anada Lakra (00:26:28) :
What didn't I do? In the early days, it was really crazy guerrilla approaches across everything that we could think of. Well, first of all, we were part of YC, right? With a lot of international founders. So we were constantly asking people to join the app and try it out, and see if it could help them nail the pitch that they were going to give at the end of demo day. I was joining Reddit groups where people were talking about this and commenting. There's this Reddit thread called Judge My Accent that was the perfect environment for something like this. Where people were already soliciting that kind of feedback on their accents and their speech. I joined Facebook groups with various ESL teachers. I got on calls with them. I started DMing them. Can you get on a call with me? Is this something that would be helpful to your students? I was reaching out in those early days to international student groups on campuses to see if they wanted to, you know, try a free version of it and the only thing is that they would give me feedback. So in that time, I was constantly in calls with a lot of these people to try to understand. They were like, oh, they weren't liking, and how we can make the app better for them. And again, a lot of the people I spoke to in the early days were not, didn't end up being our target audience anyway. But it was a process of discovering who is the target audience for this.

Pablo Srugo (00:27:47) :
What were some of the things that you learned through those calls about who was and who wasn't your true ICP?

Anada Lakra (00:27:53) :
Yeah, we quickly learned this thing that was, to us, counterintuitive at the time. Which is, you know, students, international students, seem like such an appealing audience. Because, you know, they're easier for us to reach. We can reach out to international student groups, and they'd be interested in something like this. It made sense, theoretically but then when we reached out to them, we realized, A, the willingness to pay isn't there. When I was a student, yes, I couldn't afford $200 accent classes but I also could not really even afford $10 a month to pay for an app, right? So the willingness to pay profiles were pretty different. Two, they were in a social campus setting surrounded by a lot of other international students, and they could pick things up a little bit more quickly on their own. And it was a much more international environment, so maybe they didn't feel the pressure of their accent as much as someone in a work environment would feel. So we started off being maybe overly focused on students and then quickly realized that no that's not our ICP. Our ICP is someone who is a working professional. They're in their 30s and 40s. They are pretty advanced in their career and they want to take the next step, and they're frustrated that even though they're an amazing, smart, capable professional. People still overlook them because of their accents and they also have the willingness to pay because they see the ROI of something like this. Because if they get even marginally more likely to land a client pitch or investor pitch or get the job promotion, that's going to pay for the app investment for itself, a hundred fold. So, that was the learning and, again, it took us a while of doing user interviews to understand that and also, then seeing who is actually paying for the app.

Pablo Srugo (00:29:31) :
We have tens of thousands of people who have followed the show. Are you one of those people? You want to be a part of the group. You want to be a part of those tens of thousands of followers. So hit the follow button. I dug in on that because I think figuring out product market fit and so much of it is about, the subtleties and the specifics. You think about who exactly would pay for this and why. And it's hard to figure it out on paper. You just have to go out into the world and do this stuff. But as you're talking and telling me some of these things. It seems now obvious, right? But it's, for example, for this, you need to be truly a top of mind problem for your customer and for this. Probably the best situation, as you can imagine, is somebody who's just had a situation or is having these situations. They're in a business meeting, they're trying to get a point across, they can't get the point across, and they can feel they're not being respected to the same level, and that leads to a certain level of frustration. Where it's like, I need to figure this out, I can't not versus the international student, you think, you know, you're gonna need to figure this out. But they're also in another mode, they're doing other things, whatever, later. You know, and so that affects the need, and obviously the willingness to pay has to do with that. But it's like if you can hit them right when that frustration is at its peak, that's when they're most likely to say, you know what, I can't ignore this anymore. I have to do it because it just, it's truly pissing me off today. Which I don't think is something you could have figured out at the outset, right? But once you clue in on this, obviously it drives a lot of the go to market and the messaging, and all the other stuff.

Anada Lakra (00:31:12) :
For sure, and that's why I think it's so important to launch early. Because then you're not talking in hypotheticals. You're not, again, doing potential user interviews where people are saying one thing, but then when they get access to the app, they do something very different. The app is out there, you can see these types of people are using it. Who's actually coming back the next day and the next day. Who's actually paying for it, who has churn. So it gives you real data instead of just what people say, because people are very, not intentionally so, but are not very good at describing what they want and their priorities, and it's their actions that show it. So you want to get to the point where you can get a read on their actions on the actual product as quickly as possible.

Pablo Srugo (00:31:54) :
So as you figure this out and you're trying out all these guerrilla tactics, what are some of the things that really work and move the needle early on?

Anada Lakra (00:32:01) :
Yeah, I remember this was in probably late July. We were a few weeks away from Demo Day, and the big thing that we did that was part of the YC launch playbook was launching on Product Hunt. And that was a bigger thing, I think, back from the day. I don't know if Product Hunt is still nearly as active now, but yeah, we took it seriously. We did a founder video where, I showed how the app works and, some nice screens but still we had no idea whether it was going to work and we were actually really surprised. I think we hit number two of the day for that day and it got us, I just remember us looking at the charts, and it's hundreds of people. But when you have, a dozen subscribers, getting hundreds is a wow moment and that was really cool.

Pablo Srugo (00:32:46) :
It's surprising though, just because you'd think that the Product Hunt audience is like early tech adopters. So I'm actually surprised that that drove that many paid installs.

Anada Lakra (00:32:57) :
Yeah, but a lot of people in tech are immigrants, are non-native speakers. So it ended up being actually a really good audience for us. But yeah, again, it wasn't necessarily something that we were expecting. So we were very pleasantly surprised. But again, it's like, you can't launch on Product Hunt a million times. You can only launch once, so it wasn't a repeatable playbook. But it was really great to, again, to go from, OK, we're making hundreds of dollars of revenue to what we're making thousands of dollars of revenue and then, most importantly, in the early days, what you really care about is just, again, narrowing down on who's your ICP, who's paying, and who's coming back. So the more people you get in initially, the closer you get to the answer. So it was really impactful for that and also, being able to say this is our ARR. When we did Demo Day and reporting a much higher ARR than we had before was also very nice.

Pablo Srugo (00:33:51) :
What were you doing to get more data out of users? Were you calling them? Were you talking to them on the phone? Or was it just kind of, they gave you some data as part of their profile or whatever it was, and that's kind of what you used?

Anada Lakra (00:34:01) :
Yeah, I mean, we did all of the tracking and Mixpanel analytics and everything that every product does. But I was also sending emails to all the users asking for user interviews or just feedback via email. We also had this "give us feedback" in the app back then, which is still currently there. It has not changed. If you go to the Boltwoods app and the profile, you'll see my face and Co Founder Ilya's face that says, "Hello, we're the founders of BoldVoice, and we'd love to hear from you." And then people will reach out, and it's so much more, I think welcoming for them to see the founders' faces. Caring about their feedback, than to just see a random chatbot or an email that is going somewhere anonymous. So very, very focused on doing user interviews then and getting feedback in all the ways possible. And I think this is still, a habit that I continue to have where Ilya and I, and the rest of our team will do these user interviews. The first time you users, people who have been using BoldVoice for three years and still use it, to know exactly what our users are perceiving and not just rely on customer operations, for example, to tell us what's going on. It's super important, I think, to always know how your customers feel because sometimes, I think, over the years, you may develop blindness to certain areas. So it's so illuminating to have a first-time customer say, "This thing makes no sense." and you're, like, "Oh, it doesn't make any sense, we've just had it this way but actually, this is really good feedback." And I think it's also very energizing to hear directly from users. I think it's a really good habit to build in the early days but also, you always need that voice of the customer to make the product better throughout the company's lifecycle.

Pablo Srugo (00:35:44) :
So after Demo Day, shortly after, you raised a $3 million seed round, but as you mentioned, you're still, relatively small. I mean, it's growing but you've got, kind of, thousands in MRR and I assume you're trying to figure out that go to market playbook. What are some of the repeatable things that you start trying, and what are some of the ones that start to work?

Anada Lakra (00:36:05) :
We are trying to do more of the guerrilla approach but the thing with that is, that it's hard to scale, for sure. You're kind of starting from scratch every time. So what we take our next bet on is ads and figuring out, can we reach people scalably through different social channels? And we try different ones. I think we tried, in the early days, some Google search ads. They don't really work. We have a couple of different channels going, YouTube and Instagram as well and we have no idea, is the channel even going to work for us? What is the formula that works? Also, this is a mobile app. This is post-iOS 14.0. If you don't know what that means, as a listener, I'm happy you don't ever have to deal with it. But for mobile app developers, that is a privacy framework that creates a lot of headaches around tracking and measurements. So you don't actually really know how your ad dollars are converting to subscribers and so there are a lot of things at the time that made this very much of an open question of whether we would be able to crack it or not. And it took me taking a lot of different shots on goal to try to figure out what is going to work here. Make a bunch of videos, some more graphical videos about how the app works and then I remember reaching out to this one creator and paying her, I don't know, maybe $200 or something, to make a video on YouTube that she posted on her YouTube. And they were like, OK, well, maybe this segment could actually be used as an ad. So we tried that, and that worked super well. And it was incredible to see something that is like, oh, this is not just a one-off thing that lasts for five days. This continues to drive clicks and installs, and trials, and real revenue. And then we knew that, OK, this channel can work for us. So then it's like, OK, how do we get the next influencer to work with us and expand that channel? And then maybe testing the same video on Instagram also works. Because then we can expand and have a multi-channel strategy. Even though in the early days, I don't know, if you find one channel that works, you should just focus on that. But yeah, unlocking paid in a way that made sense for our unit economics was key to starting to really scale this thing.

Pablo Srugo (00:38:18) :
This started with videos you said on YouTube? Is that? Were those the first, or on Instagram?

Anada Lakra (00:38:24) :
The first one was YouTube for us.

Pablo Srugo (00:38:26) :
And basically you just get an influencer to create a naturally sounding video and then you just pay to promote it. That was kind of the strategy?

Anada Lakra (00:38:33) :
Exactly, and you've got to figure out what is the thing that you need in that video for the creator. And for us, I think the big wow moment of the app is that you can record your voice, and instantly you see it grading. You see the colors, you see the green of what you did correctly, you see the reds and then you can say it again. You can fix that mistake, and you get the green and the high score, and the confetti. So it's a very audio-visual kind of experience. It's something that's hard to describe in just words, which is why I think the Google ads were not as impactful. The search ads, but the YouTube is impactful because you can show that visually, and the person can hear, "Oh, I heard a mispronunciation. Oh my gosh, the app caught it." And then she was able to fix it. And then it correctly caught that that was the correct pronunciation. So that kind of showing that core loop and the wow moment in the video was super important to have that be something that works for ads. Because if we had just had her talk, I don't think that would have worked. I mean, I tried different variants. So try to figure out not just what channel works and influencer collaborations, but what is that one thing that you need to show in that video that is key to get people to convert. And when we discover that, then we can repeat that formula with a lot more creators.

Pablo Srugo (00:39:48) :
Yeah, tell me maybe more about. Because a lot of people I think want to use influencers to grow their business. Once you figured out the wow moment and kind of had one video from one influencer that seemed to be promising. What did you do to make that into more of a formula and really scale it?

Anada Lakra (00:40:03) :
I was personally finding people who are close to our ICP. We didn't need, Americans to talk about this because it made no sense, right? So we were trying to find people who had an accent, but also their accent wasn't, you know, it was aspirational, right? Someone who was a believable person would be an ambassador for the app who had clearly very confident, clear communication. So finding that person takes manual effort because there's no filter that you can use on YouTube to figure out the accent level of the creator. So it's about digging and finding who is the best representative for your product and for the problem you're trying to solve that will resonate with your target customer. And then we'd create a script. And initially, the scripts would look pretty similar because, again, we're just starting to do this. So we were nowhere near saturated. Most people, ninety-nine percent of people, were seeing us for the very first time. So we're focused on just, OK, let's find a formula that works. Let's just try to replicate that but then, at some point, you also get into diminishing returns from the same exact thing. So you also have to get creative about trying out different hooks, trying out different angles. Maybe one can be more about personal life confidence. The other one could be more about professional. In some, we might be talking about pronunciation; in some, we might be talking about accents because different terms matter to different people. So you've got to figure out what are those different angles and different influencer personas that you can bring in that will appeal to different audiences. And then, it's just a matter of testing and iterating. And turning this into, also with the more budget you have behind a channel, the more you need to always have a creative pipeline going at a higher volume of creatives. So you can test various things, ideally on a weekly basis, to find a new winner. Because even your best creative, after a while, will get fatigued. It will have been shown to everyone in that group, and you need something fresh to take its place. So you need to get into a very good creative testing framework and velocity. But first, you need to figure out what is the first winner, and then only go to the next to find the next winner when you feel the first one is starting to show fatigue.

Pablo Srugo (00:42:12) :
And then besides the fact that these creators obviously weren't Americans, what did you find was the most successful type of creator? Were they somehow in this language learning or pronunciation world or were they completely from somewhere else?

Anada Lakra (00:42:25) :
This was also very counterintuitive initially. We're like, well, there's a lot of English teachers out there, a lot of whom are non-native speakers. Let's try to work with them and I think it goes back to the insight we had in the very beginning about how different this is from language learning and from Duolingo. Our user is not still learning the language. They speak English, they know English, they're using English at work. They're not going to be following these basic English learning influencers or creators. The people who are following those are much, much earlier in their language journey, and they're not looking for an app like BoldVoice. So coming to BoldVoice, I'd be like, I don't understand what's going on, and churn right away. So it's definitely been trickier to find these people who can be good creators for BoldVoice. So I don't have a formula to share there. I think it would be very, very specific to us and to our brand. But again, it's kind of, I think it's the same exact thing that we're saying earlier about, you know, the difference between language learning and communication training. The difference between what sounds good on paper and what's actually true in real life, the insight around the international student versus the working professional. So the same thing applies here, and I think it took us a while. It took us some tests to figure out, OK, maybe this person who is a language creator doesn't work. But this person who has been talking about being an immigrant and their progress of adapting into the American culture, and workplace. That's a much better person but it's, again, so specific to each company and to each product that I would say the most important thing is to really understand the user mindset and where they're at. And if you were to scroll their feed, what would they be following? What would they see? And then plan along those lines.

Pablo Srugo (00:44:04) :
Can you give us a sense of what a great paid channel like this looks like from a metrics perspective? Are you looking at ROAS? Are you looking at LTV to CAC? Are you looking at conversion to pay? What metrics are you looking at, and what were they in those early days as this strategy started to play out?

Anada Lakra (00:44:21) :
Yeah, you need to have a really good handle on unit economics because you can have something that you think works and then doesn't work. Because churn is way too high, and you don't realize that until you're three months in, and then you thought these people would have stuck around for six months, and you're planning for that but they actually only stuck around for two months. So one thing that we did not very early, but once we did, it was really transformational, was have the default plan be annual instead of monthly. Because that gave us a much, much clearer insight from, you know, the very first purchase moment, once they converted from the trial, that this is going to be their year one LTV. So we didn't have to guess about how much they convert month to month and then our goal was for the acquisition cost of that user to be well below how much they paid us in that first year. So we'd be getting this instant CAC payback. You know, we sacrificed some LTV over CAC there for the sake of the instant CAC payback. So you really got to know what you are optimizing your business around. I think the annual works really well, especially for the types of products in consumer, where, you know, churn industry-wide is a problem. Because this gives you a level of predictability that is otherwise really hard to get. But if it's more of a utility product, where ongoing usage is something that is much more predictable, then I think monthly is also fine, but then you have to have a really good sense of when you get CAC payback and having that window be as tight as possible. When it came to the ad platforms, I got to look at all the metrics. What matters the most for us is cost per trial and then ultimately cost per subscriber. Cost per trial was the item that we got on day one because cost per subscriber has a one-week lag. Because of the one-week free trial. So that cost per trial was the thing that I cared about the most. Then you also have to look at all the earlier funnel metric to kind of get a gauge of the quality of the ad, and sometimes you might get conflicting things. Where the cost per click is really good, but the cost per trial is really high, and you really have to use that data to fine-tune what ads are working and what seems to be working but actually isn't.

Pablo Srugo (00:46:32) :
And then once you figured this out, and I'm sure layered other things on top. Do you remember your growth trajectory, like revenue wise from then till now?

Anada Lakra (00:46:39) :
If you zoom all the way out, it's up into the right, right? But, if you actually zoom in, there are a lot of moments where we maybe hit a ceiling with a certain channel, and we have to go and find a new channel or we hit a ceiling with, you know, the creative for that channel, and then we have to find a new type of creative that works for that channel. So it's never, you know, up and to the right as it seems. You always have to kind of reinvent something and I think a big learning was what I shared earlier. Which is around velocity, the infrastructure that you have, and the things that you have that support, you know, $1,000 a month spend is going to be very different from what supports $1,000 a day or $10,000 a day or more than that. So you need to always make sure that you're upgrading the infrastructure around your channel. Which means a diversity of campaigns, maybe not just one campaign, a diversity of campaigns. It means a diversity of creatives. Because different creatives will apply to different users. A higher volume of producing these creatives so that you can have more chances at finding a winner. Because creative fatigue happens faster when you spend more. So we had to kind of get into those points where you hit a ceiling, and then we were like, OK, the playbook that got us here maybe doesn't work anymore. And we have to figure out what is the next playbook to get us to the next step. And it's an ongoing process.

Pablo Srugo (00:47:59) :
And where are you at now, ARR-wise?

Anada Lakra (00:48:01) :
We are around $10 million in ARR.

Pablo Srugo (00:48:04) :
And when would you say that you felt like you'd found true product market fit?

Anada Lakra (00:48:09) :
I don't think of product market fit as this on or off toggle. I think of it as a spectrum. For sure, if I had to pick a number, I think the one million ARR felt like, OK, this is not just a novelty, or we found a few people who really like this. But this is something that clearly a mass of people are starting to use and I think it was not just in the revenue metric, it's also in the usage metrics. Because we're not just having people who maybe have such a big problem that they're paying for it, but they're not coming back the next day. We also started to see really healthy usage metrics of people coming back, and then the qualitative feedback they were sharing with us about their improvements. So the ARR is a nice number, but backed by actual metrics, it was really powerful for us. But again, I think we're always trying to, the product is not done, right? We're now a couple of years out of that one million in ARR, and we're still improving the product, we're still trying to get that day zero to day one number coming back to lift. And I also talk a little bit about, making this less of an education one and done app, but more of an ongoing utility app. And I think that's the work to get more, and more into that one hundred percent side of product market fit.

Pablo Srugo (00:49:21) :
And was there ever on the flip side a time where things were clicking, things weren't working and you thought maybe this wouldn't work out?

Anada Lakra (00:49:28) :
Only about twenty times or fifty times, I don't know. I think if you're, as a founder, you're always in the trenches. Things are always going wrong in some way and, you know, I already shared that there were various moments where we hit various ceilings, and it wasn't clear how we're going to break out of that. And we figured out how to do it. And I think any founders, being honest, will say that the same happened in their business. But what's really important, I think, is to also know if, yeah, what you're working through is something fundamental, like the product has been resonating. We thought people wanted this, but they don't. We thought people care to fix and to improve their communication confidence and their accents, and they don't. If that had been the case, if that's what we found, I think we would have stopped. But I think what we found was things like I described. Which is more like, OK, this creative has kind of had its, run its course. We need to figure something else out and that feels like a solvable problem, of course with a lot of work and focused execution to take a lot of, experiments and shots on goal. So I think you've got to be realistic as a founder about what's a solvable problem and what is a more fundamental business problem. And today we've come across difficult but so far solvable problems.

Pablo Srugo (00:50:39) :
And last question, what would be a top piece of advice you would give an early stage founder that's still in that early phase, trying to find product market fit?

Anada Lakra (00:50:47) :
This is so cliche, but I think a lot of people get so caught up in building a startup and building a business, and finding the lawyers, and the cap table and the, I don't know, a million things that come up when you're building a business. And the only thing that matters is do people want what you're trying to sell them? And so really focusing on that and minimizing all the other distractions is super important. And I think that was the YC advice, is talk to your users. And I remember every time we'd go into those YC partner check-ins or batch check-ins. It would be, how many times did you talk to your users this week? So just keeping that in mind that at the end of the day, nobody's giving you gold stars for building a company. You really need to solve the user problem. So stay as focused on that as possible and also be honest with yourself about whether you're doing that or not. Because you can always pivot and yeah. You need to find that thing that the user wants and you also feel like, OK, this is the kind of thing where I could work on this for ten years and still feel excited to do it every day.

Pablo Srugo (00:51:52) :
Perfect, Anada, thanks so much for jumping on the show. It's been great.

Anada Lakra (00:51:55) :
Yeah, this was fun. Thanks for having me.

Pablo Srugo (00:51:57) :
Wow, what an episode. You're probably in awe. You're in absolute shock. You're like, that helped me so much. So guess what? Now it's your turn to help someone else. Share the episode in the WhatsApp group you have with founders. Share it on that Slack channel. Send it to your founder friends and help them out. Trust me, they will love you for it.