-5.png)
Just a few months ago, Grammarly announced the acquisition of Superhuman in a push to build out its AI for its productivity suite. Prior to the acquisition, Superhuman raised more than $114 million in funding from backers including a16z, IVP, and Tiger Global, with its last valuation at $825 million.
Gaurav Vohra was employee #1 and built Superhuman from the ground up — having led Growth, Product, Marketing, and Analytics. Now, Gaurav advises founders on growth with a focus on product led growth, working with fast growing companies such as Clay, Replit, Readwise, Wispr Flow and many more.
In this conversation, we discuss:
1. Gaurav’s background and journey to startups + Superhuman
2. Lessons from building a high-leverage RevOps function and transitioning into consulting
3. What early growth looked like at Superhuman—and why they didn’t A/B test for 4 years
4. Hiring the first 9 engineers through a hyper-instrumented, founder-led recruiting funnel
5. Superhuman’s “hard mode” GTM strategy: targeting CEOs, VCs, and power email users
6. The rationale behind their white-glove onboarding model and the decision to scale it
7. Creating intentional FOMO through waitlists, referrals, and Twitter virality
8. How Superhuman built a localized monopoly among high-expectation customers (HXC)
9. Why onboarding was led by ex-teachers, and how analytics drove quality assurance
10. The bottoms-up growth framework: fixing churn → activation → demand gen
11. Why paid acquisition failed, and what replaced it as the key growth engine
12. Building for team expansion, not just individual users, to achieve high NRR
13. Designing opinionated, interruptive onboarding flows—like video game tutorials
14. The “PLG trap” and why horizontal tools need clear vertical buyers
15. Gaurav’s #1 advice for PLG founders: fix your pricing and packaging first
youtube: https://lnkd.in/epz-6UBQ
spotify: https://lnkd.in/eDxfgfmd
apple: https://lnkd.in/e-BtVEP4
Cleaned up summary of our discussion for reading below too - 45 mins:
Charley Ma: Let's start with the basics. Walk me through where you grew up, where you went to school, and how you initially ended up at Superhuman.
Gaurav Vohra: I'm a small town boy. I was born in the UK, as you can hear from my accent—Manchester and Birmingham, not the big city, not where it's all happening in London. My upbringing was pretty sheltered. I didn't really see much going on in tech or any of that. It all seemed quite distant.
You probably know this—my brother Rahul, CEO of Superhuman, is fairly older than me, so I was fortunate enough growing up to learn from him in terms of tech, software, and computers. When he was a teenager, I was still a small child basically. But he was working with computers, programming, getting into building software, computer games, everything. I was soaking it all in like a sponge.
Over the course of my childhood, I got really technical through that influence and also through hobbies. It was a blessing in disguise that I was living in a place that rained 80% of the time. I didn't want to go outside, so I'd sit inside all day and play with computers and learn how to be really good at everything from photo editing to video creation to spreadsheets—you name it. Any software at the time I could get my hands on and become very proficient at.
I did weirdly end up in economics at university though. I didn't do computer science. Not that weird because I guess the standard path would have been medicine. Slightly left was economics. Far off the deep end would have been computer science at that time, demographically speaking.
Then it was very clear that tech was interesting and where I wanted to be. My very first internship was in my freshman year at a company called Redgate that did productivity software for database developers. If you needed to write loads of SQL in the late 2000s, early 2010s, Redgate was your best friend.
Charley: How did that even get on the radar? How'd you get that first internship?
Gaurav: It was weird because no one does internships in their freshman year. I wanted a valuable experience. Turns out my brother was actually working out of the Redgate offices in this little corner that they dedicated to startups. They had a tiny incubator—it wasn't even an incubator. It was more of a free desk and chair because they didn't get proceeds from the startups or anything. It was just "come chill here."
There were 10 startups, maybe 15 founders hanging out there. Rahul was one of them. This wasn't for anything that anyone recognizes today. Rapportive was born out of that office, but only two or three years afterwards. He was working on some five startups before even Superhuman.
By the way, that's also where his co-founders for Rapportive were. And it's also where Conrad, who's our CTO at Superhuman, was hanging out at the time too. But I was interested in an internship. I applied to one or two in London, and he said I should also apply to Redgate. He introduced me to the sales manager there, and the sales manager brought me on for a PM RevOps role. It wasn't called that—it was called analyst or sales analyst.
But it was basically RevOps PM: build software processes, systems, functions, and do data analysis to make the sales team make more money. I was deep in it with the renewals team, driving renewal, upsell, reactivation, all that stuff. And I learned SQL, which was amazing. Learning SQL was the best thing ever.
Charley: That's a great first internship.
Gaurav: What a banger of an internship. Six things checked off in one: productivity, SQL, sales, RevOps, PM, engineering. It was great. I was like, "Cool, I should definitely be in tech, but I also want to learn about businesses." This is where I ended up doing a little bit of the vanilla path. I ended up doing management consulting for five years.
But being someone who showed up on day one with pretty good SQL, they staffed me on technical projects from day one. I had a compounding advantage on my skill set where I was probably the most technically capable out of the entire London office when it came to SQL and building tech. At the time, we didn't call them ETL flows, but it was basically a serious investment in ETL and data stacks.
There was this team coming out of the Boston office at Oliver Wyman that was really tech-focused and was running servers for clients and building front end and creating web apps. They were basically internally recruiting all of the technical people, myself included, to build more software.
That was awesome because they put me through a dev bootcamp style program where it was "here's how to be a PM and here's how to be a front end software engineer." You're going to learn these skills so that you can either do it yourself or work with engineers on these projects. I was building real software for clients. Some of my clients were even tech companies. And I was like, "What am I doing? I should be in tech."
I convinced that firm to move me. At the time, my girlfriend, who's now my wife because she worked at the company too, said, "You got to move us to SF. You either move us to SF on an L1 visa or we're going to probably quit and go ourselves somehow." And so they did.
It was about a week after getting here—January 1, 2015—I'm hitting up my network of people to chat with, including my brother, obviously. He hasn't started Superhuman at this point. He's in the process of incorporating the company. He's finished Rapportive, sold to LinkedIn, exited LinkedIn. He took a year and a half off to recuperate his energy and relax. Then he was thinking of Superhuman.
I asked, "What are you up to, man?" And he said, "I'm actually starting another company again in email. It's called Superhuman." He showed me the pitch deck that he raised the first check off, the seed fund or pre-seed even. I was like, "This is amazing. This is what I want to work on. It's the thing that I have a ton of experience in. Email is such a core part of management consulting. I'm doing it for five years. I'm super passionate about productivity, I'm super passionate about building software. I can be the company data person, the growth person, the commercial, whatever. I haven't done tech startups or marketing formally, but I can learn that stuff."
Basically we agreed that I would found and build this company from the ground up together with him, with a growth focus being what I was primarily responsible for. There was one period where it took flipping forever for me to get my visa to actually start. So I was doing evenings and weekends for a solid 14 months. But anyway, got myself in and very rapidly became the growth guy at Superhuman.
Charley: Everyone throws around growth and distribution and GTM. When you first joined, what did that actually entail? What did growth mean and how did that change and evolve as the company started to change?
Gaurav: My internalization of growth was "what do we need in order to grow?" This is such a high level question. Honestly, when I advise startups these days, it's the question I often lead with if things come to an end or we're coming to a lull. I'm like, "Okay, so what is getting in the way of growth? What do we need in order to grow?" That's the single question that can kick start so many downstream discussions.
I would always ask that. You're looking at things like strategy: What is the pricing model? How are we charging money? Or is it free? If we're charging money, how much are we charging? Is there a free trial? Who is our customer? Why would they pay? Do they have the money to pay? How long are they going to stick around for? Are we viably going after someone who's going to pay for a long time?
At the point where most of that was figured out, the next thing that's getting in the way of growth is not having a product. We needed to actually build and ship something. So I was like, "Man, I can do marketing stuff. I can be on Twitter talking to these prospective customers. But truth be told, we need to hire engineers." So let me apply my growth brain to the problem of hiring engineers in this very competitive 2015 engineer market.
You've got companies like Instacart and Uber absolutely vacuuming up all the good engineers. And we're like, "Shit, how do we get good engineers to come join this thing?" So anyway, long story short, I got nine engineers to join in about nine months.
Charley: How'd you do that?
Gaurav: It was an aggressively tracked, instrumented, and optimized sales funnel. It's basically you've got a top of funnel—how many impressions are you generating? And then you track every damn step along the way: people responding to your first email, people getting on a first interview, and then all the way through to offer accept.
It wasn't a process of sales and closing, obviously. I had to do the vetting. I directly did the vetting myself: phone screens, on-sites. I was even vetting the vetters. I was pretty suspicious of some of our internal interview processes. I was like, "These questions suck. We need to improve."
It got to the point where we hit a groove. There were six channels that we were all over: AngelList, cold email outreach, in-person office events, Waterloo internships, word of mouth referrals, and—I'm forgetting one. Those were the six that yielded actual results for us.
The crazy thing is at least three of those engineers—one of them's still at the company. One of them did five years, left, and then I brought him back on a contract for two and a half years. He's an absolute crusher. Another one's a best friend of mine who I'm working slowly on making him come back for a contract here and there. Another one was Tristan, who's the CTO of Readwise, a co-founder of Readwise. There's a lot of great people in that cluster.
I was very glad to have learned how to hire because I then subsequently used that skill for the next 10 years. Hiring engineers was probably the hardest thing I've ever hired for. Everything else felt easy from that point onwards.
Charley: Going back to when you were first in that early strategy phase, trying to figure out customer ability to pay—how did you think about setting up the experimentation framework for trying to figure out who should we build the product for, what should the product be, pricing, all that?
Gaurav: We didn't really run proper experiments for four years. Everything was more of a pre-post, try it and qualitatively assess and see if it works, iterate based on qualitative feedback, and a lot of willing things into success. Having such first principles conviction around it that we beat down every objection and made it work for better or for worse.
We weren't product-led growth or A/B test minded for literally up until 2020 when we even considered running the first real test, and that's okay.
Breaking it down, Rahul had the vision of "we're doing this on hard mode. We're going to go after CEOs of companies with crazy bad email problems." Zach at Plaid was one of our earliest customers actually. He's actually very organized with his email. If anything, he's hard to win over because of how on top of it he is. But Will was another CTO whose backlog of email was way worse, way harder to be on top of.
So we're like, "We're going to do this on hard mode. These are our ideal customers." Also, by the way, VCs—they also have a shitload of email and they're the influencers of these founders. If you can get the VCs, then you reach hundreds of founders.
If it weren't for that, I think things could have gone really differently. If he weren't first principles pushing us towards that market, because I think we would have gravitated towards sales, recruiters, outbound professionals, BD, that stuff. Then you might build a very different product. You end up building Mixmax or Yesware based on those people's needs. But the VCs and founders have no need for long elaborate email sequences and templates and such. They want the fast thing that works really well. Whereas the recruiter needs the 5-touch email sequence. So different products.
We didn't not serve the go-to-market person or the recruiter. They can obviously use it, but we were hyper-optimized around this—we called them the HXC, highest expectation customer: the founder, the VC, etc.
Now imagine this: You've got really high profile people as the earliest possible customers. So we were like, "Shit, we're also charging way more money than anybody else. And there's Gmail, which is free—you can always go back to Gmail. So we really don't want to screw it up for any of these customers. We want to make sure it's going to be mind-blowingly amazing." 11 out of 10 Airbnb type experience, world class. A lot of inspiration from Tesla and Apple: start at the high end of the market, deliver the best, and then use that momentum and really the money to then deliver a more accessible or cheaper option for more of the masses.
That's where all of our super structured, super white glove, high touch onboarding came from. We were turning away probably 99 out of every hundred signups based on a series of indicators. If they were—we did a Clearbit lookup on every signup and if they met very stringent criteria, we would consider onboarding them and then have a dialogue back and forth to validate that they were ready to be onboarded based on the product's readiness. Only then did we go ahead and actually bring them in.
But by doing that, we created this corpus of insanely passionate first hundred, ten thousand customers that kept growing, but was so passionate about what we were building. And they were all speaking the same language in the same circles, talking to each other. That was really the first generation of the company's growth.
Charley: I remember you built this amazing localized monopoly across that very specific customer set. This is also a little bit before the era of founders on Twitter and everyone talking loudly and pontificating on productivity. It's right timing too, around that. Even getting access to Superhuman was almost a bit of a status thing at some point.
Gaurav: There were a few things that I think were of its time, probably wouldn't work—I don't know if it would work now. Rahul was very much the customer success, customer support guy at Rapportive. He was all over Twitter personally hand responding to every customer of Rapportive. That was something we emulated and copy-pasted into Superhuman. Me and Rahul spent a lot of time on Twitter and email manually responding to people in a very white glove way.
Could you do that still to this day? I think so. I mean, I think I see that with founders still doing it.
The next piece was the FOMO factor. The fact that there was a waitlist and then that people on the inside could refer people off the waitlist—that was the status factor that led to organic behavior on Twitter, particularly people sharing their referral code and then other people seeking a referral code.
We would tell people, "Sorry, you're not allowed to use this product. Go and find a referral. You'll find one on Twitter." That would create demand for referrals. Then we would create supply for referrals by advertising it in the product and making it a status thing. So then Twitter became the marketplace where that supply and demand equilibrated.
As an example, Nick Abouzeid was an early, early super referrer. He put his code out and advertised on Twitter and was like, "I have access to this thing. If anyone wants it, hit me up." I think he referred two or three hundred people in the span of a month. It's insane. So many people hit him up.
I don't know if that could still happen. The concept of a product as access to a product as a status symbol exists today. Superpower is one company that is doing a little bit of that with a waitlist and I see it with a few companies, but it's a very precarious and delicate thing to balance this idea of a waitlist. It's a little house of cards. You have to carefully prop it up and keep it feeling fresh and exciting. I think the max it runs its course is 12 to 24 months before everyone gives up.
Charley: When you started with that super hands-on onboarding and being really specific on the qualification criteria, was that started off as something where you said, "Hey, we are going to try to figure out how to scale this as much as we possibly can," or was it more, "We know we want to get our early customers, get them really excited, and at some point we'll figure out how to automate more of this"?
Gaurav: It was more the latter. It was more "nothing matters if we can't get people loving the product." JCal is an early investor and customer and sends a lot of email. If he has problems with the product, we got to damn fix them. One voice being unhappy was enough to have an engineer stop what they're doing and go and fix the issue.
The thing with email is there's so many issues to fix, especially if you support offline mode, you have a mobile app, you claim to be the fastest experience. Things like drafts disappearing, synchronization across devices, losing work, things being the same in Gmail and Superhuman. There's so many random edge cases that take a lot of time to get right.
Actually we saw our competitors—Polymail, Astro, there's so many other email startups that tried and died. They did the exact opposite, which was "let's scale PLG. It's self-serve, get on board." And they got tens of thousands of customers, many of whom paying. Their revenue curves probably looked pretty good for three to six months, but I think they probably churned 90% of that acquisition.
Charley: Some trigger thing doesn't get fixed and then it's "I don't want to use it anymore."
Gaurav: And it's one person experiencing a dropped draft actually means a hundred who don't tell you about it. But if we had one person experience it, we heard about it. And then we also had monitoring and instrumentation and stuff. So there was a lot that went into early R&D of "man, make this thing really good" such that we weren't even thinking about scaling because we were like, "We have to get to product-market fit and make this thing bulletproof and then we'll figure it out."
Now there were several aspects of the operation that we sought to scale as quickly as we could. Delight responses like "we don't have that feature, we're building it"—obviously build the feature, close those loops, make sure that users aren't encountering those frictions. But the onboarding was this bit that persisted for way longer than I think any of us internally expected.
The initial sense was we'll onboard the first thousand customers and then transition. It ended up being several orders of magnitude more than that. I wrote about this in the First Round article. I was in charge of a team of 25 onboarding specialists who did 40 calls a week, every single one of them. That was the revenue generation machine in the company.
That machine moves its way through customers at a huge clip once you're up and running. You've got the demand of customers coming in, the supply of onboardings. It's basically you're almost running a high velocity clinic, a doctor's office. People in and out, getting their experience, getting their productivity boost and then being successful.
The thing that I would probably say in reflection is we could have started down the journey of automation and scaling it sooner than we did. I think there were several times where we were evaluating "should we build this core feature or invest in this thing versus automate the onboarding." I think we chose the former probably a few too many times and should have gone for the latter probably sooner. But would it have been dramatically sooner? I'm not sure. I think we're talking 12 to 24 months sooner, but not three years sooner.
Charley: When you decided to start to scale out that onboarding team, what were some of the more critical pieces of infrastructure or processes or things that you had built in place to be able to feel confident, "Hey, let's start to open up that funnel pretty dramatically"?
Gaurav: The first thing was qualitative signal. I did the onboardings—I did 200. I think my count's 280 or something. Rahul's similar, 200, 150, something like that. And he was like, "Okay, people love this, man." I remember onboarding you and you were like, "This is so great." I went to the Plaid office and I learned a shitload about Plaid by being in the office. I saw Will and we chatted a little about Superhuman. I think I saw Zach on the same day. The learnings for the company and then the value for the customer—and this is everything I wrote about in the First Round article—enormous.
Then it's like, "Is that going to translate from founding team to hired onboarding specialists?" And the answer was yes. Especially teachers. Former teachers. Such a great, interesting category of individual. Such great talent to help onboard customers to a product. I think any solutions engineering, implementation team, any B2B SaaS that has a hands-on component should consider former teachers looking to get into tech.
Charley: I love that.
Gaurav: Because they understand things from first principles and they can break things down and they can understand where the end user's lacking understanding. Whereas a salesperson may or may not do that—good ones do. And then support or success—support's often a little bit timid, they don't want to challenge you as much. And success is probably the second best category I would say because that's what onboarding is—success, isn't it?
Infrastructure-wise we needed sick analytics. As soon as I'm no longer doing the onboarding, seeing the qualitative results, I need to see if these customers are coming back on day two and what are they doing. So observability of what each customer did was critical. And we had these goals around: Are these customers sending all their emails—a high volume of emails through Superhuman in the next one day, next seven days, next 14 days?
The onboarding specialist's KPI was the number of delighted customers that they brought on board each week. And it wasn't the number of calls they did. It was of those calls, which transitioned fully into Superhuman. So it was a volume and a quality thing. And it was like, "Okay, that's the success of onboarding. We also need to make sure the demand gen, the marketing side is working well enough to keep feeding onboardings in." Otherwise we end up with capacity here and no demand. But at the same time we don't want too much demand because then you end up with a big backlog. So that became some of the friction that precipitated wanting to shift over to self-serve.
Charley: How did you start to think about that delicate equilibrium between demand gen, new customer growth versus onboarding versus onboarding activation versus obviously retention and renewals? How do you start to think about that? How do you start to resource and think about the team that you build around that too?
Gaurav: So I was responsible for all of growth, which actually included all of that. And because I came up against this question all the time, I had a framework that I used, probably still use to this day, which is working from the bottom of the funnel upwards and making sure that I'm shoring up obvious gaps and problems in that funnel as I move my way upwards to the point where I've reached, let's say, 80th percentile of market in terms of goodness for that conversion step.
So if our retention is, for B2C productivity or prosumer productivity, 80th percentile, there's some headroom, but that indicates we've probably hit diminishing returns. It might be time to move to activation. If activation is 80th percentile, then it might be time to move to demand gen. So I basically did that. And the reason is that if you have a leaky funnel, obviously you're throwing water into a bucket and it's leaking right out.
But you also learn fastest that way. If you fix a problem in your activation, you will see the impact on revenue and downstream metrics way faster than the top of funnel. So there's a learning loop thing there too. And this is something I think small, tiny, constrained teams need to do. It's a pretty reliable way to think about the funnel: start at the bottom, churn, and work your way up. Get to a point—and each startup's terminal point is different. If you are going for volume and you're in a knife fight with a competitor in a land grab situation, your churn at 50th percentile of market or whatever is good enough to move on to demand gen because there's too much opportunity, there's too much volume on the table to sit on your hands. Then you come back to churn later on.
So where you stop is the strategic decision. For us we wanted to stop at a really good place with all of these metrics because of the brand, because of the promise and the PMF and all that and the price point. So even in some places it was more 90th percentile, even 95th percentile is where we want to get it to before we move on.
So that's how I thought about it. I didn't do much active retention or churn work. The core, the whole of the core product's team is working on retention and churn. I didn't do much growth-related work around that while I was working on activation. Meanwhile, once I'd figured out activation and the onboarding team and I had managers who were hiring the team and scaling that operation, I was almost entirely focused on demand gen and marketing. I fully shifted my attention to marketing. Let my mid-managers run onboarding. In fact I gave that team away after about six months so I could fully focus on top of funnel.
Charley: Now you have this streamlined onboarding process, you have the observability platform, you feel good. Now your job is "we want to 10x the volume because we have the capacity to be able to do so and start to create that funnel." Where'd you even start?
Gaurav: The starting point is double down on what's working. So the things that were working: word of mouth, casual contact through the signature sent via Superhuman, founder-led content or social media, the interplay of social media content. Rahul's PMF article. He did some live roadshow style things—Slush, the conference in Helsinki. So all of those things combined. PR was a big one for us, getting on the front page—the New York Times I think was for our Series B announcement. Those kinds of big PR drops.
Those were the initial channels that we were like, "We have to do all of that times 10." See how far those things go and how saturated they can get.
Then there was a couple of twists and turns as I think most marketing organizations face. By the way, I will candidly throw my hands up and say I'm not—I shouldn't be a head of marketing. I'm not a long term head of marketing for any startup by any means. But I can talk about growth and the marketing strategies to drive growth to no end, but to actually go and execute and run marketing, I think you want someone who is very attuned to the product you're selling and then also has obviously depth of experience in three or four key channels.
So we hired a head of marketing who invested a lot in paid and it wasn't the right move whatsoever. We dumped loads of money into performance and it didn't work.
Charley: How long did it take and how long do you think it should have taken to figure out that paid wasn't working? And when you say paid wasn't working, what does that mean?
Gaurav: It took about eight months. It should have taken about three. And it wasn't working—crazy CACs, man. Unreal CACs. And bad data, observability of that problem such that people weren't aggressively looking at it enough. The company was hiring really fast in several areas. So a lot of exec time was on the sales side of things, not on the demand gen side. It should have been more focused. And I went on parental leave. I handed over the marketing team in Jan. I went on parental leave in June. Came back in September. And shit was really weird. The whole marketing thing was unrecognizable.
Charley: What happened here?
Gaurav: I started building the analytics team and digging into the numbers. By the time I finally shined a light on the problem with the numbers, it was like, "Yeah, this is really bad." I think it was January when I ended up with the marketing team again. I was like, "Okay, back on my plate. Cool. First things first, let's stop paid. Not ready."
And the reason it wasn't ready was because we were still onboarding manually. And so paid relies on no friction, super fast, annual at checkout, ROAS immediate. That was one of the interesting learnings. But what we didn't see early enough, and this is another learning, is that the growth engine of teams is actually something we should have invested in as early as possible. Because yes, you can do performance marketing to reach more singletons. Yes, you can do SEO content, whatever. But the thing is, when you get accounts with NRR over 100, that is a fundamentally different and better business than singletons churning out.
Even if the terminal retention is really high, the fact that your cohort shrank, as opposed to grew, means that you're fighting gravity at a certain level. So the biggest channel, the growth marketing channel, that works from that point onwards, it was individuals wanting to bring it to their colleagues and pay for them. And it's like, "Obviously, I mean, that was always the plan. That's why we went after the CEO in the first place, because we wanted the whole fricking company to buy the thing."
So I'm like, "Okay, but we need to make the product really easy to expand to your team, add to my team flows, sign up organically, find teams when they sign up. Flows. Admin controls, role-based access and then security and reporting and analytics and such." That's all PLG stuff in the product. Then you can go ahead and find more logos. You start thinking about it in terms of "where do I find logos? Not individual, not people. Drop logos into that system" and that's been where the highest quality growth has come from, I would say.
Charley: I remember I think it was at some point during when everyone was cost cutting and I think it was even an email from Zach at Plaid where I was like, "Hey, if you're buying this software, you don't have approval for this software. We're cutting software spend." But Superhuman, that is an approved vendor. You can still expense that. And I was like, "That is genius." I think a lot of people think about PLG as one of the first things to get cut because these are small experiments or people trying things at the lower IC level. But by having that top down initial person or persona, it also gives you that exact buying because hey, the CEO finds it useful and relevant, increases their productivity. Should also increase productivity across the rest of the company too.
Gaurav: Yeah, exactly. It's both—and I think hard mode is any productivity software selling to this mythical Chief Productivity Officer that doesn't exist. Really what you're doing there is if you have something the CEO loves, then they can hopefully be the internal champion. But I will say this: products that solve specific problems for teams or indeed the whole company—that's where it's at.
If you can find that then you've really found a moneymaker because there is this thing called a PLG trap. Oliver Jay, who was a CRO at Sarna and then I think he was in a go-to-market leadership role at Dropbox, and then now he's leading international at OpenAI. He's very knowledgeable about PLG, having worked at all these places. And he wrote this series of essays called "PLG Trap."
If you're Trello or you build this great horizontal tool, but who's your buyer? And you hit the ceiling where those buyers are probably buying something that's arguably could be a worse product, but because it's clearly solving their problem. Jira as an example—a VP of Product buys Jira, not Trello. Trello is a better product. So then the argument there is, well, if you have a beautiful horizontal product, that's a good start. But figure out what your vertical is. Who's the VP? If you're selling to businesses, who will put their neck on the line, who will put their budget out for you? They will fight for you on your yearly renewal. That's what you got to figure out. If you happen to have started with a horizontal productivity tool, that makes sense.
Charley: Switching gears to the last bit—it's now starting to shift over to self-service. How do you think about taking all the learnings and insights that you have on the team, starting to scale that, bring that into the product, let the product do more of that? What was that feedback loop? What was the team that you built around that?
Gaurav: We also had a downsizing of the team. So the onboarding team was one where we were like, "We need to go from 25 back down to 12." And it was pretty painful and tough. But as part of that we were like, "This is ripping the band-aid to push us towards something more 50/50 immediately, self-serve white glove."
So the feedback loop was almost like, "Okay, we've got observability of the onboarding through our team. They can tell us how good the self-serve experience is, but then also we need to analytically be able to look at the other group of customers and compare them and evaluate how they're doing, what their usage curves look like, what their activation looks like." So that gave us an initial body of data almost immediately.
Initially the results were depressing to see the activation rates. Man, these people don't activate, they don't understand how to mark emails done. And yeah, there's no surprises because the product didn't explain anything. We were like, "Okay, what are we doing in our onboarding?" It's highly persuasive, opinionated, almost obnoxious. You have to speak to someone for half an hour. It's all of these things.
Self-serve onboarding has devolved into checklists and tooltips in most startups. So we were like, "No, man, this has to be—this is Navi in Ocarina of Time. This is your helper yelling in your ear if you're screwing up." Saying like, "You idiot, you missed the left turn. It's right there." So that's where I came up with this idea and I didn't coin it for probably another year or two internally. But this idea of being opinionated and interruptive.
You have to do that. You have to be opinionated, you have to be interruptive. Opinionated: you're not dropping people in and letting them figure out how to use it. You're telling them how to use it. And I wish more software was opinionated. Trello should be way more opinionated about—there's so many software tools that let you come up with your own crap way of using it.
Interruptive is checklists and tooltips—they don't interrupt you enough. They sit in the corner and act as clutter. I even did an onboarding review of ChatGPT for the team there—I know a couple of the PMs and there were four pop-ups that got in the way of each other. And I'm like, "No one designed this. This is not interruptive. This is random tooltips that are sitting and it's horrible."
So some of the even some of the biggest and best software products fall into this trap and I'm like, "Full screen. Look at an Apple onboard. Look at iPhone onboarding. It's Face ID. You have to flip and do this thing."
Charley: Can't skip it.
Gaurav: Exactly. Good luck. And then the pushback is obviously, "Well, that's no fun for users. You don't want to be obnoxious." So it's like, "Okay, well how do you make it fun and interactive?" Make it something they can play with. And then you think about games. Video games do all of these things beautifully. They drop you into the most complex environment.
Charley: Here we go.
Gaurav: Yeah, exactly. And you're playing the game while you're learning the game. So yeah, chronic video gamer. That childhood I mentioned growing up in rainy Birmingham—I'm playing a lot of video games and so channeling a lot of that energy towards the design of this thing to say, "Look, make it fun, make it interactive, teach it to them while they're learning a new skill and reward them as they go."
So that was the foundational philosophies underneath it in terms of the concrete building blocks. I mean, I wrote about some of them. You can experience them. It's things like the hands-on keyboard shortcut version of the game. It's Mavis Beacon. What's it called? The typing thing.
Charley: The typing, yeah.
Gaurav: I don't know, you'll have to look it up. So it's like, "Okay, that module, all right, get me to zero." It's this nudge that tells you to archive. It's nuking your whole backlog of emails. That's a super opinionated, super interruptive. Opting into AI or inviting all your team—these are things that are full screen. You basically have to do it.
So we started standing up these modules and chaining them together in a first run, a first sequence. You asked about the team, really? It was one PM—this guy Ben, who I called out at the end of the First Round article—and a team of design, analytics and engineers with him. Kaelin, designer. Lily, analytics. And then four or five engineers, pretty stable over time. And it's not a huge pod. Some core product pods are 10 engineers to 1 PM. This is four to five engineers. But they can ship a lot.
And I think the ratio has to be smaller because the PM has to be so close to what they're doing that it's a smaller unit moving faster to build and ship things.
Charley: Last question I have for now. We're obviously now in a post-ChatGPT LLM world. If you were able to go back and think about rebuilding Superhuman, building out the team in this post-LLM world, what would you perhaps change? What would you perhaps spend more time on or rebuild?
Gaurav: I think the core product would be way more AI powered to start and that would probably change a lot of the assumptions about what you need to opinionatedly onboard someone to. Because now the product can be opinionated. Now the product can read your emails and tell you what to do with them. So that was probably the foundational thing that would change.
I think that you'd also have an opportunity for an LLM to be part of that handheld onboarding journey somewhere. It's contextually aware of what you're doing, it's doing that part of that hands-on onboarding for you, a little assistant in the corner of the product. I'd probably investigate that.
And I think I shared some of the things that I would have done sooner, faster: transition to self-serve, transition to teams. I would get those things up and running sooner. Those are more reflections on good B2B SaaS in a way.
But yeah, I think in a post-LLM world the ability to have fuzzy inputs and fuzzy outputs but still be somewhat reliable is probably the biggest change that affects both core product and the go-to-market motions.
Charley: Cool. I think that's—we covered a lot. This is awesome. Anything else that you want to share or what's the most common thing that when a PLG founder comes to you and says you have five minutes, what should I spend time on? Where are you usually pointing them to?
Gaurav: Pricing and packaging, man. I always go straight to the pricing page. I don't even go to the homepage. I type their company name slash pricing. Let me see what's up here. Because there are usually so many mistakes and I was looking at one earlier today. I love the product but they charge less per seat on the team's plan than on the pro plan. And I'm like, "What are you doing? It's $12 down to $10 and I'm like, "It should be $12 up to $15." Teams are price insensitive. They don't give a shit.
And so it's like, "Let's unwind some of these things." Quick wins, man. These things are worth probably a couple hundred K in the bank right there. But then there's always the more sophisticated nuanced discussion of "okay, where do you put your core features, what do you withhold from the higher plans? Let's talk about the freemium side of free trial. Is that cannibalizing or is it good? And then what's the hop from free to pro to teams to enterprise?"
And so that construction is for PLG companies—I would say, 9 out of 10 PLG company founders I talk to have some opportunities by looking at the pricing page alone.
This interview was conducted by Charley Ma and has been edited for length and clarity.