Growth Hacking Is Dead. Growth Architecture Isn't.

Machine-generated transcript from the episode video. Speaker labels are not included and some names and product terms may be transcribed phonetically.

[0:00] I've learned that the best founders are always selling. Welcome to the GTM Vault podcast hosted by Rick Kleta. We uncover the strategies, the pivots, and the breakthroughs turning startups into giants. Let's crack open the vault and find out how. Welcome to the GTM Vault podcast. I'm Rick Ketta and today we're talking to Shawn Ellis, the man who helped build Dropbox, Eventbrite, and Log Me In into billiondoll giants, coined the term growth hacking, and now educates the next generation through his book growth hacking, his podcast, and Gopractice.io. Sean, your journey spans from hands-on growth leadership to shaping the playbooks that thousands of companies now use. Let's dive in. You were the first hire at Dropbox, Eventbrite, and Lookout, all of which hit billion dollar valuations. What did those early days have in common, and how has the playbook changed since then?

[1:01] Yes. Uh, thanks, Rick, for having me on. It's it's uh it's great to be with you. So, the um Yeah, I mean, probably the biggest thing they all had in common was they already had product market fit. um if uh if they didn't have product market fit, I mean with with Lookout, we had to we had to dial it in a bit, but uh if they didn't have product market fit, then I couldn't have been really effective in in helping them with those early growth efforts. So, that's probably probably the biggest thing. Um the other piece is that the founder CEOs all took a a super hands-on role working with me on growth. And it wasn't it wasn't something that they just said, "Oh, this stuff's confusing. I'll I'll I'll let someone else take care of this for me. They were they were really a partner in figuring out the engine that that ultimately took them to those big valuations. And you coined growth hacking in 2010. How do you refine that definition today, especially as teams now have AI and far more data than you did back then? Yeah. I mean, I think that it it's really around the fundamentals of growth. And so with more data, with more automation, with all the

[2:04] things that AI brings to the table, the fundamentals still don't change. that if you if you can't have a product that retains customers that that must have value that that we need for product market fit. If you don't have a smooth path for customers to experience that value and so kind of quickly before they give up uh if you're if you're not able to to retain those customers long term ultimately ultimately AI doesn't really matter that much if you if you don't have those kind of fundamentals. AI can help you run tests faster, can help you process data faster, and ultimately, you know, later on can help you can help you kind of do do more personalization at scale and some some other parts. But, you know, for the for those early days, it's really about building that early growth engine around uh around product market fit that's been validated. You've taught growth at top universities.

[2:59] What's the most persistent myth about scaling that even smart founders still believe? Well, yeah, I just had a call actually today and one of my with with a founder and one of my key messages to that founder was that most founders they they they are given some some numbers they need to hit from their investors in order to kind of clear those milestones to to ultimately scale the business. and they know that I'm not going to hit you 10,000 MR and then 100,000 and then a million MR if I am not able to acquire customers at scale. And so they immediately start thinking about how do how do I test and optimize customer acquisition channels and and in my experience that really comes last. you you need to have some flow of users to be able to to get your your engine humming. But the obsessing on on uh finding scalable customer acquisition channels is a mistake too early and many founders make that mistake. Dropbox became the fastest SAS company to $1 billion in ARR. What's one data driven

[4:02] insight from that time that still surprises people today? Oh jeez. Um there's there's so many you know pro probably probably the you know for for me one of the first things so there was seven engineers at the company when I joined I was the first non-engineer to join the team and and so yeah all I did was study data for for the first you know few weeks before I didn't there was a lot of things that were working well and I didn't want to break them. So yeah the the the important piece of data that I saw coming into it was most people were discovering the product through existing users. So existing users were either through natural word of mouth or through you know in inviting them to a shared folder were were essentially exposing the product to other people.

[4:46] And so uh yeah it's famous for the referral program but even before the referral program there was just some some natural word of mouth and verality that that were were helping the product a lot. So that was a really important insight. But another datadriven insight was that most of the people or many of the people who were trying to use the product the first time were were running into barriers to getting started with the product. And so we studied a lot about what was present pre preventing new users from from effectively experiencing Dropbox. And so a lot of that was through initially studying the data and seeing where where people are getting stuck and then running surveys to understand why they're getting stuck.

[5:30] and then using those insights to to drive experiments to get more and more people to a great experience more quickly with Dropbox. At LogMe In, you scaled a premium model profitably. What's the number one mistake companies make today when trying to replicate that model? Yeah, so the interesting thing with Log Me In was that we there weren't really any other premium models at the time in SAS. So, it was pretty early. There was there were a couple of companies that had free versions, but they they were less likely to be SAS products. And so I I think for us one of the one of the big things was we couldn't do it when we initially wanted to do it. Our unit economics did not support it. And so I think that's the that's the starting point. Do you even have unit economics that support it? So as a as a hosted software. So log me in is a software you install on one computer and you can remote control it from any other computer using a web browser. And uh our the way we were architected meant that those sessions were going through our servers. That's why why it was such an easy product to

[6:32] use was that we would we would essentially connect the two computers together through our servers. And that was expensive bandwidth to be hosting all of those sessions. And so it was my initial push that got us to do premium. But it was not possible until the engineers made a breakthrough on rearchitecting the product and finding a way to not keep the long-term connection between the two computers going through the data center and instead roll those over to a peer-to-peer connection. And so when they did that, they took our per user per year cost down to it was like $5 per user per year maybe when we first started and got it down to maybe 5 cents per user per year in terms of like bandwidth costs or might have been per month, but you could say it was just it was a massive a massive cut and and so you know by by the time I left there we had a 100red million active devices plugged into the network. So you can imagine if we were giving a hundred million away and we were paying, you know, $5 per device, it just that that

[7:36] would have not been a sustainable business. That that was the starting point was making sure that the uh the uh just unit economics supported it. And then and then the next piece was it was just really really challenging to uh to execute. I mean we we have a much lower value per customer than someone has a premium only model. we have a more compelling value proposition, but a lot of the channels like we we we focus a lot on paid search at the time. It's an auction and it meant that we we couldn't pay as much to get the attention of of customers in the first place. We did have the benefit of a lot of word of mouth once they started using the product. But ultimately we had to be really really efficient in our in our conversion and activation and and that's that was my first big focus on learning how important onboarding and activation were. And so when we initially tried to scale, we couldn't find paid channels that worked to scale that business. But we focused on improving that that new

[8:39] user experience that that new customer onboarding. And it was really a a full team effort because it was product, marketing, engineering, design, really all working together. But we improved our sign up to usage rate by uh by about a,000% in 3 months. Once we made that the focus once we had that really efficient activation, then we went back and and the channels now could scale really effectively. Log me in sold for it was about $4.6 $6 billion uh a few few years ago and and I think um we we would have been a very uh very small company had we not had we not kind of got all of those different pieces working. Your must-have survey is legendary beyond NPS. What's counterintuitive signal you're you look for to predict whether a product will scale? Yeah. So, so most people and I get I still get it from founders now.

[9:32] there. They want to focus on all the ones that get away. And so they they want to figure out like why am I losing people? Why why why do people why are people kind of lukewarm on my product? And it makes sense. Like that's that's how you kind of build your business to start with. You identify a problem. You solve that problem. And and and that's that's how you kind of get to product market fit in the first place, but you really hone product market fit not by focusing on the problem people. You focus on the successes. And that's what this survey does. It helps you identify people who would be very disappointed without your product. So you you ask, "How would you feel if you could no longer use my product?" And you give them the choice. I would be very disappointed, somewhat disappointed, or not disappointed. And you want to focus on the people who say they would be very disappointed because those are the people who really hold the keys to helping you build a big scalable engine.

[10:24] Those are the people who you have learned how to give them a great experience that makes it a mustave for them. So by truly understanding that and continuing to double down on the things that are working for them, you build a much more sustainable growth engine than by saying, "Well, what what are all these other people that are lukewarm? What do they want? How do I how do I adapt the product to their needs?" There's a good chance if you do that, you're going to break it for the people who it's already a mustave for. And so you make a product that's kind of good for everyone instead of great for the right type of people. And so that's what the survey is really powerful at doing. So the counterintuitive piece would be you know focus on the successes rather than the problems. You said growth is about high-speed learning. How do you build a culture where teams want to fail fast rather than fear it? Yeah. I think the starting point especially in the early days so that culture building becomes harder and harder later in the business like once you once you have a few hundred employees your culture is pretty fixed at that point. So you have a big opportunity really early in the business to to build a the right

[11:27] culture. And the initial starting point is just recognizing that nothing that you've created so far is perfect. That every single you know some people don't want to maybe move toward testing but they need to recognize that version one of everything you've done is already a test. It was it was your best guess at well what will work and the next test you run is an alternative that lets you know how close you were on that first one. And so when you open up that mindset that that nothing's perfect, that there's room for improvement in everything, you're just more likely to you're more likely to to keep seeking that. And so it's not about failing. It's about it's about learning. If you if you run a test and it's doesn't beat what you started with, then then that's not like a failure of a test. That's a reinforcement that you did pretty well on your first one. And then then you need to do another one and another one.

[12:17] And so you have to constantly seek a better way. And that's how you how you get into a position of being more successful. It starts with the starts with the the mindset that we're we're not perfect in anything. There's there's room for improvement in absolutely everything we're doing. I want to talk a little bit about GTM and growth in the AI era. If you were designing Dropbox's referral program today, how would AI change your approach to incentivizing shares or measuring verality? Boy, I don't know. I don't know that it would really I have not used it as much on the analysis and personalization side mostly because because you know a lot of the companies that I'm working with at this point are are super small still where there is not a lot of data to work with.

[12:59] If I was if I was working on Facebook I'd probably be leaning into that personalization a lot more and they can they can squeeze a lot more value out of it. for for Dropbox, you know, I think I think we we probably could have run faster tests in optimizing the program. And um you know, like one one of the things that uh that was really cool with Dropbox's referral program is that um people people didn't even necessarily know they were enrolled in the program. um they just you know some somebody signs up through a shared folder that you've set up and you just get a little notification that you just got some free space because they signed up and it's it kind of pulls you in where there's a lot of companies that would probably say hey if they didn't know why should I give them any free space but in our case we we thought you know that's a that's a a great way to introduce them to the program you've already earned this earn some more and and then a lot of people got kind of in that mode of of wanting to learn more and so but how did we message that like would there have been a better way to message that? I think with AI we we probably could have cycled through getting getting to optimal

[14:02] levels a lot faster, but but the truth is that we we we got massive acceleration off of the referral program pretty quickly as it was. So, I'm not sure it would have moved a whole lot faster, but hard to know. That's the challenge with anything that uh you know, even even when I when I lay out every decision we made on something, do I know those were the best decisions? There may have been a set of decisions that were even better that could have led to more success. There's definitely a set of decisions that would have led to uh a worse outcome. And so, but you you never know the path not taken. AI can now generate ads, emails, and even sales decks. Where should humans still own the creative process in growth marketing? Yeah, I the way that I personally use it, I don't I don't tell AI to do these things for me. I don't I don't program it to to to come up with tests. I think of it as a sparring part partner. You know, I'm I'll ask for 10 ideas and those 10 ideas will spark 10 more from me and then I'll I'll roll

[15:04] back and say, uh, you know, here here's two that I'm feeling really good about. What else you got? And it's just I I have found myself I don't think I go five minutes in a day now where I'm not using AI 90% of the time. within a five minute span I'm I'm using chatbt or claude or coding with lovable or there's so many new capabilities and and tools that are that are available now. You've seen companies stall after early traction. What's the most overlooked step when moving from zero to one to hyperrowth? Yeah, I think people probably don't spend enough time on activation. It's just, you know, retention. A lot of the stalling happens because of churn. you know, you you can have a fairly high churn rate and when you're when you're in those early growth days, the churn doesn't much of a drag on growth. But if you're turning 5% of your customers every month, when you get to when you get to, you know, a thousand customers, that means that, you know, I got to I got to add 50 customers just to replace the 50 that that churn. Like if

[16:08] you're in a B2B business and, you know, throw a bunch more zeros behind if you're in a B to Z C business. It's uh it definitely it definitely starts to it starts to add up if you if you got a high churn business. And so that's where and then most people when they think about having high churn, they they immediately go into what are the tactical things I can do to stop that churn. So how do I how do I rescue those people who are about to leave? But most of those people are about to leave. It's it's a function of we didn't give them a good enough experience to start with. We didn't help to build the habit early on. And so so much of long-term retention is really cohort-based. and how do I how do I get a cohort to experience the product in a way where where they're going to stay long-term retained. And so that helps you in two ways. One, you don't have that churn that's dragging your growth rate, but two, your unit economics end up getting so much better that your ability to acquire customers goes way up. That that suddenly there's a lot more ways to be able to acquire customers. Happy customers are telling other customers. You have the organic word of mouth engine. Really, that's

[17:10] that's the piece that I think the companies that plateau and and ultimately hit that growth stall is that they're they're not looking at growth as this integrated system and and you need you need to really understand the system before you can effectively scale the system. And so so much of the early days is around is around building that system and and then longterm it's around okay, how do we how how do how do we continue to optimize and scale that system? You've worked with iconic founders like Drew Houston. What traits separate founders who scale from those who plateau? Again, I think it's I think it's that recognition that uh that the the CEO has to play a big role in growth. The CEO, you know, growth growth is crossunctional. Growth is not about one one person who, you know, figures it out. It's about it's about all people on the team understanding that that growth system and and working their part of that system and coordinating each of those people and keeping them working in in in concert together requires a CEO who strongly understands that and uh and

[18:16] sets a vision and and sets a northstar metric. And it's not just simply, hey, I'm going to set this big target. You go out and hit it and they hit it. I'm going to set an even bigger target the next time. But it's it's more about understanding what is what is possible within the system you've created and being just really really diligent about you know how how do you how how do you where are the risk points and being able to hit the numbers that you want to be able to hit what's what's possible and where where's the where's the leverage to where we can we can accelerate things more and because it happens across so many teams that the CEO has to be a big partner in that process. Dropbox scaled with relatively low burn in today's tough funding climate. What's your rule book for balancing speed and efficiency?

[19:01] Yeah, I mean the good news is that's where I think AI really shines for for companies is that uh you know you AI gives everyone on the team so much more leverage. So you can you can do more with a team today of five people than you could probably do with a team of 20 people two years ago if they're really leaning on AI in the right way. And so I I actually think like seed funding will probably dry up as a class of funding and and it's really going to be much more of of growth funding. There'll be some some categories of products that might require more investment, but a lot of categories of products um will require such light funding that um that they could probably be self-funded. You interview hyperrowth companies on your podcast. What's the new pattern you're seeing in 2025 that didn't exist 5 years ago? I mean AI is obviously so much a part of every conversation that we're having now. So I think that's the but you know again AI AI will help you execute faster but the probably the biggest thing that we found not just in 2025 but across all of these years is

[20:04] that it product market fit is ultimately the the fuel for growth in every company that's growing quickly. So you really it all starts with a deep understanding of that product market fit and then building your growth system and engine around amplifying accelerating delivery of that product market fit. And so that's where that's where the AI piece helps much more. Some might even argue that that it can help you with with getting to product market fit. probably it can as well in the sense that you can you can prototype a lot more things a lot more quickly and and and and pivot and get validation at speeds where in in the past you know maybe had for for a certain amount of funding two or three pivots. Now you've got yeah 10 to 50 pivots that you can do and that that's going to increase your chances of of getting to product market fit m much more effectively. And and then once you have that product market fit, again it's it's about how quickly can you can you understand, construct, understand, scale, improve that that system, that

[21:08] engine that that ultimately leads to growth. And um that's I just I I think our book has has about uh 800,000 copies that we've sold now. And and so there's a lot more people that understand the fundamentals well now. And but I can tell you that even even if I understand the fundamentals perfectly, if I'm not working with product market fit, I will fail every time. And so that's why foundationally you need product market fit before any of the of the growth approach is going to work. And that's the hardest part for sure. But then once you have that, I think I think it's just it's it's just like speed and leverage is probably the biggest biggest change and benefit that we're seeing now with with fast growing companies. Sean, thank you for sharing decades of hardearned wisdom. For everyone listening, grab hacking growth. Check out gopractice.io and follow the breakout growth podcast. More thing too, I'm I'm actually launching a course in the next month or so. I'm opening it up. It's already been announced uh on Maven that is growth for founders. And uh it's it's

[22:12] really the idea that that founders do need to be kind of the center of of growth. And and so it's a it's a two-week accelerated course. And so they can they can find more about that on my personal website, shanellis.me. But uh I thank you for mentioning go practice, but I also wanted to to get that in there because for for founders, the go practice material might end up being a little too heavy. It's about an hour a day for 10 straight weeks where this course will be more accelerated and kind of build that baseline for for those founders. Sean, one last question. If you could only give one piece of advice to a founder starting today, what would it be? Understand growth. That's gold. Until next time, keep breaking out. Thank you, Rick. Tech founders and VCs, careers, lessons. GTM