# Why Product-Market Fit is Not Enough Revenue Does Not Create Usage, Metrics Must Be Before Strategy, Why it is Always Better to Concentrate than Diversify Marketing Channels and Secrets from Hubspot's Growth Engine with Brian Balfour @ Reforge 20Growth · Aug 16, 2023 · 58 min · 11,062 words Speakers: Brian Balfour, Harry Stebbings Source: https://www.996.fm/episodes/20vc--ep-eaaca419/ ## Cold open **Brian Balfour** [0:00]: Revenue does not create usage. To understand what your metrics should be, you have to understand the qualitative underpinnings. You have to have something called product channel fit. Then you need to understand channel model fit. Focus your firepower on fueling that thing as fast as possible. **Harry Stebbings** [0:18]: This is 20 growth ## Intro **Harry Stebbings** [0:19]: with me, Stebbings. Now 20 growth is the monthly show where we sit down with the best growth leaders in the world to discuss their tips, tactics, and strategies to starting and scaling growth teams. Today, we're joined by an o g of the growth world, Brian Balfour. Now Brian is the founder and CEO of Reforge. Previously, he was the VP of growth at HubSpot. And prior to HubSpot, he was an EIR at Trinity Ventures and founder of Boundless Learning and Viximo. He also advises companies today including Blue Bottle Coffee, Gametime, GrabCAD, and others on growth and customer acquisition. But before we dive into the show today, ## Sponsor read **Harry Stebbings** [0:52]: listen to this. **Unknown** [0:53]: Mercury has been a breath of fresh air. Getting started was maybe one of the most delightful onboarding experiences I've had. Mercury is just so easy to use. The aesthetic of it is actually quite relaxing. For me, it was less a choice and and more finding a kindred spirit. **Harry Stebbings** [1:09]: Imagine feeling this way about business banking. You could if you join more than a 100,000 startups on Mercury, the powerful and intuitive way for ambitious companies to bank. Start building momentum and leave the friction behind by visiting mercury.com/20vc. Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group and Evolve Bank and Trust, members of FDIC. And speaking of incredible products like Mercury there, the question, do you know the impact of your marketing efforts or how much you spent so far this month and what you got as a result? Funnel is the world's first and only marketing data hub that takes all the manual effort out of data collection and lets you blend data from different platforms to unlock new deeper insights. And the best part is this can all be achieved without writing a single line of code. That's why it's used by more than 2,000 of the world's top brands and media agencies, including Samsung, Adidas, Sony, and Havas Media. And now you can start for free with a plan that is not time bounded and allows you to see the full power of the platform without any time constraints. Simply head on over to funnel.io/20vc. That's funnel.io/20vc. You have now arrived at your destination. ## Conversation **Harry Stebbings** [2:32]: Brian, this is such a joy to do. I mean, I feel like I've interviewed, like, everyone from Reforge, and so I've been waiting for this show in anticipation. So thank you so much for joining me today. **Brian Balfour** [2:42]: Thanks for having me. I'll do my best to live up to all the rest of the folks you've interviewed, but we'll see how this goes. **Harry Stebbings** [2:48]: Listen, I had nothing but wonderful things. I want to start though with a little bit of an entry point, which is how did you first make your first foray into the world of growth as a starting point? **Brian Balfour** [2:58]: I made my way into growth where I would say like 70% of like the growth OGs made into growth, which was the early Facebook platform social gaming days. If you look at like some of the top people in growth, there's like a lineage, almost like a coaching lineage, like all the way back to the social gaming days. And so I had started this company shortly after school called Viximo, and we were in the social gaming space for a while, and then it kind of transitioned to mobile gaming. And that was like the perfect petri dish to create growth people. What the Facebook platform did was that it opened up a ton of viral channels, as well as like, paid acquisition channels. The games were very driven by like, product driven levers, like virality. It was a highly quantitative game, you were playing this high, like, arbitrage game, and games are so focused on the psychological aspect of users to a degree that other software products aren't. And when you combine all of those things together, that's really kind of what growth turned into, which was understanding how your product grows, which is different than how your business and company grows, and we can talk a little bit about later, combining it with the quantitative elements of growth, as well as with the psychological elements of growth. Those three things combined, especially in a high pressure, high competitive environment, is what really formed, I think, the like, initial injection into the growth scene. And so, as a founder of that company, you know, there was like nobody who really knew what to do and how to do it, and so a big part of a founder's job is just like, go solve the problems that you don't have other people for, and figure it out, and that's how I got into it, and I just, I loved it. I actually loved the game of trying to find an opportunity that other people weren't seeing, and figuring out how to exploit that, And so I just loved it, and I got into it. Been there ever since. **Harry Stebbings** [5:00]: I mean, it sounds like you should be a VC as well, but it is fascinating to see actually how many came from those early Facebook gaming days. So totally get you there. I do wanna ask, I think so much is gained in in hindsight and with years of experience. If you could go back to your first day prior to your first role in growth and tell yourself one piece of advice, cool yourself up that night before, what would you tell yourself knowing all that you know now? **Brian Balfour** [5:23]: Probably two things. One is like, there's not an infinite world of growth options. It's fairly well defined and constrained. And it's not like there's an endless list of ways to grow. There's actually a fairly set menu of things, and you can innovate within that menu, but it's actually pretty rare that a new dish gets added onto the menu. Knowing those sets of things helps really start to hone in on what your realistic opportunities are and not. And then the second piece of that is conviction and patience. When you look at a lot of the highest flying growth strategies, right, what is underneath them is what we call Air Reforge growth loops, where other people call flywheels, but they're basically systems that work like compound interest, right. They're things that feed itself and grow over time. And the really hard part about those systems is just like compound interest. It looks like tiddlywinks at the beginning, and it doesn't look like much. And so what that does is it puts a lot of pressure, and I see a lot of founders doing this, of like, oh, this isn't working, so I've got to move to this next thing, and this isn't working, and this isn't moving to next thing. When actually the right strategy is to have conviction in one of those bets, and invest your way through it, because compound interest, right, it pays off later. The hockey stick happens later, right? And so you have to look for early signals of like the fire starting and have conviction that as long as you keep adding like kindling and fuel to the fire, this little campfire is gonna turn into some kind of raging thing that you you can't even control. You **Harry Stebbings** [7:01]: mentioned that the compound interest and bluntly having conviction in a strategy or a channel. Yes. And I think content's a great example of that. It's about consistency and keeping going and keeping going, and it's a game of who can survive the longest. Sometimes things just don't work. How do you determine when something just isn't working and won't work versus consistency. Just keep going. **Brian Balfour** [7:23]: I think the mistake that most people look at, and the reason they kill things too early, is that they're looking at the output and not the inputs, and seeing like how those things are improving. And so, in the content game on SEO, the things that you're actually wanting to start to look at is like, well, is my domain authority increasing? Is my rate of like new pages starting to increase over time? And these are like the types of signals that as they grow, they feed each other, and things like start to compound. But this is also kind of gets to probably the biggest mistake I see across all of growth, which is that before anything, whether we want to talk about metrics, or this question, or all of that kind of stuff, you have to have a hypothesis about how your product grows, which is captured in what we call as like a growth model. And this is very different than a business model. A business model or a financial model is saying, when I put a dollar in, how do I get a dot more than a dollar out? And what a growth model says is, when I put a user in, how do I get more than one user out? And the levers between those two points, a and b, are actually very different than a financial model. If I'm looking at a product like, I don't know, let's call it, Loom is one I was an early investor in, and advisor in, right? Money does not help that company grow that much. The things that are helping that product grow are its organic viral and content loops, right? I record a Loom video, and then I send it to somebody, and then that person kind of picks up the product, and they start using it. Understanding what that system is, what that loop is, I can then start to identify, well, where are my biggest constraints in that system right now, and how might I start to unleash these constraints? So an example in Loom's case is like, hey, I could onboard somebody onto a use case where I'm recording a video and sending it to one person, like a one on one communication, But I could also educate the user on a use case like company communications, like I use it at Reforge, right? And that's a one to many use case that has a lot more virality built into it. And so I can start to influence like these steps of the system by understanding what the system is, and where the constraint in the system is. The constraint also tells you who should you hire, and when should you hire that person, as well as what metrics to track along the way. But getting back to your original question, how do I know something is not working? It's not about the output, it's about understanding the system, and whether or not you're able to continue improving those levers in that system, and putting it on a trajectory that if I keep moving those numbers, if I keep improving those inputs, at some point, the system goes from me having to add a ton of manual energy into it, to it starting to act like compound interest, and it like evolving itself. And so playing that out ends up being the fundamental key part of the equation to understand like when to kill something or not. **Harry Stebbings** [10:27]: Brian, can I ask you? I think I I mean, I'm an early stage investor. I work with many, many kind of especially from PLG SaaS companies, very similar to Loom in terms of kind of go to marketing business model. And I think all the founders would say they understand the machine and the system, but they don't know what the constraint is. What do you advise founders who don't know what the constraint is? **Brian Balfour** [10:47]: I would first go back, which is I actually think a lot of people don't understand what the machine is. And the reason is is if you ask them to draw you a picture that answers how does the product grow. They'll either struggle to draw that picture, or it'll look like such a jumbled mess that you, as the recipient, can't really understand it. That's actually a signal that not only have they kind of boiled it down, that they understand it, but they're not able to communicate it in a way that the rest of the team understands. And then the rest of the team doesn't understand it, that's where you run into these problems like, oh, we're gonna, you know, come up with these 10 different tactics, try them all, hope one of them works, because they don't really understand how things like, actually like map to the equation. But understanding the constraint ends up coming down to the quantitative part of this, right? The exercise I just mentioned is more of the qualitative part, is like, hey, can I draw a bunch of boxes and arrows that say, as a new user enters the system, they walk through these four steps, and if they complete these four steps, another new user comes out of it, right? That that's kind of what that picture looks like. The quantitative is now taking that picture and saying, okay, this is the metric that maps to this step, this is the metric that maps to this step, this is the metric that maps to step c, right? And then saying, okay, where are we performing at on these metrics right now? And then you need to go through the exercise of, okay, well if I move this part of the system from A to B, what happens to the output, and how does that differ between if I move, you know, metric b from a to b. And what that starts to tell you is like, well the system is more or less sensitive to one or the other, and then you need to layer on how easy is it to move these levers as part of the system. Depending on what the loop is, there are tactics that you can use to actually like pressure test this. The easiest one that you can pressure test are any things to do with paid acquisition and paid loops. Because what you can do is you can run what you call spike tests, Where you basically take something that's like at a low volume, and in a twenty four or forty eight hour period, you basically ramp up the volume as much as you can to understand where the system breaks. Right? Like what starts to perform less, right? So if I'm spending a $100 a day on a paid acquisition loop, for example, it might look like it's performing pretty well, and then I ramp it up to 200, 300, 400, 5,000, 10,000, right? And at some point, the performance breaks, and the performance will break in, you know, your performance of your creative, in the performance of your sign up rate, some something down funnel, and so you can like, really understand, like, where is the piece that that's kind of breaking the most. That's easy to do in some types of systems. It's much harder to do in other types of systems, like content loop, that organically take a much longer time before plotting it out on the curve, the base of the curve, before the inflection point, looks much much longer, right? Because there's just like less levers that you can dial manually as part of that. But understanding the constraint once again comes down going from your qualitative system to mapping it quantitatively, Starting to understand the sensitivities of all these levers, and then essentially trying to understand things that you can try to understand how hard it is to essentially move these levers. And the combination of that starts to give you the indication of this is where the constraint is at in the system, and this is like where we need to focus. My last point on this is that the constraints actually tend to follow patterns depending on the type of product you're at. So, very commonly with the product we were talking about, Loom, which is like a horizontal product, the constraint is activation. And the reason the constraint is activation is that the biggest strength of horizontal products is also its biggest weakness, is that it can be used for many, many different things. Right? So when a user is onboarding into the product, it's actually really hard to get them and understand what is the ideal use case to activate them on, like what is the thing that is like most relevant to them. But if you do activate on them, and it's somewhat viral like Loom, the fact that it can be used by so many people has such a spread rate that that's kind of like the easy part of the system. It's very different than a vertical product, which is very targeted at a single persona, or a single market, or a single use case. I actually know very specifically what I'm activating them onto. And so if I get them over that hurdle to like signing up or buying that product, it tends to be easier to activate that person, right? So it follows patterns based on the dynamics of the system and the types of products that you're at. And so these things end up being fairly predictable when you've seen enough of the game play out enough time across enough companies. **Harry Stebbings** [15:35]: You mentioned sensitivity of levers there. I have a contrarian thought on growth hires and timing that everyone says you should hire growth after product market fit. And I say you should hire it before because you need to throw enough people at the machine to know if you have product market fit. You need enough data to understand if some segment of people liked it and a growth high can help you get that data to understand. Do you agree with me or do you agree with everyone else that growth is a post product market fit higher? **Brian Balfour** [16:07]: I agree with you that you have to have enough volume coming into the system to really understand what these levers are in the product, and what you can play with, right? And sometimes, pre product market fit, you need to hire a dedicated person that's just focused on bringing volume into the system. Sometimes you actually don't. So to give you an example, like, I was to start an amplitude competitor, a new analytics product, right, that thing's not viral at all, right, like you're buying that thing for tens of thousands of dollars, there's a whole like setup cost, like all that kind of stuff. Stuff. There's probably a team that's really focused on building that product and building that engine, you know, engineering, and that growth model's going be very marketing driven versus product driven, and as a result, you need a totally different skill set of person required to focus on bringing volume into the system, so that you can understand if what you're building is on the right track or not, right? Use a counter example, which is like, if you're working on like a new consumer social product, as an example, or even something like Loom, where the system, the machine, and the loops are actually very product driven, that person actually looks very close to the early product and engineering team that you're hiring, and those things are driven so much by virality that if you're not getting the viral loops working, well then hiring a different type of growth person, a marketing or some other thing, it doesn't fucking matter. You're dead in the water anyways. Right? So I think the answer to this comes down to what do you actually mean by a growth person? Like, what type of profile does it mean? What is the machine and the type of machine that you are building? And then three, like, you can start to match, well, do I need a person dedicated to that pre or post journey, right? Loom is the example where I would probably wait till post product market fit to start to carve out, like, a dedicated bro product growth team, dedicated marketers. Right? Those those types of things. Whereas, like, Amplitude competitor is one where it's like, hey, I need somebody dedicated on this from day one to drive some volume to understand whether or not what the hell we're building is even close to on the mark. **Harry Stebbings** [18:22]: It's a shit question, isn't it? I mean, what a bad question. We're all for interviewing. I I totally agree with you. It's totally company dependent and it sets you up for a very challenging answer. My question to you is you said about post PMF there for Loom. You said before in some of your work that product market fit isn't enough. What does that mean? Because founders chase this product market fit. Like, what should it be and how **Brian Balfour** [18:47]: could it be changed? Product market fit is not enough to build a venture scale business, meaning like a 100,000,000 plus in revenue within some reasonable time period. And so I think there's like a few few pieces of this. One is that, I think one of the mistakes is people think about product market fit as a binary thing, and it actually lives on a spectrum. There's two vectors of product market fit. There's how strong of a fit I have, and with how big of a market. Right? And there's some kind of efficient frontier on that graph of like, you cross over, and it's like, it's strong enough with a big enough market. We're on the venture viable path. But there are companies that certainly live with very strong product market fit, but with a small market that just aren't venture viable. And there are actually, I've seen companies that have a very large market, but weak product market fit, those are also not venture viable, but in neither of cases, they are viable businesses in the grand scheme of things. This little tiny world that we live in around venture is not part of it. So that's the first part of it. The second part of it is that you have to have something called product channel fit. And so the mentality before was like, ah, I get product market fit, and then I'm gonna like, bolt on a bunch of distribution that'll play off. But that is not true. We cannot mold channels to products, We have to mold the product to the channel. And what I mean by that is Google, Facebook, Apple, whoever owns all these huge distribution platforms, they do not give an f about your product. They are the ones determining the rules of the game, and so your product has to play to those rules, right? That has to mold to those rules, right? And so if you just build your product in isolation of this concept, what you end up with is like a total mismatch, and this gets back to understanding your growth system. So if your growth system is gonna be very content driven, like a Pinterest or HubSpot was very content driven with like their SEO, like all that kind of stuff, You basically have to mold your product to fit with that channel. You have to be writing the content that's gonna rank, or your users have to be generating the content that's gonna rank, like all of those types of things. You have to instrument your product in order to adapt to that channel. The second part of it is like once you have product channel fit, then you need to understand channel model fit, right? Which is essentially saying all channels don't fit your monetization model, right? So you can't use viral loops for a product that you're charging $10,000 upfront for. It just doesn't work. It's weird to share that type of product with somebody, but even if I do, the conversion rate on that is like so low that that system, that loop, it like never actually really gets spinning. And so what you see is that there's channels like virality, and UGC content, and paid acquisition that work with lower friction, lower priced products. And there are channels, like different variations of sales and other things that work with higher priced products. And so you have to match your pricing to the channel to not only match the friction, but also match the economics. Right? Because if I'm using a bunch of sales, for example, right, like I have to make sure I get my money back on the cost of that human touch. So that that piece has to work too. **Harry Stebbings** [22:02]: But when we say about pricing channel fit there and aligning the two, would an example be like a luxury brand having a core channel in first class British Airways or Delta? High price point, high luxury good. Is that what we're doing about? Because like, does that mean enterprise products should not do YouTube and Twitter? Do you know what I mean? I'm just trying to understand what that kind of means then for the high priced products. **Brian Balfour** [22:26]: You know, in your example, an enterprise product, should they not do YouTube and stuff? It's more about what the purpose of YouTube and stuff is driving for that company. Yes, it might be driving awareness and these other pieces, but that alone is not going to drive customers for an enterprise product. And even if you were doing it, if you looked at the overall mix of like, at scale of what's driving customers, it's going to be a pretty, likely a pretty small fraction. And so it kind of gets to, in an enterprise product, you have to make your sales machine work, and if you can't make your sales machine work, everything else that you're doing is just pure noise. And so you've got to get the core machine working, and then you can add these other things that help accelerate it, as an example, but where, you know, the focus piece of it is, I can do a bunch of these things, but it's actually pulling attention away in the early days from actually understanding and getting that core machine working. It's very different than I think a company where, like, what you're seeing is, if we think about a product that, like, Riverside is actually probably a good example of this, the product that we're on right now. I am naturally creating video, so I should probably be asking the question, well, where are all of the different places that people are consuming video? YouTube is one of them. And so then the question becomes, well, how do I make it really, really easy for my users to essentially post video to YouTube, potentially, you know, with my branding on it, maybe they can pay to like, take the branding off of it, right? That all of a sudden starts to look like a really interesting growth system, because I bring in a user, they're gonna record a bunch of videos. If I make it really easy to distribute to YouTube, and there's some exposure to my brand on it, that's probably gonna bring in more customers. That starts to look a lot more interesting. **Harry Stebbings** [24:11]: Two things that can actually come from Kit Bodnar at HubSpot related to channel. He said on the show before that one channel working really well will get you to 50,000,000 in ARR. Mhmm. Two channels working really well will get you to a 100,000,000 in ARR. First question, what do you advise founders in terms of channel selection? How to know which channel to choose? **Brian Balfour** [24:33]: So I agree with Kip. Once you figure out kind of that core, that thing that's working, it's almost always in a venture scale company, the better thing to do is to focus your firepower, your limited attention, your limited capital, your limited talent, on fueling that thing as fast as possible, which is very counter to the, like, lot of the common advice, which is like, oh, you don't want to be too reliant on one thing, you want to diversify, like, all of that kind of stuff. That is a bad investor's point of view, right? Like, yes, that might work for money management, right? It does not work for growing a business, because it ignores all of the complexity that comes with diversification of products, or channels, or any of these types of things. It's like, if you find something that's working, you better focus that firepower, but the second piece of that is that you then need to anticipate when that one thing might start running out of fuel, you're not like, caught with your pants down in trying to find that second thing that Kip is talking about too late, because that creates this stall out effect and is really hard to like get yourself going from a stall out effect. **Harry Stebbings** [25:44]: If we just dive on that, do you think it's obvious when channels are starting to deteriorate in effectiveness and growth? **Brian Balfour** [25:51]: This is, I would actually say, even harder than understanding your constraint in the system. Essentially, what you're trying to do is predict a point of saturation of what you're doing. It's hard on multiple dimensions. One is that I think the tendency is to actually think something is going to saturate much more quickly than it actually does, and that's because if you look at the really high growth companies, you know, taking HubSpot as example, along the way on content, their path to really establishing that content loop, they hit ceilings along the way. Right? And it wasn't like they stopped at that first ceiling and been like, oh, throw the hands up, like, we're done. Like, every step of the way, there was some new idea, some innovation that unlocked another pool, right? And so going beyond, digging beyond that, is really, really hard work, because you're basically trying things that others aren't trying, and that's like actually just very hard for both people and a company to do. So I actually think predicting saturation is one of, like, the hardest things for companies to do. HubSpot, I would say, is actually one of the biggest things I learned there is they are probably one of the best at the game in the game at this, both from a growth and product strategy perspective. If you take into account, they have not missed a earnings since they went public, which was like eight years ago. They they have never missed an earnings. A perfect record, as far as as far as I know. And that was one of the biggest things I learned. I joined eighteen months before the IPO, and we were already talking about a five year exercise of saying, like, where do we want to be five years, and how do we work our way backwards to what we should be doing today? And where that got us was, hey, our single marketing product, and our single marketing and sales motion is not going to get us to our five year goal, which was maintaining 50% year over year growth for five years. And what that led the company to do is say, oh, we need to, like, kick off a multi product strategy, and actually finding the second product is probably gonna take us a year or two to do, so we'd better start that, and we'd better start that now. And actually, the second product might not even get us there, so we're going to need a third product. So not only do we need to plant those seeds, but we need to come up with a system to start to find new products that we can expand into. And so a lot of it comes down to looking into the future, playing out your growth today, and saying, okay, well, if I wanna maintain some sort of growth, what does that mean a year for two years from now? And then you gotta start going through the exercise of working your way backwards of like, what has to be true for that to occur in our system? And what starts to feel like really unrealistic around those things? What does feel realistic? That's just a really hard exercise, I think, to go through, especially because you don't do it that often. And so it's like one of these like low frequency, low rep things, and there's just not that many people in the world that has a ton ton of reps at these things. **Harry Stebbings** [28:56]: If we get to that critical inflection moment of predicting saturation in a channel, what do we do then? Do we go, right. We're predicting saturation, and we're gonna divert a little bit of resources and really try and carve out a content strategy, a SEO strategy, a you name the channel. Or do you say, hey, we've got one channel that's saturating. Let's split the resources between four and see which one springs with hope. Which one is the right approach? **Brian Balfour** [29:24]: Biggest mistakes of this are one, not starting early enough. Two, underestimating the amount of time that it takes to get a new channel or new product going. And three, over resourcing it. And then four, essentially, this is an area where I would actually take a couple bets, but not like 10 bets. Right? So the way that we did this at HubSpot, when I came in there, my mission with a couple others was help establish a second product category for the company and a second channel. Go from the sales and marketing motion we had to more of a product led motion. We had those two missions to figure out. I did not come up with this, this really came from Dharmesh and Halligan, the two founders, which was, the philosophy behind it was, okay, we are going to start with multiple bets. We are going to treat each one of these bets like a new venture investment. We are gonna seed fund this. And so they, like, made us come and they would pitch for our seed funding for the idea, which was basically one year of funding for a small team, like a four or five person team. And then at the end of that year, and then Halligan JD, who was the COO, they would act like our board essentially, and we'd have to go do like board meetings internally to them every like month or couple months. And then at the end of the year, we'd have to come pitch for a series a. And so like the next year would be like equivalent to like a series a funding for a slightly larger team, and then so on and so forth. And then at some point, the bet gets big enough where it's like, now we have to integrate it into the core machine that's typically around the series b phase, I would probably say, like internally, and then things got integrated into the machine. But as part of that, they initially funded a few bets for the first couple years, and sometimes they would work, sometimes they wouldn't, and they wouldn't pass to their series a funding, and and it was kind of that natural evolution. But part of that I think is companies I I think probably the biggest one that kills these things is they over resource them. They over resource these things. The more people you have, it's like the slower these things go, the more voices that you have in the room, the slower that you iterate and change, like, all of these types of things. You wanna find a couple people, like a couple generalists, separate them off, protect them a little bit from the rest of the company, essentially let them loose with the right guardrails and checkpoints, so with a couple people. And look, we made this mistake at Reforge with one of our new product bets, and last year, I had to kill it, fully reset it, handpick a team of three people, and go forward with it. It's one of those painful mistakes that, especially with people who have something already going, they're like, oh, I've got resources, so let me just leverage that, and they fail to remember all of the lessons learned from the early stages, which is just like sometimes more people, more money does not help you find things faster. It's about going through the natural stages with their natural constraints to find things along the way. **Harry Stebbings** [32:15]: What do you do if this channel's working, but it's not driving back to the core? And what I mean by that is, like, I was with a company this morning, an enterprise company, and they've actually got a really impressive short form video execution, and they've done amazingly well on TikTok and YouTube shorts. But there's no obvious correlation to customer sign ups or to revenue. And so it's like, yeah. A ton of people are watching your videos, but what? I didn't really know how to advise them. Yeah. What do you advise when there's, like, a channel that's working in terms of attention but not obvious value derived back to business? **Brian Balfour** [32:48]: Yeah. It's adapt or kill. So we went through this at HubSpot. You know, the first product we built, it's now called referred to as the sales hub. At the time, it was essentially a bottoms up tool for sales folks that let them track their emails and contract opens and stuff, and at the time that thing was magical. We got to a 100,000 weekly active users or something, and then we looked at the user base, and there was a huge mismatch between a lot of the types of users that were showing up in that product, and the target market that was driving the core strategy of the business. The core strategy of the business and the marketing product was really around mid market companies, companies that were between twenty and two thousand employees, and we had high conviction on that. But when we looked at the user base of this, we were like, oh, shit. These two channels, these two growth mechanisms, yeah, we're getting some of that mid market, but there's a huge amount of these, like, very small business owners of one or two person shops, or like other pieces that were using this tool, and there was a huge disconnect between those two things. And so we had to go through this exercise of adapting. Okay. We've got some of these components working, some of these aren't. What if we held the market constant, like the mid market piece? What would we do to these other pieces around our product, and our channel, and our pricing to make the rest of this work? And so what we did is we adapted some of the product that we were doing. We redid our roadmap around some features for, like, the the sales leader and the sales admin targeted more the mid market. But then we had to change our channel, so we killed a lot of the paid acquisition. We kept a decent amount of the viral loops, but we added in more of a content motion. We layered in content and a little bit of inside sales, and then we had to change the pricing to match with that. So we raised prices in a couple areas, I think like minimum seat limits, all those types of things. And then the system started to work for that core market, which was the mid market. So I think in your case, there's this whole exercise of like, oh, there's this really cool thing. Do we have bets that like can adapt it, can mold it to what we're trying to sell and what our product is? Let's take a couple bets at that. But otherwise, honestly, with limited time, limited resources, you gotta kill it. Because what you're doing is you're stealing time and attention away from finding that core machine that does contribute. You're focused on the icing, not the cake, and you need to eat the cake first. So **Harry Stebbings** [35:19]: Speaking though, like, knowing what works and seeing the success in a channel, the decision around whether it's successful or not is largely made by the metrics that you see and how you interpret them. You said before that people often get metrics wrong. What do you think are the biggest ways that you see founders and growth teams really misunderstand or get metrics wrong? Well, first and **Brian Balfour** [35:40]: foremost is metrics before strategy. Your metrics are there to help answer the question, is your strategy working? They are not there to determine your strategy. You have to have a reasonable hypothesis of what your strategy is first, like, what is your core growth loop? Like, of these things. And then you can say, well, what are the metrics that we would use to measure that strategy, and what would success look like? So that's number one. Number two is, they basically go quant before qual. So to understand what your metrics should be, you have to understand the qualitative underpinnings. And a great example of this is retention. Like, setting retention metrics, for example, is 80 to 90% understanding the qualitative definitions of what is the problem that you are solving, what is the natural frequency that that problem occurs in your target market's life? And then you can start to match the metric to it. Like, should we be tracking this on a weekly active user basis, or a monthly active user basis? Active. What does active mean? Well, what is the behavior in the product that indicates we are solving the problem for this user? But those things are underpinned by your qualitative definition in understanding of, like, your user, of the problem, of the natural frequency. Whereas, I think most companies are like, we're just gonna track it on, like, a weekly active, or a monthly active, or, like, quarterly buyer case, like all that kind of stuff, and I'm like, well, okay, what is the problem? Does your user actually incur that problem that often in their lives? Right? But that ends up being the dynamic there first. I would say the third, especially among SaaS companies, is they focus on revenue metrics before usage metrics. So almost everything is defined on like ARR, MRR, my revenue retention, all of those types of things. But what people miss with that is that usage is what creates revenue. Revenue does not create usage. You have to understand the usage of your product first. Like, those are the things you should be focusing on, not necessarily the revenue metrics. This is a classic, like, sub example of outputs before inputs as part of that. And then I would say, finally, is like, a lot of people mix customers versus users, right? So there are dynamics in some SaaS products that are very different. Like HubSpot, we had a customer, but we had multiple users on that customer's account. And to understand how to move, grow the business, and how to grow the product, it actually wasn't about a usage metric at the customer level, it was about a usage metric at the user level, all right? This was very clear with something like the free CRM which we built, is like, we originally tracked weekly active teams, but it's like, okay, well, the way to influence a team is actually to find one of the users in that team and how we can influence their usage of the product. And, oh, by the way, there's two very different types of users. There's the admin of the CRM, and there's the end user of the CRM, and the way to move their usage is fundamentally different as well. And so just by focusing at the dollar or the customer level, you actually don't get deep enough into the things that you need to understand to actually grow the business. You're two levels high. You're too high in altitude. **Harry Stebbings** [39:00]: Speaking of actually growing the business, over the last few months, AI has been the hottest thing. And I'm just fascinated for you at the center of the world of growth. How do you think AI changes the way that growth teams operate and the world of growth? **Brian Balfour** [39:16]: This is an interesting one because I think it changes a lot and not a lot at the same time. I think the surface level tactics and systems will definitely change a lot. A lot of these analyses that seven years ago were sort of leading edge, things like finding what your habit and moment are, for example, which comes down to some form of a regression analysis, like those types of things. All those things are gonna get automated. Super, super simple. **Harry Stebbings** [39:43]: Sorry, can I just interrupt you? How is finding your moment automated? Like, is it not highly nuanced and contextual dependent on the millions of infinite options of what a product could be? **Brian Balfour** [39:54]: I think like today, it's already kind of automatically done inside tools like Amplitude and others, but I think there's another layer of that where the friction to do them doesn't require the same skill set that it required you eight years ago. But this kind of gets to the dichotomy that I was talking about. The things that don't change, that I don't think AI is gonna really help you with that much, is understanding like, all of these qualitative underpinnings of your product, of your use case, of your problem, of your natural frequency, and then the psychological levers you can tap into to move a user's behavior. AI's gonna have a really hard time in doing that thinking, but in addition to that, growth boils down to four things, which is you find an arbitrage opportunity, you use that to spark some type of compounding system, some type of growth loop, you optimize the crap out of that system, and then you repeat that, but don't wait too late, right, like, you try to anticipate the saturation, right? Like, those things do not change, right? Like, all of growth will still come down to those things. I think it's just a lot of the surface layer tactics and other components of how we do those things actually, like, will change. Maybe we get a fundamentally new distribution channel out of this wave of AI technology. If so, that is like a freaking field day for growth people. Because what a lot of people are focused on right now is like, oh, AI is going to take away traffic from Google. It's gonna reduce my marketing surface areas because people are just gonna be interacting with a chatbot versus a website, and I'm like, yes, that's true, but at the same time, it's gonna also inject all sorts of new chaos into the system. And chaos is actually good for growth people, because within that chaos lives these arbitrage opportunities, these new things that nobody else has figured out, and those sparks are what give new companies and new things life and hope. **Harry Stebbings** [41:57]: When you think about, and I think lessons learned from often mistakes, what is the biggest growth decision you've made that went wrong, Brian? And how did it change your mindset? **Brian Balfour** [42:08]: So my fundamental view is like, the only way to figure things out is to like, create. You can do some thinking upfront, but then you gotta get into like, the making, the creation of these things. And you learn through that creation of like, what's right and wrong, and then you go back to like, the start of the process and think a little bit. I get really frustrated with folks who think they can think their way to all the solutions, and I'm like, that's impossible. We we just need to get into the making of things. Right? And so as a result, I think a lot of people look back at your question that you asked, and they're like, well, what is something that I thought long and hard about, but I was wrong? And as a result, it like had a huge cost. I just try to view things as like, it's just being wrong is like a next step. I will answer your question directly. Okay. So my biggest mistake at HubSpot was I was in love with like, the virality and other things that we had going with the early product, but there was a disconnect with the core target market of the core business. I was pushing for like that virality. Kept going down that path, I created a lot of resistance. What I didn't see at the time was, yes, even though I had those things working, the fact that there was a disconnect with the core target market meant that down the road we would not be able to leverage all of the assets that we had already built up on the core business. All of the core knowledge around our content motion, our sales motion, our knowledge of the mid market, like all of those types of things. And so what that did is it probably left that product in a state that disconnect with the core business for a little too long. Luckily, I actually got vetoed over this, and we maneuvered the product and the business to a different direction. At the time, I disagreed with it, but now in hindsight, I'm like, I was the dumb one. I was stupid. And so like, that was a really bad decision. I think in the context of Reforge, I think we've probably made two mistakes. One was a few years ago, we transitioned from a transactional to a subscription model, a membership model. That had huge growth initially, like just massive growth. And that kind of put the blinders on us to whether or not we were also doing the things to make a subscription product sustainable. And as a result, we ended up being too late to following up with some of the products and some of the features that create the type of habitual usage that's required for a recurring model. And so we had to work ourselves out of like a stall point. So I think that was like a case where the metrics ended up misleading us, and we actually weren't looking at everything that was happening underneath the surface in terms of how users were using the product, how they were thinking about product, and how those things might end up leading to that future ceiling that we were talking about. That's a good point of like not anticipating the saturation. I've also done some stupid growth hacks, which we can talk about. **Harry Stebbings** [44:56]: Oh, come on. We we go before we do a quick fire, tell me about a stupid growth hack. I love that. **Brian Balfour** [45:01]: Well, early in the Facebook platform days, one of the apps or games that we had was this virtual gifting app, and where you could like send these virtual gifts. It was like dumb, but people enjoyed it. Right? And so at the time, through Facebook's platform and API is that when a user signed on to your app, there was essentially the ability to like auto message that person's friends, like up to 20 folks from that person. So what we did is when a new user installed the app, we would just pick, I think it was like 10 or 20 friends at random out of their friends list, and send them a virtual gift from them. This was still before AWS. This thing started melting our servers within like four hours. Like, was just growing, and so we had to shut it off. The problem with this, besides it being incredibly spammy, was that afterwards, we find out that we didn't think about this as like, when you're choosing somebody at random like that, in some cases, we were choosing like ex wives and ex girlfriends, and they were receiving like a virtual gift of, like, a rose from somebody. It created this confusion. And, like, after that, was like, okay, that was one of the dumbest things I've ever done. It grew. It worked. We figured out that arbitrage opportunity did not translate into a sustainable system and created some really awkward conversations in the world, but that was one of the fun ones for for a little we'll not do that again. **Harry Stebbings** [46:23]: Right. That is amazing. Listen, Brian, I would love to move into a quick fire round. I So say a short statement, and then you give me your immediate thoughts. Does that sound okay? Let's do it. So what tactic has not changed over the last five years? If you think there's one that continues super strong. **Brian Balfour** [46:38]: Word-of-mouth. If you can get any form of virality, if you can get one user to tell another user that has been around forever, will be around forever. **Harry Stebbings** [46:45]: What tactic has died a death? Oh. Wow. Many. I feel SMS. SMS people feel intrusion nowadays. **Brian Balfour** [46:53]: SMS feels the intrusion. However, it depends on how you define it. Like, in some parts of the world, like WhatsApp, for example, is just a massive growth factor for folks. I've looked at companies that complete their entire growth strategy is on WhatsApp. That type of direct messaging, I think, still works in some parts of the world. I think when we talk at the tactic level of things that died, like, you can basically tie it to some type of change in the distribution channels. Just taking something as simple as importing somebody's address book, as an example. It doesn't really work anymore because there's been so much misuse of that that people are skeptical and fatigue and like all that kind of stuff. It's like, you can't really do that unless you've like built up trust with your users and have a really good reason. Who do **Harry Stebbings** [47:39]: you think is the single best growth practitioner that you've worked with? I know that's so hard to say given Reforge. Oh, jeez. But who would you say is come on. **Brian Balfour** [47:49]: Okay. I'll name a few on the different dimensions. Strategically, be Casey Winners. The best person that thinks about the psychological elements of growth would easily be Darius Contractor. And then the person of very specialty, who I consider like a mad scientist, I call him g, but it's Guillaume Cabin. He was at Segment Drift, like a bunch of others. But those three on different dimensions is is who I would probably name. What do you think is the biggest misconception that people have with growth? Growth does not mean one thing. As we've talked through, you have to understand the different types of growth systems out there, the ones that work with what types of businesses, and then how do you map talent metrics and other things to those systems? And so I think for a lot of cases, people just, they talk about growth as it's like a single thing, and that's like trying to say engineering is a single thing, but there's so many different types of engineering that if you were trying to describe the discipline and practice of engineering, you just like lose all of the substance. **Harry Stebbings** [48:55]: Will AI change the tooling incumbents of growth teams, or will they enhance the existing incumbent? Both. Who's most strong, and who's most vulnerable from the incumbent side? It's hard to **Brian Balfour** [49:08]: not say Twitter right now from a vulnerability perspective. Strongest? I think this is a little topic du jour, but I I think you gotta go Facebook for a few reasons. They still have the OG growth talent that they did from fifteen years ago. Javier, Alex Schultz, right, like, those people are still there. And so you got that. You've got the capital advantage. You've got the advantage that you can leverage billions of users on your other platforms. They still have like the most data. So I think you put all of those things together. I would probably put them in the strongest point. I think they're probably like the least vulnerable as long as the company priorities are pointed in the right direction. **Harry Stebbings** [49:49]: Why do you think Twitter's vulnerable though? I actually don't. I think elections coming up. I think, you know, we are early adopters. The tech ecosystem's early adopters. Majority of people don't actually know about threats, if we're honest. I mean, we're in Europe. It's not gonna be here until early next year. I think Twitter's strong. Elon's brand is bigger and bigger. I wouldn't bet against Twitter. **Brian Balfour** [50:09]: Yeah. So I played a fun game with a couple folks. If you were to have to predict threads DAUs and Twitter's DAUs one year from now, threads is at about a 100,000,000 sign ups, and then Twitter's is, I think, 250,000,000, something like that. If you had to predict, what what numbers would you choose? **Harry Stebbings** [50:25]: Oh, this is so unfact. I just interviewed Adam Messeri, who's running Threads earliest. Oh, you **Brian Balfour** [50:30]: did? **Harry Stebbings** [50:30]: Oh, **Brian Balfour** [50:30]: yeah. That is unfair. **Harry Stebbings** [50:32]: Yeah. Honestly, I don't think Threads is gonna work. I think it will fail pretty quickly. I don't think they'll get retention, and I think the majority of people on Twitter are actually 100,000,000 of the threads people signed up are the 100 hardcore techies, and I think the other 900,000,000 won't move I like the stickiness. I think people are also forgetting two things. One is the friendship graph, and then the other is the interest graph, and they don't overlap. On Instagram, I like Audi. I like Chanel, visually aesthetic, inspiring. And I then I have my brain, which is Balaji. It's Marc Andreessen on Twitter. I don't want a correlation of them both in threads. This is not how a nice wave of social works. So for me, super simple to cut to your point. I would say threads does not work, and I would say maybe it teases along at 5,000,000 DAU with hardcore small numbers. And I would say Twitter has modest growth to two seventy five, 300 with election spikes. Interesting. **Brian Balfour** [51:30]: Yeah. I chose they're flat. I I think Twitter's flat, two fifty million. I actually agree. I lean in your direction, which is that I think people are overestimating, you know, the vulnerability of Twitter. But when when you ask that question, it's like, I just don't think Twitter's going to grow, and that's why I think it's vulnerable. Right? Have they here the scene, if you were to be ahead of growth at Twitter, what would you do? I think this gets back to what we were talking about earlier, which is like, you can have product market fit lives on two vectors, the strength of the fit and the size of the market. I think what you're really seeing, and I think it was like Eugene Way who captured this well in his essay, which is they actually had really strong pre m fit with a smaller market. Like, had these, like, subcultures of Twitter, of, like, tech, and politics, and other things. And what they've done with a bunch of their changes, like the algorithmic feed and all that kind of stuff, is they're essentially searching for that PM fit with a larger market, a larger market of folks. The challenge with that is that you do that and there's a risk of breaking, like, your original PM fit, and that's going to be the big question. If they keep pushing aggressively in that direction. So I think the challenge that they're caught in is the reason they're making all these changes is because of Elon's $44,000,000,000 buy and needing to make that money back. So he has to find PM fit with a larger market in order to make that back, and that's basically forcing essentially maybe unnatural unnatural plays into the game. Do you know what I'd do if I was Elon? What would you do? I would do an Instagram clone. Oh, god. I don't know what kind of tweets back and forth that would create between Zach and Elon. **Harry Stebbings** [53:06]: I would do an Instagram clone, and I would make sure that there were no hacked accounts and no fake accounts. And I would say it's like Instagram with no fake accounts. If threads doesn't **Brian Balfour** [53:17]: work out, that just goes to show so they have the talent. They've got the distribution to leverage. They've got the data, and they've got the capital. If they are not able to break a network, create and break a network effect of Twitter, that just goes to show how strong network effects can be among, like, a certain market. So I think that's the interesting thing that's gonna play out. I kind of agree with you. I think they probably went a little too fast on the demand side of this network, and now they've got to play a game of filling that network with the right creators. And I think the the question is, can they do that fast enough to, like, actually get that flywheel moving in a sustainable direction? So it'll be interesting. **Harry Stebbings** [53:57]: Brian, final one. When you look at the last twelve months, not including threads, when you look at growth strategies, what one have you been most impressed by across the landscape, and why them? **Brian Balfour** [54:09]: I have an unfortunately, like, discolored view of this, because you name a company and I guarantee somebody is a Reforge member, and so what I often see is like things on the outside that look very smart choices, where it was either a little bit luck, or I also see what all the problems are. So I think Canva would be one for me. Canva Talk, and their approach has Canva has a a very, very good job, and probably fair, the marketplace has done a pretty dang good job here in The US. The ones who had to really work for the growth tend to be the most talented teams, right? But there's these situations, like OpenAI, where they kind of captured lightning in a bottle. And I would put that more like, I wouldn't attribute amazing growth strategy or growth talent there. I would attribute that to amazing technology talent and product talent, but I think those two things are different. I think DoorDash is probably another one where they had to freaking work for it, and they took on some people who had advantages that they did not have, like Uber Eats and other ones. And so that's probably another case where they had to work for it. Shopify probably has the best talent density at the moment. Luke Lavask is basically, like, probably one of the most amazing recruiters I know, and he's built a talent density there that easily say is probably the highest talent density team across tech at the moment. **Harry Stebbings** [55:38]: I totally agree. Former guest on the show, so thrilled to hear it. Brian, I've absolutely loved doing this. Thank you so much for joining me today, and I'm glad I set the schedule a couple of hours ago, so you couldn't get too used to it, because I wasn't very aligned to the schedule. But this was a pleasure to do, so thank you so much. Thanks for having me. I mean, that was so much fun to do. I've loved Brian's writing for a long time. And if you wanna see the video version of that interview, you can check it out at YouTube by searching for 20 VC. That's two zero VC. But before we leave you today, ## Sponsor read **Harry Stebbings** [56:06]: listen to this. **Unknown** [56:07]: Mercury has been a breath of fresh air. Getting started was maybe one of the most delightful onboarding experiences I've had. Mercury is just so easy to use. The aesthetic of it is actually quite relaxing. For me, it it was less a choice and and more finding a kindred spirit. **Harry Stebbings** [56:24]: Imagine feeling this way about business banking. You could if you join more than a 100,000 startups on Mercury, the powerful and intuitive way for ambitious companies to bank. Start building momentum and leave the friction behind by visiting mercury.com/20vc. Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group and Evolve Bank and Trust, members of FDIC. And speaking of incredible products like Mercury there, the question, do you know the impact of your marketing efforts or how much you spent so far this month and what you got as a result? Funnel is the world's first and only marketing data hub that takes all the manual effort out of data collection and lets you blend data from different platforms to unlock new deeper insights. And the best part is this can all be achieved without writing a single line of code. That's why it's used by more than 2,000 of the world's top brands and media agencies, including Samsung, Adidas, Sony, and Havas Media. And now you can start for free with a plan that is not time bounded and allows you to see the full power of the platform without any time constraints. Simply head on over to funnel.io/20vc. That's funnel.io/20vc. As always, I so appreciate your support for the show. On Friday, we have the most incredible show, deep diving into the world of AI with Rich Socher at u dot com. It really is a must.