The Exchanges

Q to sales. And I think the question I have is, you know, I spoke to Fabrizio before the show, and he said that you're the master of aligning marketing KPIs to dollars. You mentioned it earlier with Evelyn in terms of that dollar. Can you just help me understand how do you approach marketing KPIs to dollars? And just take me through it. Imagine I'm a founder trying to understand.

A Reject any KPI that is vanity. You don't care about traffic. You don't care about, um, any of those KPIs. The only thing you care about is weight adjusted pipeline dollars. So all of those KPIs, engagement KPIs, your growth team, your marketing team needs to convert them based on lead score, based on company size and stage, your usual close rate, what that. Every single logo that engages with your brand At any stage needs to be converted into a dollar figure. You then sum those, and you have a pipeline by stage that is not just for your sales team, that is also for your marketing team. Your marketing team has a funnel, you know, from like TAM to MQLs, going down to SQLs, then over to your sales team. They should have a pipeline figure per stage, and based on your conversion rate, you should give them goals at every stage of that pipeline. The key learning is what has worked for sales Also works for marketers. If a marketing candidate that you're looking to hire does not own a pipeline figure, that is the wrong candidate, my friend.

AI assessment note: “The only thing you care about is weight adjusted pipeline dollars.”

Q You know, this shit's getting real when I get my pen out. You said three, three, two, two, when it came to growth. For people who don't know what that framework is, can you just explain that for us? Yeah.

A I don't know that it's still I'd say the, the golden rule, and I don't know who came with it, but I, I associate it in my mind with Jason Lemkin, the godfather of SaaS here, because he's the one who's always talking about it. It used to be said that to be in the top decile of startups in terms of performance, thus in terms of valuation and funding and outcomes, you would have to grow past the first 1,000,003 x the first year, three x the second year, and then two x For the next, uh, for every year, for the next three years. So three, three, two, two, two. Okay. And if you do that, you end up at, um, a hundred million plus in revenue in, um, in five years.

AI assessment note: “three x the first year, three x the second year, and then two x”

Q to sales. And I think the question I have is, you know, I spoke to Fabrizio before the show, and he said that you're the master of aligning marketing KPIs to dollars. You mentioned it earlier with Evelyn in terms of that dollar. Can you just help me understand how do you approach marketing KPIs to dollars? And just take me through it. Imagine I'm a founder trying to understand.

A Reject any KPI that is vanity. You don't care about traffic. You don't care about, um, any of those KPIs. The only thing you care about is weight adjusted pipeline dollars. So all of those KPIs, engagement KPIs, your growth team, your marketing team needs to convert them based on lead score, based on company size and stage, your usual close rate, what that. Every single logo that engages with your brand At any stage needs to be converted into a dollar figure. You then sum those, and you have a pipeline by stage that is not just for your sales team, that is also for your marketing team. Your marketing team has a funnel, you know, from like TAM to MQLs, going down to SQLs, then over to your sales team. They should have a pipeline figure per stage, and based on your conversion rate, you should give them goals at every stage of that pipeline. The key learning is what has worked for sales Also works for marketers. If a marketing candidate that you're looking to hire does not own a pipeline figure, that is the wrong candidate, my friend.

AI assessment note: “The only thing you care about is weight adjusted pipeline dollars.”

Q Oh my god, we've got so much more to talk about, but I just want to start with some context. Growth is a weird world. How did you first make your way into growth, and what you think is your first growth role?

A Yeah, uh, it depends how much time you have, but I think there's two, I think there's two founding stories here that are important. One is, ah, I'm an, I'm an old man now. I started in the early 2000 at Apple. And if there's one thing you want to know about Apple, especially, you know, I was based in Europe, is that it's a very centralized company, and so you need to ask authorization for everything, and you never get it in marketing. You just never get it. And so my boss, who was the head of marketing for Apple France, he was not running campaigns because he was not getting authorized. He was running experiments. Because for experiments, he did not need to ask. We're talking about 2005 here, 2004. He didn't need to ask for anyone. And then when he got caught, he said, oh, no, we're just running an experiment to see if the audience of the market reacts well to that. I'm like, that's smart, right? And the second founding story, a couple of years later, I was in an IT security consulting firm, leading marketing over there, and that's when I realized that leveraging engineers to build quasi-products, to build, uh, let's say, demand gen, drive people to the product that is different, so have people build something which is a quasi-product, That works really well, especially for technical audiences, right? Both of those things merged, um, in 2014, 20 15 when I joined Mention, and I …

AI assessment note: “when I joined Mention... at that time, it just started to be called growth”

Q Yeah. No, I, I do agree with you. I think too many kind of use commoditized products that they kind of slap on as calculators. You mentioned segment there. You grew revenue for active segment. I have to ask, what was your biggest lesson from your time at segment?

A Segment is a, it's a technical product targeting a engineering audience. And you know, I often Get founders who have a similar audience, want to sell to engineers, and, and they're telling me, gee, like, you know, outbound emails doesn't work. Engineers don't want to be sold to. And by and large, that's, that's true. That's absolutely true. Though, I want to say, if you're lucky enough to have a really good product, let's put aside those that have like a bad product. We have a truly good product that is useful. You have an incentive to have a free plan, a startup plan, whatever you want to call it, because engineers Well, stay engineers as they change jobs, and they no longer are in startups, and they join a real legit company, and they will bring your product along with them. That's what we found at Segment. Really crazy attribution logic, where we found engineers across jobs by, and that's the unique lesson, keying not on the link, not on the email, because the email changes, domain changes, not on the name, that's not relevant, on the LinkedIn URL. Because unless you're a psychopath, you don't change your LinkedIn URL between jobs. It's the same, like, slug, right? And so we just took all the LinkedIn URLs of other people, and we said, hey, has that person, um, came, uh, three months ago, four, five months ago? And if yes, did they come through a startup plan? And then we di…

AI assessment note: “that's the unique lesson, keying not on the link, not on the email”

Q You know, this shit's getting real when I get my pen out. You said three, three, two, two, when it came to growth. For people who don't know what that framework is, can you just explain that for us? Yeah.

A I don't know that it's still I'd say the, the golden rule, and I don't know who came with it, but I, I associate it in my mind with Jason Lemkin, the godfather of SaaS here, because he's the one who's always talking about it. It used to be said that to be in the top decile of startups in terms of performance, thus in terms of valuation and funding and outcomes, you would have to grow past the first 1,000,003 x the first year, three x the second year, and then two x For the next, uh, for every year, for the next three years. So three, three, two, two, two. Okay. And if you do that, you end up at, um, a hundred million plus in revenue in, um, in five years.

AI assessment note: “grow past the first 1,000,003 x the first year, three x the second year”

Q What are the biggest mistakes you see companies get, make, or do when it comes to activation?

A They take a working growth tactic like that one. And they tone it down a bit. I'll give you an example. You often see emails where they have this, like, almost tone-and-cheek personal from, like, the CS person or, like, a sales engineer, um, making it, look, the text is personal, but it's wrapped in an HTML layer, which makes it very clear it's not personal. And I tell them, like, you gotta pick, folks. Either this is a corporate email, And it's corporate-looking, or this is a text-based email that somebody sent through Gmail. Which behavior are you trying to, like, get your users to, to do, right? What do you want them to believe? Is this coming from a human or is it not coming from a human? And they haven't thought about, they haven't put themselves in the shoes of the recipient. They don't think, like, what feeling am I trying to create? So the biggest mistake is they don't think of the humans on the other side. They don't think of the psychology enough.

AI assessment note: “So the biggest mistake is they don't think of the humans on the other side.”

Q from founders is like, hey, this channel's working. It's working well. How should I think about the other channels? And I'm normally saying, hey, don't get too cute. Facebook ads is working. Just pummel it. Keep going, keep going until it doesn't work. Obviously, you want to transition quickly, but Don't be too cute too early. Do you agree with that? Or do you think actually diversification is important early?

A I'm in between. I'm in a, in business. I don't like to be surprised. And what that means is, um, if I'm going to raise a series A or B, whatever your stage is, uh, I want to know, like, can I actually spend that money? Like, do I have like more than like hypothesis that I know how to spend that money? And so I recommend Increasingly as you go through the stages to just test the cack and the depth of those channels. I'll give you one example. Let's, let's take like paid on, on, on Facebook as, as you name it, right? Do you really know what's the depth of that channel? No. Right? How do you do it? I recommend companies to two X the budget every week until it's that lines. Just do it for like Four weeks, six weeks, it's not gonna, like, burn your budget, your entire, like, capital, right? But just, two x on week two, two x on week three, two x on week four. Very quickly, you're gonna find out where the, where the cap is, right? You're gonna find you're not getting more, like, engagement, more clicks, more revenue, and then you can scale back and can tell your board, hey board, we've tested, we've pressure tested the depth of that channel, and we know how much money we can spend on that channel. And you then go and do the same on every other channel. I did the same for me on emails. It's like, what's the number, maximum number of emails somebody can receive before there's like nega…

AI assessment note: “I'm in between. I'm in a, in business. I don't like to be surprised.”

Q Can you just unpack that for me? Relevance creates reciprocity. Reciprocity gets you responses. Can you just break that down?

A There are, there are two things that work when you try to reach out to someone that you don't know, or a prospect, all right? Whether it's, like, by email, by whatever, whether the communication channel. True value. Your communication. Let's call it an email. Contains something that's really helpful. I can give you an example. And the second is Reciprocity for humanness. They truly believe it comes from a human, and thus they will feel compelled to respond because somebody else has spent some time. Let's unpack those two here, ok? True value. So, um, give an example. Gorgias again, right? A support platform for merchants. Merchants sell, you know, these days a lot, of course, on TikTok, but also on Instagram. Instagram is a major platform for merchants, right? And what we found out is that sometimes Some customers, potential customers of the merchant, of the brand, posted negative comments on a brand's post, and nobody took care of that. We said, hey, that's a miss from the brand, but wait, can we scrape that? Can we automate that? We found out that we could. And as we started scraping and automating the detection of negative comments on brand posts on Instagram, sending it to the brand owner, to the company owner, to the marketing leader, he says, hey, Harry, Love your products. Love what you do. But just letting you know there's like a couple of negative comments here on your…

AI assessment note: “Let's unpack those two here, ok? True value. So, um, give an example.”

Q Yeah. No, I, I do agree with you. I think too many kind of use commoditized products that they kind of slap on as calculators. You mentioned segment there. You grew revenue for active segment. I have to ask, what was your biggest lesson from your time at segment?

A Segment is a, it's a technical product targeting a engineering audience. And you know, I often Get founders who have a similar audience, want to sell to engineers, and, and they're telling me, gee, like, you know, outbound emails doesn't work. Engineers don't want to be sold to. And by and large, that's, that's true. That's absolutely true. Though, I want to say, if you're lucky enough to have a really good product, let's put aside those that have like a bad product. We have a truly good product that is useful. You have an incentive to have a free plan, a startup plan, whatever you want to call it, because engineers Well, stay engineers as they change jobs, and they no longer are in startups, and they join a real legit company, and they will bring your product along with them. That's what we found at Segment. Really crazy attribution logic, where we found engineers across jobs by, and that's the unique lesson, keying not on the link, not on the email, because the email changes, domain changes, not on the name, that's not relevant, on the LinkedIn URL. Because unless you're a psychopath, you don't change your LinkedIn URL between jobs. It's the same, like, slug, right? And so we just took all the LinkedIn URLs of other people, and we said, hey, has that person, um, came, uh, three months ago, four, five months ago? And if yes, did they come through a startup plan? And then we di…

AI assessment note: “That's what we found at Segment. Really crazy attribution logic, where we found engineers”

Q I totally get that, and I love that in terms of an understanding. Four months is very, very rare though. I never see four months. When it's 12, it's like, okay. Like, what is a good payback to you?

A Anything below 12, I'd say, is good. Now, when you think of that, generally, like, that's a very simple framework. If you want to be a bit smarter, you're going to, like, CAP to LTV. Again, going back into, you know, I mean, FinTech these days with Ramp, um, when people onboard a financial, you know, platform, global platform like, like Ramp, they tend to stay, to stay for very, very long, multiple years, and so you can afford It's a something much higher in terms of payback time because the LTV is huge. People stick for three, four, five at the mid market, eight years. So when they stick around for eight years, you can start being very aggressive. You can start spending like more than one year, maybe two years.

AI assessment note: “Anything below 12, I'd say, is good.”

Q A lot of people don't actually get good learnings, I don't think. What's that, and they say, yeah, it didn't work. What's a good learning and what's a bad learning?

A A good learning is a statistically significant deviation from your, uh, status quo. It's hard. I'll give you one example, okay? People often Change their onboarding email, their welcome email, you know, the whole sequence, like post sign up to get people to use the product. And they test and, you know, it's hard to get a lift there. You know what I tell them? Have you tested against a, a holdout, a control audience? And they tell me yes. And they tell me, oh, the control is, is the standard email. I said, no, you're wrong. The control is no email. Have you tried not sending a welcome email? Not sending, you know, the first three, four, five emails to get people to use the products. And of course they haven't. And I tell them, try it. Take 2030, whatever it takes to be statistically significant and don't send emails. And let's see what the true deviation and activation and retention is. And you know what happens, Harry? Most of the time, it's not statistically significant. The people who are going to activate will activate regardless of getting an email or not. Right. And so the email adds no value at all.

AI assessment note: “A good learning is a statistically significant deviation from your, uh, status quo.”

Q different channels and getting to, say, fifty million in ARR, I always think to Kip, At HubSpot, who told me that it takes one channel to get to fifty million an ARR, and two to get to a hundred. From your experience here, you've been at Segment, you've been at Drift. Is that the case? Is there much broader diversification? How do you think about that when you hear it?

A I mean, I've talked to Kip, and, uh, I mean, I'm a big fan of HubSpot and what they've been able to build in the past couple years. No one, almost no one is HubSpot. And they have one of the biggest addressable markets in the world. It's almost any business needs a CRM, ok, or an automation platform. That's one. Second, their market is almost an oligopoly. There's Salesforce, there's HubSpot, maybe there's like Freshdesk, Zoho, that's about it. There's like maybe less than five companies that are like, that are decent when you look at the entire suite, ok, which is extremely rare given the size of the market that is of so few players in that market, right. And you need to have one. It's not that you can say, no, I'm good. Like, I don't need a CRM. I don't need a marketing opportunity. You need to. So you will buy one of those, ok? Which is why it's a great company and it's, it's, they're doing so well, right? Almost no one has the combination of those market attributes. And I have never seen a companies I've worked with, neither at Segment or at Drift, Gorgias, or Ram, and not in one of the companies I engage with, where we, we were able to get to fifty million on one channel. Not one.

AI assessment note: “I have never seen a companies I've worked with... get to fifty million on one channel.”

Q I get you in terms of running a process. So we hire a former founder, probably one that hasn't worked out in the earliest stages. Great. Yes. Is that it? Do we just hire one at a time?

A You want for a growth team to work out. First, it exists after your product team, obviously, otherwise you, you don't have a product. Generally after your first few marketers, right, who are doing the traditional campaigns, the core campaigns, and you're going to create this startup in the startup, this team within the team, um, and they're going to take risks as we talk as like a VC, okay? What do they need to be able to do that? They need to be able to build and to ship. That means you have this former founder who's like the Head of growth slash PM running the process. You're going to have at least one engineer, maybe data analyst, if the founder can do it, but you, you, the, the head of growth is the analyst in the early days. So you have like head of growth, engineer, and one marketer slash copywriter. That's, that's your team.

AI assessment note: “So you have like head of growth, engineer, and one marketer slash copywriter.”

Q Does AI make this much, much harder? Because cold outbound, done well like you've done there, People think that it's actually a human and it's not. Kind of the wrapper of AI makes people just inherently more cynical that everyone's a robot. Does it make it harder?

A It will. Or I should say, I have a conviction. I have a, my personal conviction here is two things will happen that will be true at the same time. Um, some part of the population will start rejecting communication coming from anonymous people. Because they won't know if it's a human or not, and thus they will, by default, think that it's a robot, and thus that there is no reciprocity that should be expensed. And some other part of the population, maybe the more textile people, will not care and say, hey, is this information valuable? Right? The problem is that because the marginal cost is so low, high-quality text, high-quality communication about you will become, like, It's going to be all the time everywhere, and you just won't have the time for it. So it's very likely that we'll start ignoring messages from real humans at a much higher rate, because the quality of AI will be better than the quality of humans.

AI assessment note: “It will. Or I should say, I have a conviction.”

Q love that in terms of kind of an alternate tool that drives DemandGen to the core product. Well, I, you know, I automatically think of actually HubSpot and the website grader, which did it so well. Do you think that's, again, fuck it, I had the schedule, sod it. Do you think that still works today, or is it too commonly trodden path now where everyone's tried to do it?

A It's hard to, uh, come out with a really good product, um, that you're not monetizing directly. Uh, first, because many people If there is interest for a product, people have built a better version of it and have monetized it. So, like, a lot of those niche have disappeared, right? And the second is that the quality has gone down, and people have become desensitized when they say, oh, there's, like, a free, uh, you know, cost calculator, return on investment calculator, like, one of those typical things, right? Or website grader. Like, most of those are terrible now, and so people will think, like, oh, I'm not gonna, like, sign up or put my info, my PII there because I'm not gonna get my value's worth of it, right? Those negative effects there. But if you have something truly amazing, does it still work? Absolutely. Absolutely.

AI assessment note: “if you have something truly amazing, does it still work? Absolutely.”

Q I totally get that, and I love that in terms of an understanding. Four months is very, very rare though. I never see four months. When it's 12, it's like, okay. Like, what is a good payback to you?

A Anything below 12, I'd say, is good. Now, when you think of that, generally, like, that's a very simple framework. If you want to be a bit smarter, you're going to, like, CAP to LTV. Again, going back into, you know, I mean, FinTech these days with Ramp, um, when people onboard a financial, you know, platform, global platform like, like Ramp, they tend to stay, to stay for very, very long, multiple years, and so you can afford It's a something much higher in terms of payback time because the LTV is huge. People stick for three, four, five at the mid market, eight years. So when they stick around for eight years, you can start being very aggressive. You can start spending like more than one year, maybe two years.

AI assessment note: “Anything below 12, I'd say, is good.”

Q have horizontal SaaS products, it could be Airtable, it could be Dropbox, it could be any of these, um, they can be used by dentists, they can be used by product managers, and the use cases are so different. So to get messaging that resonates across that spectrum is so tough. How do you think about that horizontal product marketing being done well and resonating and getting in their psychology?

A You know what's fantastic? We live in twenty-twenty-three, and now you have AI products, you know, chat, GPT, you name it, all the ones that you love, and you can build personalized content with a scale that you never dreamed of before, which is at a quality that was unachievable before. What people care, especially in horizontal products, is that you talk about them, their problems, their challenges. I might go back when I said just earlier, you asked me what was the biggest mistake. Actually, it's probably that companies, marketers, talk of them and their product first. They don't talk About you, the audience, and your problems. They say, hey, here's my product. Here's what we do. Here's our customers. Here's why we are great. And you know what? As a recipient, I don't care. I really don't care. I've got other things to care about in my day than like a random company I never heard about, right? But if you, if you reach out to Harry and you have context in his business, what he does, what he loves and care, what challenges, you have a really good assumption that he's facing right now. Like, it's much more likely that you can pique his interest. These days, with large language models, you can inject the data about, hey, is this a dentist? What's the size of the dentistry? Where are they based? Are they in the mission, in the SF? You know, what problems might they have, right? I…

AI assessment note: “with large language models, you can inject the data about, hey, is this a dentist?”

Q love that in terms of kind of an alternate tool that drives DemandGen to the core product. Well, I, you know, I automatically think of actually HubSpot and the website grader, which did it so well. Do you think that's, again, fuck it, I had the schedule, sod it. Do you think that still works today, or is it too commonly trodden path now where everyone's tried to do it?

A It's hard to, uh, come out with a really good product, um, that you're not monetizing directly. Uh, first, because many people If there is interest for a product, people have built a better version of it and have monetized it. So, like, a lot of those niche have disappeared, right? And the second is that the quality has gone down, and people have become desensitized when they say, oh, there's, like, a free, uh, you know, cost calculator, return on investment calculator, like, one of those typical things, right? Or website grader. Like, most of those are terrible now, and so people will think, like, oh, I'm not gonna, like, sign up or put my info, my PII there because I'm not gonna get my value's worth of it, right? Those negative effects there. But if you have something truly amazing, does it still work? Absolutely. Absolutely.

AI assessment note: “if you have something truly amazing, does it still work? Absolutely.”

Q from founders is like, hey, this channel's working. It's working well. How should I think about the other channels? And I'm normally saying, hey, don't get too cute. Facebook ads is working. Just pummel it. Keep going, keep going until it doesn't work. Obviously, you want to transition quickly, but Don't be too cute too early. Do you agree with that? Or do you think actually diversification is important early?

A I'm in between. I'm in a, in business. I don't like to be surprised. And what that means is, um, if I'm going to raise a series A or B, whatever your stage is, uh, I want to know, like, can I actually spend that money? Like, do I have like more than like hypothesis that I know how to spend that money? And so I recommend Increasingly as you go through the stages to just test the cack and the depth of those channels. I'll give you one example. Let's, let's take like paid on, on, on Facebook as, as you name it, right? Do you really know what's the depth of that channel? No. Right? How do you do it? I recommend companies to two X the budget every week until it's that lines. Just do it for like Four weeks, six weeks, it's not gonna, like, burn your budget, your entire, like, capital, right? But just, two x on week two, two x on week three, two x on week four. Very quickly, you're gonna find out where the, where the cap is, right? You're gonna find you're not getting more, like, engagement, more clicks, more revenue, and then you can scale back and can tell your board, hey board, we've tested, we've pressure tested the depth of that channel, and we know how much money we can spend on that channel. And you then go and do the same on every other channel. I did the same for me on emails. It's like, what's the number, maximum number of emails somebody can receive before there's like nega…

AI assessment note: “I'm in between. I'm in a, in business. I don't like to be surprised.”

Q Final one, and then we'll do a quick fire. Now you're an advisor in many amazing companies. When should companies hire an advisor? When should they hire full-time?

A You should always hire full-time when you can. Like, I always tell people, like, if you can hire the person you, you want, that's the right fit for you. Go for it. I'm not a good replacement for that. The hole that I plug, the problem that I fix is generally you can't make that higher, right? Because the experienced growth person or that founder that you care about that is amazing, do they want to take the risk with you at the Series A? No. They want to go build their own company, right? And so who are you going to hire? You're going to hire a more younger Let's say, with, let's say, high on the execution scale, right? But they don't have the expertise. They don't have the experience. I'm the fix for that, right? But if you can hire the guy that has experience, the skills, the knowledge, and can like ship, hire that person absolutely 10 times out of 10.

AI assessment note: “You should always hire full-time when you can.”

Q know, what worries me most though, actually, Is the fact that a lot of startups that I work with today, don't insult me or my portfolio, please. Uh, but they essentially layer on the enterprise cost base of sales, marketing, customer success onto a kind of PLG style price. And I'm going, shit, that doesn't work. Am I right or am I wrong? Cause they argue, well, it could scale.

A You're right, but there's worse. That you're absolutely right. Like, you, if you, if you don't create the packaging, the pricing plans to support the high cost of the enterprise motion, then you're killing yourself. But there's worse. It's companies that are enterprise first with the entire, like, cost structure and product structure of an enterprise software, and then want to open a PLG and go, hey, let's let people, like, sign up on their own. And like nothing works. Like the acquisition does not work. The product that does is not self-serve. You can't like build like a self-serve like feature into like something that was meant for your CSMs to build the instance like that. And that happens. I see those people come to me and it's like, this is just not going to work. It's impossible. You can't pivot that way.

AI assessment note: “You're right, but there's worse. That you're absolutely right.”

Q What are the biggest mistakes you see companies get, make, or do when it comes to activation?

A They take a working growth tactic like that one. And they tone it down a bit. I'll give you an example. You often see emails where they have this, like, almost tone-and-cheek personal from, like, the CS person or, like, a sales engineer, um, making it, look, the text is personal, but it's wrapped in an HTML layer, which makes it very clear it's not personal. And I tell them, like, you gotta pick, folks. Either this is a corporate email, And it's corporate-looking, or this is a text-based email that somebody sent through Gmail. Which behavior are you trying to, like, get your users to, to do, right? What do you want them to believe? Is this coming from a human or is it not coming from a human? And they haven't thought about, they haven't put themselves in the shoes of the recipient. They don't think, like, what feeling am I trying to create? So the biggest mistake is they don't think of the humans on the other side. They don't think of the psychology enough.

AI assessment note: “biggest mistake is they don't think of the humans on the other side.”

Q have horizontal SaaS products, it could be Airtable, it could be Dropbox, it could be any of these, um, they can be used by dentists, they can be used by product managers, and the use cases are so different. So to get messaging that resonates across that spectrum is so tough. How do you think about that horizontal product marketing being done well and resonating and getting in their psychology?

A You know what's fantastic? We live in twenty-twenty-three, and now you have AI products, you know, chat, GPT, you name it, all the ones that you love, and you can build personalized content with a scale that you never dreamed of before, which is at a quality that was unachievable before. What people care, especially in horizontal products, is that you talk about them, their problems, their challenges. I might go back when I said just earlier, you asked me what was the biggest mistake. Actually, it's probably that companies, marketers, talk of them and their product first. They don't talk About you, the audience, and your problems. They say, hey, here's my product. Here's what we do. Here's our customers. Here's why we are great. And you know what? As a recipient, I don't care. I really don't care. I've got other things to care about in my day than like a random company I never heard about, right? But if you, if you reach out to Harry and you have context in his business, what he does, what he loves and care, what challenges, you have a really good assumption that he's facing right now. Like, it's much more likely that you can pique his interest. These days, with large language models, you can inject the data about, hey, is this a dentist? What's the size of the dentistry? Where are they based? Are they in the mission, in the SF? You know, what problems might they have, right? I…

AI assessment note: “you can build personalized content with a scale that you never dreamed of before”

Q Does AI make this much, much harder? Because cold outbound, done well like you've done there, People think that it's actually a human and it's not. Kind of the wrapper of AI makes people just inherently more cynical that everyone's a robot. Does it make it harder?

A It will. Or I should say, I have a conviction. I have a, my personal conviction here is two things will happen that will be true at the same time. Um, some part of the population will start rejecting communication coming from anonymous people. Because they won't know if it's a human or not, and thus they will, by default, think that it's a robot, and thus that there is no reciprocity that should be expensed. And some other part of the population, maybe the more textile people, will not care and say, hey, is this information valuable? Right? The problem is that because the marginal cost is so low, high-quality text, high-quality communication about you will become, like, It's going to be all the time everywhere, and you just won't have the time for it. So it's very likely that we'll start ignoring messages from real humans at a much higher rate, because the quality of AI will be better than the quality of humans.

AI assessment note: “It will. Or I should say, I have a conviction.”

Q Can you just unpack that for me? Relevance creates reciprocity. Reciprocity gets you responses. Can you just break that down?

A There are, there are two things that work when you try to reach out to someone that you don't know, or a prospect, all right? Whether it's, like, by email, by whatever, whether the communication channel. True value. Your communication. Let's call it an email. Contains something that's really helpful. I can give you an example. And the second is Reciprocity for humanness. They truly believe it comes from a human, and thus they will feel compelled to respond because somebody else has spent some time. Let's unpack those two here, ok? True value. So, um, give an example. Gorgias again, right? A support platform for merchants. Merchants sell, you know, these days a lot, of course, on TikTok, but also on Instagram. Instagram is a major platform for merchants, right? And what we found out is that sometimes Some customers, potential customers of the merchant, of the brand, posted negative comments on a brand's post, and nobody took care of that. We said, hey, that's a miss from the brand, but wait, can we scrape that? Can we automate that? We found out that we could. And as we started scraping and automating the detection of negative comments on brand posts on Instagram, sending it to the brand owner, to the company owner, to the marketing leader, he says, hey, Harry, Love your products. Love what you do. But just letting you know there's like a couple of negative comments here on your…

AI assessment note: “Let's unpack those two here, ok? True value.”

Q A lot of people don't actually get good learnings, I don't think. What's that, and they say, yeah, it didn't work. What's a good learning and what's a bad learning?

A A good learning is a statistically significant deviation from your, uh, status quo. It's hard. I'll give you one example, okay? People often Change their onboarding email, their welcome email, you know, the whole sequence, like post sign up to get people to use the product. And they test and, you know, it's hard to get a lift there. You know what I tell them? Have you tested against a, a holdout, a control audience? And they tell me yes. And they tell me, oh, the control is, is the standard email. I said, no, you're wrong. The control is no email. Have you tried not sending a welcome email? Not sending, you know, the first three, four, five emails to get people to use the products. And of course they haven't. And I tell them, try it. Take 2030, whatever it takes to be statistically significant and don't send emails. And let's see what the true deviation and activation and retention is. And you know what happens, Harry? Most of the time, it's not statistically significant. The people who are going to activate will activate regardless of getting an email or not. Right. And so the email adds no value at all.

AI assessment note: “A good learning is a statistically significant deviation from your, uh, status quo.”

Q different channels and getting to, say, fifty million in ARR, I always think to Kip, At HubSpot, who told me that it takes one channel to get to fifty million an ARR, and two to get to a hundred. From your experience here, you've been at Segment, you've been at Drift. Is that the case? Is there much broader diversification? How do you think about that when you hear it?

A I mean, I've talked to Kip, and, uh, I mean, I'm a big fan of HubSpot and what they've been able to build in the past couple years. No one, almost no one is HubSpot. And they have one of the biggest addressable markets in the world. It's almost any business needs a CRM, ok, or an automation platform. That's one. Second, their market is almost an oligopoly. There's Salesforce, there's HubSpot, maybe there's like Freshdesk, Zoho, that's about it. There's like maybe less than five companies that are like, that are decent when you look at the entire suite, ok, which is extremely rare given the size of the market that is of so few players in that market, right. And you need to have one. It's not that you can say, no, I'm good. Like, I don't need a CRM. I don't need a marketing opportunity. You need to. So you will buy one of those, ok? Which is why it's a great company and it's, it's, they're doing so well, right? Almost no one has the combination of those market attributes. And I have never seen a companies I've worked with, neither at Segment or at Drift, Gorgias, or Ram, and not in one of the companies I engage with, where we, we were able to get to fifty million on one channel. Not one.

AI assessment note: “we were able to get to fifty million on one channel. Not one.”

Q isn't, I, as a Brit, I love simple frameworks. It helps bring a lot of clarity to the thought process. When we channeled before thinking of kind of that growth mechanism, you said boards want, I'm just skipping to CAC. I love CAC. You said that boards want low CAC and high scale growth. I just want to kind of Break that down. How do you define low CAC, G?

A Yeah, it's a good question. So, at first, like, why do boards want that now? Well, it's because we're in this, like, kind of recession or tough times, I want to say, in a way where capital is scarce. And when capital becomes scarce, there's two ways to win. You either grow fast out of it, well, that's tough, or you reduce cost, right? And founders try to do a bit of both, right? To de-risk. And so, when you think of all of the people asking about PLG, for PLG, that's another way of saying they're looking for low-cac tactics. They're looking for customers that bring themselves, that bring other customers for free that you don't have to pay for. Because that sounds just great. If you go beyond that, if you can't do the PRG, which most can't, as we discussed, Well, I'd say low CAC, if you look at the, um, you know, the KeyBank Capital deck that they release every year, look at the top quartile of startups in terms of, um, performance, um, post five million revenue, you're gonna see that the, roughly, it's about a dollar of revenue, sorry, a dollar of cost for a dollar of revenue. So you spend a dollar to get your dollar, right? That's the, that's, that's the top quartile, and that's fully loaded CAC. What that means is that takes into account, like, the marketing spend, the marketing headcount, the sales team, bonuses, and whatnot, everything but building the product, ok? Marketin…

AI assessment note: “roughly, it's about a dollar of cost for a dollar of revenue”

Q know, what worries me most though, actually, Is the fact that a lot of startups that I work with today, don't insult me or my portfolio, please. Uh, but they essentially layer on the enterprise cost base of sales, marketing, customer success onto a kind of PLG style price. And I'm going, shit, that doesn't work. Am I right or am I wrong? Cause they argue, well, it could scale.

A You're right, but there's worse. That you're absolutely right. Like, you, if you, if you don't create the packaging, the pricing plans to support the high cost of the enterprise motion, then you're killing yourself. But there's worse. It's companies that are enterprise first with the entire, like, cost structure and product structure of an enterprise software, and then want to open a PLG and go, hey, let's let people, like, sign up on their own. And like nothing works. Like the acquisition does not work. The product that does is not self-serve. You can't like build like a self-serve like feature into like something that was meant for your CSMs to build the instance like that. And that happens. I see those people come to me and it's like, this is just not going to work. It's impossible. You can't pivot that way.

AI assessment note: “You're right, but there's worse. That you're absolutely right.”