MIKE DUBOE

In today’s episode we had Sam Ross, who is CEO of Numeral, and an old growth friend. He previously ran growth at Numeral and Airbnb. This was a very interesting conversation where we leaned more tactical on performance marketing, and specifically there was one interesting thread on how to apply DTC consumer marketing tactics to building a B2B business that I think got us to some really interesting territory.

RISHABH JAIN

Yeah, I think one of my big takeaways that Sam really spends a lot of time on is thinking much more tactically and avoiding the generalizations of what can work at any point in time for every channel. And he gives this analogy of how a lot of the product-driven founders are afraid of investing into growth. And I think that there is an important lesson there that I encourage everyone to listen to how he thinks about investing in growth as a founder of a product — of a SaaS product.

MIKE DUBOE

Hope you enjoy it. Feedback welcome. Sam, thanks for joining. You’ve been a friend of mine for over a decade now and I think you’ve known Rishabh as well. You were initially doing growth at Teespring where we kind of were running competing growth programs back when I was at Tilt; Airbnb, which ended up acquiring that company; and then you ran your own studio of brands where we used to trade notes when I was an eCom marketer. And so it’s been very fun knowing you on that journey. And now running Numeral, of course, which is sales tax software for eCom merchants and software startups, you continue to be a very talented marketer running that business. So I thought this will be an interesting conversation today, the common thread being your role as a marketer around this journey. And of course we’ll get into Numeral as well, but you’ve really sat on every side of the table in the eCommerce and marketing world and so I think we’ll have a lot to cover with you. So, welcome.

SAM ROSS

Excited. Thank you guys for having me.

MIKE DUBOE

Maybe to kick it off, I want to talk about your journey from being a performance marketer to really becoming a full-stack operator. To rewind, at Airbnb, you were reasonably early doing growth there, you added about a billion of GMV by really — and I say this in a positive way — exploiting early-days Facebook ad arbitrage. That was obviously the case at Teespring too. I think you all were really OGs to some of the precise last-name targeting tactics that kind of propelled your growth early on, and we tried to do some of that at Tilt too. Maybe looking back, what were some of the benefits of business building via these ad tactics and what were some of the negatives, and maybe expanding from that, what’s the best way to take advantage of paid media today?

SAM ROSS

Yeah, good question. So I think to set the context, when I joined Teespring back in the day, which was this very hype company with Sam Altman on the board and coming in every week telling us we were going to be … us and Stripe were the next $10 billion companies, and Keith Rabois on the board and Andreessen. So it was like this hype company. Really what happened there, it was a platform for selling T-shirts. so a little bit akin to Shopify. And as soon as the News Feed ad came out on Facebook, the business just exploded because, at that time, I dunno if you remember, there was all these articles about how Facebook ads don’t work. Proctor and Gamble was cutting their spend. Everyone’s like, “This thing is a joke,” like Meta ads or Facebook ads are a joke, but with Teespring you could just reverse-engineer all of its ad targeting.

So somebody, they’re a nurse, you could just say, “I’m a nurse,” or as you said, we could do even crazier tactics where we could scrape it and I could have an ad, “Hey, I’m a Mike!” and target all the Mikes. And so I think where we got a lot of learning was, you could target all these sub kind of micro TAMs — target nurses, chihuahua owners, people named Mike — and it was a lot of demand testing. And so you really got a feel for what it’s like to have fine product-market fit in a niche. And so I think you really got us … it is this pure capitalistic thing where you were really feeling the demand curve. It was easy back then too. So you’re getting conversions for … We were upset if our cost-per-conversion was over $10 and you’re getting two-cent, three-cent clicks.

We just got a lot of exposure and learnings, and I think as a performance marketer, that’s one thing you get is you just get so much … there’s just so much feedback you’re constantly getting on what’s working, what’s not working. Whereas you have people who’ve been doing brand marketing for years and running these award-winning campaigns for big companies and they honestly, there’s no feedback, there’s not really much of a feedback loop. They find out a couple quarters later and it’s hard to correlate these things. And so I think you just were able to really get a lot of a feedback cycle. And similarly, Airbnb as well, although it was always a little bit harder to measure when you’re doing these things at a much bigger scale where there’s a lot of other demand channels and a lot more word of mouth.

MIKE DUBOE

So maybe let’s stay in the Teespring example for a minute, because I think it was working so well, at a certain point it starts to decay. What are the signs you’re looking for or when did you know that it was not going to continue to work as well as it had been? And what do you do in that moment?

SAM ROSS

The business broadly or …?

MIKE DUBOE

I’m talking about these tactics that you tapped into, namely exploiting Meta ads — er, Facebook ads at the time — so early.

SAM ROSS

Yeah, good question. When I was there, what ultimately hurt the business was Zuckerberg, we found out later, didn’t like … There was a College Humor article, “Top 10 creepiest Facebook ads,” and it was T-shirt ads with their name on it was number one. So literally there was a day where all — the whole platform — all the sellers accounts got banned. And we found out later, Zuckerberg had his team make hard code, different blockers for T-shirt ads to downrank them and stuff. And so I think we saw it get hard early, and I think it took me a long time to realize [or] understand that just having quick growth is not the end-all to things. Because what happens is, yeah, you can get a ton of quick growth, but the faster you rise, the faster you fall if you don’t have retention. I mean, I remember after I left Airbnb, I decided to do the digital nomad thing, live abroad, so I just needed a quick asset business without investors. And I started selling this cheap jewelry online. I remember coming to your office at Greylock, you’re like, “What are you doing? You’re selling 50-cent jewelry from China for $50 — What is this? — to moms in the Midwest?” And I was like, yeah, I don’t know what the long-term thing is, but I didn’t really put a lot of thought into it. I figured out how to cashflow it very quickly. And then I also remember our friends, hanging out with Gustav and Casey Winters — they’re these … They’re the big growth influencers, particularly at the time. And their whole thing was like, oh no, retention is so important and you got to think about retention for your business. And you hear these aphorisms like that and you’re like, yeah, yeah, that makes sense. I get that retention … But until you actually own a business and feel how painful churn is … So yeah, I sold tens of millions of dollars of jewelry, but then when all your customers churn and never come back and you have to keep reacquiring them, then I think, yeah, you realize that yeah, maybe there’s something … just being really good at Facebook ads or really good at some whatever ad tactic strategies, it’s not sustainable unless your business is retained.

MIKE DUBOE

So let’s stay on this for another minute because I think there are some important lessons here. These categories we’re building in are not naturally retentive. Buying jewelry, T-shirts, you could still build a good business if you have that understanding upfront. I think the issue is when you start paying as if these users are going to be retained. But I guess knowing what you know now, would you have started in different categories? Would you have managed to different ROAS expectations? What would you have done differently with this understanding of retention?

SAM ROSS

Yeah, I think it’s a good question. One example would be, we knew that at Teespring, nurses were maybe the best category because there was this weird demographic that they had a lot of disposable income, but they tend to just love to buy nurse gear. If you know that, you could either go create … instead of selling a T-shirt — “I love being a nurse” — you could actually go and sell the scrubs and do what FIGS did and actually create a public company selling gear for nurses. Or my friend Lenny, who I knew at the time at Trusted Health, what he did was he’d worked at Hired, which was helping recruiting engineers. And we saw that … I think I talked with him about, I’m like, “Oh man, nurses is this great category that loves to be on Facebook and click Facebook ads.” So he basically built what Hired had done for nurses. Again, you end up building these businesses that, okay, when I acquire a nurse for a T-shirt, I make $20 off of them, but if I acquire them to get a job or I acquire them to be buying scrubs for the next 10 years or build a foundational brand, the amount of value you get is so much more. And actually, on the performance side, the interesting thing is you also can just spend way more, right? So when we’re selling T-shirts that I know it’s a one-time purchase, the problem is your CAC has to be $5 or $10. And so when the market moves, these things become more efficient, the cost acquisition gets more expensive over time, things like this, eventually I couldn’t sell my jewelry at a certain point because just the cost to acquire customers became … the economics just didn’t pencil out anymore. But if you’re trying to acquire users for it’s going to be 10, 20 years or it’s this longer journey and your LTV actually is a lifetime, it’s a much different story. Or you sell more expensive goods too. You have the people like the Eight Sleeps of the world, etc., where I buy one of these $5,000 mattresses, maybe it won’t be a 20-year customer, but $5,000 is a lot more than a $20 T-shirt or whatever. So I think you don’t really see a lot of performance marketers selling $20, $30 items, whereas 10 years ago there was a lot more of that.

RISHABH JAIN

Yeah, totally. And just to sort of pivot off of this question of, hey, those categories themselves have particular properties that are not retentive in nature, but let’s use your business today. I messaged you over X about a bunch of your LinkedIn ads that I’ve been seeing. And so there’s still some of that DNA around paid media excellence coming into your business DNA today. How do you think about … What are some of the learnings you had over time? How does that play into how you think about growth today? Whether it’s software, what you’re doing with Numeral, or product businesses, like the customers that we serve?

SAM ROSS

Yeah, good question. It’s been a learning… I think my career is mostly on consumer marketing and selling items where … Again, I was very performance-marketing focused and so I could always measure everything pretty well. It’s a lot of tactics. I think for us, what we learned when we started doing some paid ads at Numeral, was a couple of things. I think one is … I think a lot of founders are, they hear all these horror stories of spending money on Google ads and how Google’s abusing them or things like that. And so they’re scared to spend money. And I was just speaking at YC the other day, and a lot of these companies, they get their first 500K and they think they have to covet it and be so careful with their money, but I’m like, what’s the point of sitting on this money if you’re not going to grow your business? The point is, the reason you’re on this venture capital rat race or treadmill is you need to deploy it. That’s the whole point. So I think I was always a little bit more aggressive about deploying the funds. And you get to measure, obviously, how much you’re getting back. And then I think the other thing is there’s always just tactics, whatever tactics are working. We’ve done stuff like whitelisting and things like this. So I think you always just … There’s a lot of learnings you get from doing DTC, so it’s things like whitelisting, it’s things like making your ads obvious, making the ads look native, not just doing some boring creative. You’ll see ads on LinkedIn that are, “Download your KPI checklist,” and maybe it looks professional, but it’s the most boring ad and you scroll right past it. Any direct-to-consumer marketer, if that’s what you’re running for, if you’re running some very boring generic ad for your DTC brand, your brand’s never going to succeed. You’re going to die. You need to have thumb-stopping ads, as Facebook used to call them. There’s always a balance. I think there’s also this thing that’s like, “ugly ads work,” and so sometimes you have … And yeah, on a performance metric, they might. The difference you have to think about obviously with building a brand and especially building something more where you have a sophisticated buyer, I’m selling to controllers at big fancy companies, they’re not necessarily … Just because this tactic gets their attention doesn’t mean it’s going to convince them to go and make this big purchase and things like that. So I think I’ve had to adapt somewhat. And it’s always the learning process too, because I think the hardest part is just, it’s way harder to measure things and it’s a lot more just trying to look at crappy data and try to make best-guess decisions.

MIKE DUBOE

This is a very interesting point that — and it’s obvious from a distance, but hearing you say it crystallizes that — the foundation of your marketing at Numeral is really rooted in your DTC DNA. That’s evident when you look at what you’re doing on LinkedIn, like whitelisting other influencer content. Do you feel like that works across … You’re in eCom SaaS, which arguably has more consumer-like properties than elsewhere. Do you think your approach works for other B2B categories? What advice might you have to other B2B founders that are inspired by what you’re saying in the consumer growth side?

SAM ROSS

Yeah, I would say — it’s hard to generalize any of this advice, but what I would say is you have a lot of these founders who, they come in, they’re very … Gustav likes a lot of times to speak to YC companies and they’re like product visionaries. And then when it comes to go-to-market, it’s just like they delegate it away to someone. In reality, there’s this whole founder mode thing. You want to be … most good founders, is any detail of your … You’re obsessed about every detail of your product. And why is it — I’m sure Rishabh has this thing, you get mad — why was it designed like this? And you go and you write a whole spec and you redesign it. And I think you kind of have to have the same approach to a lot of your go-to-market. So even our first channel that we worked on was, for go-to-market, was cold email. It was something I had never done before, but at the same time, we just took a really intense approach to it. I did talk to a ton of people and we made everything super tailored, the way we collected the data, all of this. And so I think just at a high level, the thing with growth is it’s a lot … It’s very tactic heavy. People try to synthesize all these principles and spend this time — Yeah, sure, there’s a handful of principles, but generally, sadly, it’s a lot of tactics that last for six months and then you have to move on to the next. And so I think the challenge is, yeah, it’s easy to want to delegate that away, or find … And then I also think a lot of the time the talent you’re putting onto this is not top-tier talent. So your product manager might be some intense Stanford kid who’s high achieving and did all this and that and worked. And then for your go-to-market, you just find someone who couldn’t find another job and fell their way into B2B. Because B2B growth is, particularly on the B2B side, people aren’t like, oh, in college, what do you want to do? I want to be a B2B marketer, that’s nobody’s dream. And so there’s often a low … It’s not elite talent wanting to do this stuff. And so you have to really try hard to find good people. And my strategy has always been finding people who are incredibly talented and training them and putting them into seat. So we have two founders running all of our growth efforts versus someone who’s a career B2B growth person.

RISHABH JAIN

Can I ask?

SAM ROSS

Yeah.

RISHABH JAIN

There’s two questions that I want to explore separately, and this is awesome. The first one is, given — and I believe … I agree, with you by the way, on it being very tactic heavy and you have to be very thoughtful about the details at excruciating levels. You might find this interesting so I’ll share this story. Recently, for every inbound lead that we get, I have asked our team to send a response to the warm inbound within X time. And then, two, send me a screenshot of the email that they send and of the email that they get back and it must be screenshotted and posted in a public channel. And I’m like, as you were saying this, I was like, ah man, this is totally true. And people don’t understand how much everything is in the nuance. Given that, I am curious how you think about these various AI tools — the AI SDRs, the AI marketing email automation — and if you’ve tested them and why or why not for the growth side, because specifically on the growth side is where we’ve seen an explosion of these AI tools. That’s one question. And then another one that I want to come back to is on the hiring of a growth person. But yeah, let’s start on the AI tooling. I’m very curious to get your take on it.

SAM ROSS

Yeah, interesting. So I think, I dunno, we built all our cold-out … So it started like AI SDR, so I built all my cold-out email infrastructure before there was a bunch of hype and then all these tools came out in the market. You had the things like ElevenX and all these things come out. And so I remember I jumped on the phone with now-disgraced founder of ElevenX, and I was like, well, “Hey, here’s my stack” — because I had some issues that weren’t working — “should we use you?” And he’s like, “Oh no, your stack’s the same as ours.” We’re using all the same tooling, we’re just basically, he’s like, we’re basically just inserting AI at small points, but it’s basically an agency-type model, but we kind of streamlined things and we’re calling it SaaS. And so I think some of these AI SDR tools, maybe they’ve evolved since I’ve last looked at them, but my sense is they’re doing some AI stuff at the margins maybe to write copy. And sure, we use … We will plug in different AI things maybe to tailor copy to an extent. But a lot of it, it’s not like it’s fully owned AI, you’re just giving the agent autonomy and it’s going and figuring things out. And so I don’t know, I haven’t personally encountered these AI — maybe you’ve seen better stuff — where it’s just like, wow, this is really owning the job end-to-end completely. It’s a lot of at the margin, it’s making things slightly more tailored and I think that’s helpful and you always want to be using … Again, these tactics are always evolving. You want to stay up to date with what’s working because that’s how you get your alpha. And so there’s probably stuff I should be looking into more. (laugh)

RISHABH JAIN

Yeah, I have unfortunately not found — I mean, I would love to find something, I want to be very clear, and I want these companies to work and be successful — unfortunately, I have not found anything, and if I do, I’ll let you know. Yeah. (laughs)

SAM ROSS

Yeah. It’s like there’s all these AI chatbots on your website and all these things and … I don’t know, I still think until we get … until AI can actually … I don’t know if you … Dwarkesh had a recent blog post about this where it’s like, the problem with all these AI things is your workers, they get better over time. On your first week at work, they’re not going to be good, but six months into the job there’s kind of this continual learning, they’re continually building context. The problem is, every time the AI starts a conversation with something, it’s kind of starting again for the fresh context. You can try to context stuff it as much as you can, but it’s not continually getting better. So maybe the AI SDR-type tooling, it’s better than my SDR for the first week or two. But once my SDR has been in share for six months, it’s just going to be way better than any of the current AI tools. So I think eventually we’ll probably get there, but there’s this paradigm shift that has to happen.

MIKE DUBOE

So you made a point earlier — and Rishabh was going to ask a question on team composition and the changing nature of what you’re looking for in hires. You made a point earlier that I strongly agree with, which is being overly planful with growth could really be detrimental because ultimately it’s about how quickly you could experiment, iterate, learn, move to the next tactic. And I think there’s, yes, you want some guardrails or some process that actually ensures that your learnings are compounding, but actually creating a three- to six-month roadmap or something, you’re going to miss the window. I very much agree with you on that and that’s maybe one of the differences in running growth versus traditional marketing. But as you’re thinking about the changing nature of growth, have you thought about hires differently? Like, I’m with you. The best people I’ve hired in my previous seats were like pirates. They were just not necessarily pedigreed. They tended to be highly analytical but really just ran at stuff and didn’t really ask for permission. So that’s an evergreen skill set, I guess. But what’s changing now? How do your best growth hires today look different than they did five, 10 years ago?

SAM ROSS

Yeah, good question. I think … man, in the past when you look at how teams were built, it was kind of like trying to accumulate all these specialists. So you hire your SEO guy or gal, you hire the PPC person, you kind of accumulate. And maybe at bigger scale there’s some level that you have to do. But I think, first of all, I think teams are way more lean. You used to have teams, like the Machine Zone was famous [for its] performance marketing team. You would just recruit these banker kids and had 50 of them managing these … in Excel all day. And it’s like, I think you can run … You could probably run a billion-dollar budget with a four-person team or something today. Talking about AI, where AI has been really effective is how Meta and Google have deployed it for their ad systems. And that’s probably the biggest impact in the whole economy is literally there for AI — maybe it’s not LLMs but it’s some form of AI. I think people … It’s a very high-leverage … Anyone doing growth is highly, highly leveraged. And I also think the difference between a 50th percentile controller and an 80th percentile controller or 90th percentile controller is not going to change the trajectory of your business. You always want to hire the highest percentile, but the difference between a 50th percentile, someone doing growth, and 90th is actual … Your business, the trajectory of your business is changing. So we hired our first head of growth, I was searching for months and I just happened to have my friend Nate, who’s one of our angel investors, two-time YC founder, just emailed me. He’s like, “Hey, I’m thinking of shutting down my company and wanted to catch up.” And I’m like, within … I responded instantly and we gave him a job offer within two days. I jumped on him, I have to hire this person. And he did have some experience at SEO, so it’s not like … He had some context. He also had a personal brand and he was very well-known in the YC community. So I could see that he knew … It wasn’t that he had growth … He’d never done paid ads, he hadn’t done these things, but I was like, he had the right mindset for figuring out how to capture attention, how to get stuff done. And then these things like paid ads these days could be learned pretty damn quickly. And so if someone like me who had that experience, I wasn’t worried that we would … that he’d be fumbling around. So as he got busier, there’s a lot of stuff we needed more engineering talent on, he was getting blocked. And so we had someone else in New York, this guy on my team, another ex-founder who was somewhat … was not, like, deployed a lot of production code but was technical, had studied CS and could code some. And so I was working on some other onboarding projects and we’re like, all right, you seem like you have the right skill set where you’re very high agency, you just get … You hear … I think it’s people that you give them a small prompt and hey … So Neil, I was like, hey, my friend’s doing all these LinkedIn influencer work, you should look into it. And he comes back later and he’s like, “I have 10 influencers on LinkedIn lined up,” like within … And it was just some small prompt. Similarly, Jonah on our team who’s more on the growth engineering side, I explain the problem and I don’t need to micromanage the details. It’s sort of like having mini-founders within your company. I think, again, because you need people that are going to be highly leveraged. That’s what I found to be successful versus someone who has deep years of experience and they’re going to be able to talk about the differences between an ad group and a keyword. A lot of stuff … Again, you can hire contractors and you should hire lots of contractors. So if I need a specialist on PPC, I don’t need that to be in-house. I can hire a really smart contractor as long as I’m asking the right questions, I think, and it’s someone you trust. And so I think you want to hire generalists who can also understand how all this stuff ties into your product, ties into the roadmap, things like this.

MIKE DUBOE

So this is a very particular type of growth culture that you like to build. And I think that maps to actually the culture at many of the most successful growth teams. How does that extrapolate to other functions with Numeral?

SAM ROSS

Yeah, good question. I probably hadn’t thought about it as much. I mean, I think broadly a big part of our culture is everyone … We want them to be high ownership, high agency. Our job is to … Our clients have this nasty problem of sales tax. They don’t want to think … The whole promise is like, we’ll think … stop worrying about it, stop thinking about it. We’ll think about the nasty details for you. And so I think broadly, we always try to push, there’s a, really, culture of ownership and high agency. I mean I think all startups need this, but this is something that we just think about a lot. We also hired a lot of founders. We just aqui-hired another YC startup, Ruby Card, three YC founders, and I think they’re all incredible engineers, but we didn’t hire them just because they’re good engineers. I knew them from the batch. They just hustle. They figure stuff out, they’re creative-thinking. We don’t have any PMs in the entire company and we’re a 60-person company. It’s because good engineers, I think, can do a lot of that PM work or good designers can do a lot of that PM work. Again, I don’t know, I’m probably overly … I tend to hire generalists and it’s cognitive bias that I’ve actually tried to … maybe I need to slow down on a bit, but I think broadly I tend to be attracted to people who can make connections across things and are not just stuck in their lane because I find that people that are stuck in their lane, they may be really good specialists, but then they have trouble pulling in insights from other teams and working cross-functionally.

RISHABH JAIN

Yeah, that’s super interesting. And just to empathize, interestingly, we have only one PM at the company and, yeah, 75 people. So I can totally relate and agree with that. I’m curious to learn more about how you think about Numeral’s product as you continue to grow. And when you made this statement about a controller — 50th percentile controller, 90th percentile controller — is not going to massively inflect your business, I would suggest that you guys are there to make sure that, hey, the back office, we will take care of it for you.

SAM ROSS

Yeah.

RISHABH JAIN

You don’t have to worry about it. When you think about the vision of, hey, you don’t have to worry about it, how do you think about painting what the product has to evolve into over time? And how much of that becomes, for lack of a better framing, agentic in nature —

SAM ROSS

Totally.

RISHABH JAIN

— in terms of your entire backend can be agentic and actually you don’t need to worry about it. Or do you have a different vision for how you move from sales tax into other ways that it helps improve your business and then the leverage that your software provides to that business over time?

SAM ROSS

Great question, I love that. Our business — we help startups and eCommerce companies with their sales tax compliance — and one problem I had with … So the way I found this problem set is, I was running a bunch of my own eCommerce businesses, still own a vitamin gummy business, does about $10 million a year in sales. And so while running it, I had to go use these incumbent tools. The famous one is this company called Avalara. They’re owned by private equity now, going public again soon, actually. What I found is they were selling me a tool to solve sales tax, and I was like, I’m a founder of a DTC brand. I want to spend my time on Facebook ads, I want to spend my time on product development, I don’t want to log into this tool and try to understand the nuance of West Virginia’s tax. I was literally on the phone with West Virginia at one point yelling at some poor woman there about some penalty waiver. And I’m like, I really want a service. And so I think what … For a lot of categories, what I think we’re going to see is instead of selling you a tool, you’ll be sold more of a service. A lot of the ways it works in sales tax is, as you get bigger, you might hire an accounting firm or some sales tax consultant and then they manage an Avalara tool for you. And I don’t think you’ll need these kind of at-the-edge services anymore. So maybe it’s similar for CRO, I hired some CRO agency in your space and then they manage some landing page builder tool. It should all be consolidated down to one. So I think you’re going to start seeing some merging of services and software. Broadly, that’s the pitch of Numeral. Like, hey, you get physical mail from the states, we’ll scan it. We’ll read it. Again, AI is doing a lot of this work in the backend, but you don’t care as a user. We’ll call states on your behalf, things like this. And so all those small things just go away. And so a busy controller can go and spend very little time on it, it just runs in the background and then they get more leverage on their time. As we evolve, we continue to add this in, we’re starting to really now focus on — particularly for SaaS businesses — international markets. And so again, you talk to … How does a big SaaS company handle international sales tax? And they have six or seven different agent … accounting firms across the world. They’re working with KPMG Korea, some other firm in Kosovo, and you’re managing all these different service vendors, you manage all these different softwares. Again, I think everyone’s trying to get more leverage on their time. That’s how the economy gets more productive and you have leaner balance sheets. And so I think broadly that’s the direction we’re going, is it feels like a service, but most of what’s powering it is software. And then the reason you can now compete with service businesses is because AI lets you … It can read the mail, it can do a lot of these … It can figure out what category of product you’re selling, things like this.

RISHABH JAIN

Is your sense —

MIKE DUBOE

You’re hitting on a —

RISHABH JAIN

Oh, sorry. I was just going to follow up on the “it has to feel like a service” point, which is … And then Mike, I’m curious what you got curious about too. Do you think that becomes the expectation? So enough companies do this and all of a sudden the rest don’t have a choice essentially? Essentially at some point we get to the point at which, yes, there’s a set of things that can remain as quote-unquote tools, like your email, but enough companies do this and you flip, right? Do you think we flip and if you think we flip, at what point do you think we flip? And what would you say to other founders? Yeah.

SAM ROSS

Yeah. I think it depends on the category. So some categories maybe it’s always going to be a tool because it’s … I think if the function is something that’s really core, you really need to own it in-house for some reason, maybe like Figma, maybe, I guess the AI used a service firm to do that, so I don’t even know. But I think certainly anywhere that you tend to hire an agency to do work for you. I mean even … You can even see … We’re talking about Facebook ads, right? In the past, you had to hire an agency to go and they weren’t even doing the creatives. All they were doing was moving budget around between ads, and now someone on your team can set up a Facebook ad much simpler. And so I think, yeah, the bar is always getting raised for what you have to do in software. The expectation — I think people often ignore this — the expectation of what the software was like 20 years ago was way lower. You can have a really crappy piece of software, but because the competition was like service firms doing things by hand or things like that, even just having something just good enough, and the bar is always going higher. And so I think, yeah, if in your industry there’s services firms or implementation specialists or things like this that manage … that tend to often make it easier for your software, you need to figure out how to close that gap and be able to deliver that. And I think you might have to … It could be a change in the business model broadly too. Some of these, even if AI can’t do it, you might just have to have more services with your business and take lower margins. But if that’s just what the customer expects, you might have to do it. Now, I don’t really want to manage a big services firm. That sounds like a headache. I’ve run many agencies before and I don’t like doing that, but you have to meet the customer where they are.

MIKE DUBOE

Yeah. I was going to ask about this topic. It’s a very zeitgeisty topic in AI software right now, I think for good reason. Replacing spend on services is happening, I think faster than many people had anticipated. In your industry … I think the value prop for Numeral is very clear, and what you’re saying makes complete sense. How do you actually staff the team to support some of these transitions? I know you said you don’t want to run a massive services arm, but what are the right levels to invest in that right now? And how do you think about either leaning further into that or weaning off?

SAM ROSS

Yeah, good question. I think it’s hard. You’re always trying … I think all founders right now are thinking, “How do I insert more AI into my business and automate these workflows?” When we started our business, we were doing — I think I talked to you earlier, Mike — we were doing everything by hand. I didn’t even have good … We didn’t even have … We had basically no software when we started. We were just doing filings by hand. I had a team in the Philippines I had worked with when I used to sell T-shirts. I hired this guy for $3 an hour to design T-shirts for me a decade ago. Then I had him now going and submitting tax filings in Alabama’s portal. Over time, we built up a large international team just to quickly bootstrap stuff and just brute force and get stuff off the ground. And so, now we finally automated all that work, and a lot of it is just automating without AI, it’s just, okay, the state finally gave us access to their API or SFTP portal, whatever, and we can automate, but there’s a lot of states that don’t have anything like that so we’ve had to do a lot of work with these browser agents, we use Browserbase and go and do that. And so we’ve eliminated … it was something like, in the last three months, something like 30 full-time roles just with … A lot of it’s just like data entry, but still, it got the job done. Or similarly, I had two guys that were reading physical mail for all of our clients and trying to categorize it, and they were getting overwhelmed. They’re like, “Hey, we need to 3x our team in the next month or two” — because we were adding a whole bunch of clients — “or else we’re going to die.” And so my co-founder, Matt, is like, “What the hell? We can’t keep adding folks abroad.” And so we actually … A lot of it’s just prompt engineering. It wasn’t like we had to build crazy-sophisticated tooling, but all of a sudden we could automatically go and categorize thousands of pieces of mail that were coming in each week. And I think there’s still parts of our business where there’s human operations, but I think you just take a hammer to it and each week, each month, you find ways to automate. And there’s some things that, frankly, the technology is not quite there yet. And I think you just keep a close eye on it and these things keep getting better and the capabilities keep getting better, the infrastructure keeps getting better. But it’s a hard balance because I think as you’re building a startup, you get requests from clients, you want to be able to serve them, you want to be able to — again, our pitch and my vision is if the client has a headache here — sometimes the team will be like, “Hey, should we do this? I don’t want to do this. This sounds like a pain.” And I’m like, “No, that’s the whole thing we sold them, and that’s the whole point. We have to go do the painful stuff.” It’s a big tension we think about a lot, and I don’t have a clear answer for everything.

MIKE DUBOE

Are there other opportunities? You’re sitting in the midst of a pretty interesting topic in finance and compliance right now, which … There’s a bunch of opportunities for AI to kind of inject into compliance-related problems. What opportunities do you think are under-discussed, and maybe we could zoom out a little bit from Numeral in the tax space and just more broadly within finance ops and compliance, what are you seeing that you think is ripe?

SAM ROSS

Yeah, I am so probably focused on my little niche of the world, I don’t have a great vision, but I would say broadly what I’ve seen is I think compliance specifically is a really good area right now where … Anywhere there’s a lot of paperwork, there’s a lot of rote, fairly deterministic decisions having to be made, I think tends to be … There’s a lot of room where you have a lot of back office workers or you have people doing low-level work. I think anywhere that you have huge swaths of team abroad doing kind of, again, work that doesn’t require complex decision making, is probably pretty ripe. I think, yeah, the advantage AI gives you too, especially anywhere that you want speed too. So we could have our team doing … Having our team doing manual filings, a lot of the problem or a lot of problem we were solving for was not even a cost-saving problem or even accuracy problem. A lot of it was speed. Because now, again, instead of having you have to wait, okay, we have to get all our filings done every month by a certain due date, but now it’s like a single button. The filing’s just done instantly. So I think areas that having more speed on the compliance I think is also, there’s probably a lot of interesting opportunity.

RISHABH JAIN

Yeah, that’s incredible. I was curious a little bit about your past. I know Mike has had the privilege of seeing more of your journey over time. I’m curious, when you were running the house of brands, it was roughly at a time where there was a bunch of other large aggregators that got stood up, right?

SAM ROSS

Yeah.

RISHABH JAIN

And I wonder, how did you think about that model and, yeah, how do you think about that model now more recently as the ecosystem shakes out?

SAM ROSS

Good question. Yeah, I mean, I looked very closely. I had a whole pitch deck of doing this model, and I even almost joined my friend as a co-founder where they raised tens of millions doing this model. So I looked very closely at it at the time. The promise that was exciting is whenever you find a new private equity asset class, the first movers can make a crazy amount of money rolling things up. There was a lot of — on paper — why it was exciting. I think what happened, I think there was a couple of things that went wrong there. I think one is you have this sort of adverse selection bias, you said even with Open Door where it’s like the people most likely wanting to sell their businesses are the ones that probably know that … You want to sell it when you don’t have a lot of confidence in that business and you think it’s going to go wrong, whereas the founders who are really bullish on their business are probably the least likely to want to sell. And so I think you had that problem. And I also think there was this promise that you could get all this operational leverage or I could have one team manage five or six different brands, but I think it actually requires a lot of focus to get this stuff right. Especially, again, I think maybe doing this for a business with high retention like SaaS or even dentist’s offices or HVAC companies, whatever it tends to be, it is maybe easier and you’re just creating operational leverage. But a lot of these brands, they were probably run fairly well and there wasn’t a lot of low-hanging fruit to improve them. I remember my friend, he acquired this Amazon business. It was selling some commodity thing like staples or something, something just super commodity, but he was ranking number one on Amazon and all these keywords and so he was selling a ton. And then as soon as my friend bought it, the rank dropped and he called the guy and he’s like, “How come the rank dropped?” The guy’s like, “Oh, I have this list of 30 tactics I do every day. If I fall number two on Amazon, I have to go buy a bunch of fake orders and use these sketchy black-hat tools to get it back up to number one.” And if you’re a founder and you really care about the cashflow, you’re going to do all this insane tactics to get it to work, but as soon as it’s owned by some employee who doesn’t care, it’s not going to work. So there’s probably other reasons, obviously interest rates, etc. I don’t think the opportunity … There’s probably still different cuts of it where there’s opportunities — buying bankrupt businesses, I see some operators doing this quite well, or maybe trying to pull together … Again, I think the problem also is, eCommerce is this very broad topic, and it’s not necessarily … I don’t know if eCommerce is even an industry, it’s almost more like eCommerce is a way of selling, but what does Eight Sleep mattresses have to do with some Olipop sparkling water, whatever? They’re very different. And so I think maybe if you focus on a certain vertical or things like that. The other thing, too, is I think eCommerce … In the past you could operate these businesses with not a lot of discipline, but if you were just really good at growth in marketing, it covered for everything else, because you’re acquiring customers so cheap and you might not have that efficient operation where I think now these businesses, it’s a lot more a competitive advantage to actually be really good on the supply side, which is interesting. So maybe there’s something there if you’re really strong at some supply-side thing and you can get some operating leverage that way.

MIKE DUBOE

That last point is very interesting and resonant. And maybe … I think one of the premises that these roll-ups started on, or at least the ones that I spent time with, was that there would be some audience overlap and — outside of removing costs from your marketing org, having one centralized marketing org — actually be able to cross-sell. And that thesis you should have lower CAC and essentially higher retention across your portfolio of brands. I’m not sure that really panned out. And I’m curious, you don’t buy that thesis? Do you think there’s anything there at this point? Is it, hey, Meta is doing the lookalikes anyway for you, there’s not really much of a benefit in doing the cross-marketing?

SAM ROSS

Yeah, I mean, I definitely don’t buy that. I don’t know, I’ve always been someone, it’s like, yeah, you turn on your email, you turn on your SMS, but that’s never the key driver. It’s always number three, number four reason your business is growing. At the margin, that stuff’s nice to be able to cross sell, but I don’t think that’s … Ultimately all these brands are just like, they’re addicted to the Zuck’s heroin and you can’t … Until someone else figures something else out there, there’s no way around. These things are nice on paper and look good on a pitch deck, but if you talk to a real founder that’s actually sold these things, they’ll be skeptical, I think.

MIKE DUBOE

So maybe let’s spend the last few minutes here going deeper on a couple of the tactics. Feel free to not give away all your secrets, but I think it’s been very clear to me that you’ve tapped in on LinkedIn as a really powerful growth channel. That in itself is not unique but I think the way you’re running it is unique. And so talk to us about — maybe at the highest level — what’s one marketing channel you believe is going to be 10x as important as today within the next five years? And maybe what’s one that you think will only be 10% as important?

SAM ROSS

Yeah, good question. So yeah, I’m always … Whenever it’s working for me, I’m always, “I’ll tell you in six months, I’ll tell you in a year from now.” I would say, looking on to the future, I don’t know, I don’t think there’s any obvious things that we see right now to me that it’s like this is the trend, this is going to change. My guess would be generative AI allows for some new, what’s the next TikTok, what’s the next kind of social media app? And I’ve really focused on the media side of it versus the social side, meaning I think as we’re able to generate content much cheaper and easier, there’ll probably be just different … It’s like how are people spending their time during the day? And there probably should be different apps on your phone that you’re going to be spending time on. And so I think I would look out for whatever starts growing there. And then as soon as something like that monetizes often there’ll be probably a lot of opportunity there. At the margin, it seems like people are shockingly spending more time on their LinkedIn feed, which is like, it’s super unsexy, everyone makes fun of it, but a lot of people spend a lot of time there. And so we’ve invested on that and taking a whole series … There’s tons of different tactics you end up doing there on the ad side, on the organic side. In broad strokes, again, I think our approach has been don’t create boring content that you’re going to scroll right through. Try to create stuff that’s engaging. I think it’s always a little dangerous. Sometimes we will post stuff that’s a little at the edge of maybe what’s appropriate or is this really good for our brand? And so I think we’re spending some more time like that thinking about how we reconcile creating engaging content that’s humorous and fun. But you have a brand like Cluey where they’re really good at getting attention, but I think a lot of people are like, these guys are jokesters. I’m never going to use their product. And ultimately the goal isn’t to get impressions, it’s to get people to actually go and use your product and have some respect for what you’re building. So, yeah.

RISHABH JAIN

That’s super interesting. I was wondering, we actually recently talked to Bryan Cano, who I guess, similar to you, is thinking a lot about how he’s going to use gen AI tools to inform his marketing processes and what types of content he’s producing and things like what is catchy, what is in the zeitgeist. And a lot of this, the way you’re talking about it feels like — and your own background — it’s like you’re borrowing consumer tactics and bringing them into B2B, right?

SAM ROSS

That’s right, that’s right.

RISHABH JAIN

As you do that, how do you think about the assist? What are the mechanisms that you have found that are kind of interesting assist mechanisms of using some of these AI tools to come up with some interesting ideas? So without giving away any particular tactic, like, hey, this particular idea of a creative or this particular thing, hey, here’s a way in which I thought about it and how these tools enabled me to tap into something special here.

SAM ROSS

Good question. To be honest, I don’t know if we’re using a lot of AI right now. I think we’ve thought about it for more content generation or content creation, but I think ultimately the quality of content you can produce right now is still below the bar of something that we could write ourselves. But I do think going after your point about the zeitgeist, I think the only thing that’s interesting is, again, we’re talking about the level of expectation for software has gone up, but I also think the level of expectation for your marketing as a SaaS founder has gone up. In the past, again, it’s always a joke at these B2B SaaS companies how cheesy and bad their marketing is. Like our competitor Avalara, it’s this big orange ugly company, you see them at the trade shows. It’s so cringey and embarrassing. And so I think the bar is also … The bar is always going up everywhere. We’re in a purely capitalist ecosystem. And so I think more and more, I think a lot of these SaaS companies are taking … Consumers always sit, I think, at the forefront, at the avant-garde of marketing. And so I think you’re going to start seeing a lot of these SaaS businesses — because they have a lot of money, they have to compete — are trying to take more and more a page from consumer brands.

MIKE DUBOE

Maybe one last question to wrap here. So we’ve spent … I think one of the themes of this conversation has been taking consumer marketing tactics and applying them to your B2B space. If you flip that, what have you learned in the B2B world that you think would make you a better consumer marketer today?

SAM ROSS

Good question. I think a couple things. One is just B2B teaches you — again, I keep harping on this, but retention is the only thing that ends up mattering. And so if I start another consumer company — remember, I started off, I sold jewelry, and I sold $30 million of jewelry, but almost zero retention. Insane. People would buy, they’d search for like, “gift for my daughter’s piano teacher,” we’d have a perfectly designed jewelry, it says “to my piano teacher.” And so we were able to have all these growth hacks, but they would never buy another necklace from us ever again. And so I was like, okay, I need to build something more retentive. And we started selling supplements online, vitamin gummies. We’re like, wow, people, our average will buy three or four times from us. That’s amazing. But then at the end of the year, 90% of your customers have churned. And so now to go to a business where 95-plus percent of my customers are still there at the end of the year, you really understand compounding. To me, it’s a little bit like, any time I do a consumer business, I’d be so obsessed around retention and almost not even think about anything else from there. In terms of the marketing, I think it’s a good question. I don’t know if I have a lot of obvious stuff. I think the biggest thing is I’ve just gotten more comfortable with doing things that I can’t measure well and just having more … Like whereas when I was doing … You’re running an eCom brand, you can measure things so well, it’s easy just to only invest in the kind of anywhere you can measure a CPA. And so you end up putting all your money on Meta and Google or wherever it is, and then you just never spend any money on brand marketing. We used to always … A lot of performance marketers trash … It’s like this war between the brand marketers and performance marketers, and they each think they’re idiots. And so I think I’ve learned a lot more to … there’s just certain things we’ve done. You’re like, wow, that definitely worked. And so I think being able to take more of a leap of faith and entrust that sometimes doing certain sponsorships or certain other tactics that you can’t measure, just taking faith in that I think has been a learning for me. And again, that’s a me thing. It’s not like all consumer, but it’s performance-focused consumers.

MIKE DUBOE

Yep. Sam, thank you. It has been a great conversation. I’m excited to check in six months once your current portfolio is decayed and you’re on to the next. And so, always fun chatting with you.

SAM ROSS

I’ll tell you all the tactics that are working. Thank you guys.

MIKE DUBOE

Thanks, Sam.

RISHABH JAIN

Thanks, man.

‍