When Gross Margins Matter

In turbulent times, it's critical to to understand the fundamental finances that drive the business. One of the key metrics both investors and founders look at is gross margins. This conversation digs into the what, when, and why gross margins matter for both early and late stage startups.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  In turbulent times, getting down to the fundamentals of how your business operates is critical. One of the key metrics to understand is gross margins. This week in an episode from June 2020, A16Z General Partners Martin Casado, David George, and Sarah Wang discuss all things gross margins, including what it is, strategies for calculating and defining it as a company matures, and how this one core business metric drives everything from cloud costs to valuations.

• 0:29Speaker 01

  The content here is for informational purposes only, should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. For more details, please see a16z.com/disclosures.

• 0:48Speaker 00

  Hi, and welcome to the A16Z Podcast. I'm Amelia. Gross margins, basically a company's revenue for products and services minus the cost spent to deliver that revenue, are one of the most important financial metrics in any startup. But figuring out what does and doesn't go into margins, the ideal ratio for a growing business, and how this ratio changes from early to later stages of a startup is not as simple as it may sound. So, A16Z General Partners Martin Casado, who invests in early-stage enterprise startups, and David George, who leads our growth fund, along with Sarah Wang on the growth investing team, share why margins matter from their vantage points, as well as how modern software

• 1:32Speaker 00

  business models and technology shifts are changing the way we think about them. We also briefly touch on how margins can impact valuations. But we begin with why gross margins matter at all, especially as companies mature. The first voice we'll hear is David's.

• 1:47Speaker 02

  To us, a later-stage company is one that has found the all-elusive product-market fit, and the risk that we believe that we're assessing is execution-based risk, right? Not a binary risk of whether the product will work or not. When we're doing diligence on a later-stage opportunity, can we take some reasonable stab at what the long-term margin structure looks like? And... Do we have some sense of the market, the competitive dynamics, you know, act number two and act number three from a product standpoint, such that we can make some guess as to the persistence of the growth rate for the company over time? So is this something that can grow at very high rates for 10 years?

• 2:32Speaker 02

  If so, that's the kind of thing that gets us really excited.

• 2:35Speaker 00

  So I know this might be obvious, but why do high gross margins matter in the first place?

• 2:40Speaker 02

  Ultimately, as a company gets big, we look for companies that can generate a lot of cash flow. If you have high gross margins, that gives you more of your cost structure that you can spend on things like sales and marketing, product engineering, et cetera. So what we typically look for for a software company is their cash flow margins, right? So this has now gotten a lot easier in the last three or four years because there are more mature software companies in the public markets that are generating meaningful amounts of free cash flow, right? There's a couple of different ways that we typically try and figure out the margin structure. One is just benchmarking. So here's a really simple rule of thumb for a company.

• 3:22Speaker 02

  We typically see software companies that have gross margins of 70 to 80%.

• 3:27Speaker 00

  Right.

• 3:28Speaker 02

  And then below that, there's three big buckets of cost items. One is sales and marketing, which I'll circle back to. Two is research and development or product. And then three is what we call G and A, which is overhead functions. As a company gets more and more mature, what we would typically see in R and D and G and A are like 15% and 10% respectively. And then we say your sales and marketing expense is typically a function of how fast you're going to grow and your retention rates. And so you combine those three cost buckets with that 70% or 80% gross margin that I referenced earlier.

• 4:11Speaker 02

  And what that spits out is your long-term profitability. So we can benchmark each of those buckets against companies that are out there and more mature and say what's similar, what's different. And then we can also go deep into the specific company itself that we're assessing and say, hey, where are they at on this right now? What kind of progress have they shown over time? And how much confidence do we have that they can end up looking like? Salesforce in this regard, or Atlassian in this regard, or not.

• 4:41Speaker 00

  Right. So it sounds like a perfect formula. Is it quite that precise in real life?

• 4:46Speaker 02

  No, it's definitely not that precise.

• 4:50Speaker 02

  I think we way oversimplified it. Every business is different.

• 4:53Martin Casado?

  So I think this is a very important discussion because gross margins is far more applicable to mature businesses. And businesses that have very specific COGS. COGS is cost of goods, which is just like how much it costs for the product that's shipping out the door. And the reason I say that is, early on in a startup, sometimes it's hard to know how much of your COGS is core infrastructure. And so there's always this blurry line between what is effort that's going into building the product that's, you know, non-variable versus what is customer-specific, but still looks like R and D versus what is core infrastructure.

• 5:36Speaker 04

  I've actually got an example, or a quick story from my earlier investing days on, to illustrate that point, and also kind of highlight a mistake that we made. So I think there's a tendency to categorize things like software licenses and R and D, sometimes G and A. But I looked at a company that actually needed these software licenses for their product to be deployed. So this is literally the definition of a variable cost to serve. But the company and our team, we categorize this as R and D. And so it not only led to gross margins being off by five to 10%, but also as you think about modeling your business forward and sort of the expenses that you're going to need, R and D typically doesn't grow with revenue. It's more of a fixed cost.

• 6:17Speaker 04

  But, you know, this is software licenses, deploy your product, obviously is completely variable. So you get into danger of double overestimating your mid to long-term profitability, which also has implications for how much you can spend, or how much you're budgeting. Maybe not for this year, if you're still earlier stage, but over the next couple of years, as you move the business toward longer-term profitability. As companies think through what that looks like, you know, I think it does start with that first layer of, hey, what does it actually cost to deliver this revenue? Although I would agree with Martine that obsessing about it early on, when there is going to be likely a drastic change as the business matures, is not the right thing to do.

• 6:55Speaker 02

  Gross margins of companies often change dramatically over the life cycle of the company. So just because a company early in its life cycle has low gross margins doesn't mean that it will have low gross margins forever. So one of the classic examples of this is Workday. When they were going public, because services, their implementations were quite large, each deal was a million bucks or more, the implementation cost was a large proportion of their revenue. And so they actually had 40% gross margins. And fast forward to today, when it's a little bit more mature, they have 70% plus gross margins and look much more like a traditional software company.

• 7:31Martin Casado?

  So I actually think, I think this is great because I think these are kind of like broad situations that kind of clearly illustrate this tension between, you know, or at least the misunderstanding of what gross margin is going to pencil out over time. And that is R and D and it is cloud cost.

• 7:48Speaker 00

  So why do those two buckets matter? Can you break that down for us, Martin?

• 7:52Martin Casado?

  So the R and D, early on, you just don't know what is going to be fixed and what is going to be variable. I mean, now, maybe in some products, you know, you can have a fixed number of R and D. And by the way, I'll say, I've never seen R and D actually be a fixed cost. Actually, it's always, ever.

• 8:13Speaker 00

  It's a fair point.

• 8:13Martin Casado?

  I mean, like, just for example, so that, you know, I ran, you know, a $600 million, you know, global business and I would say 40% of our cost were maintenance costs, right? And then we kept growing R and D in order to expand sales. And so it's not linear, it's not super linear, but there was definitely a sublinear component. And even though like, like everybody, we didn't really kind of include that in COGS. But also different products just have a different level of R and D required to sell the next thing. You just don't know until you're fairly mature, how much R and D you can kind of pull out for a repeatable sale. The next one, which is very interesting, is cloud costs, right? So cloud costs have basically become variable costs. It used to be you'd buy a server and then you'd kind of sweat the server,

• 8:57Martin Casado?

  you know, over a period of time and then, you know, you'd buy a new server. But now, basically, for every bit of compute you deliver to a customer, it's likely that you're paying a portion of that to, you know, whatever cloud server costs. And, and more and more, this is becoming a significant portion of the COGS, right? Particularly in the, in the era of AI and ML, where they're compute intensive and data intensive. You just end up spending a lot of money on cloud costs. Now, if you go to an engineer and you say, oh, like, you know, this is a significant portion of our margin, can you reduce that? The engineers say, yes, we can likely reduce it.

• 9:38Martin Casado?

  But in this startup world, normally you're going for growth. You're not going for margins. And so you've got the, you know, you've got the board members saying, listen, go, let's go ahead and kind of, let's grow as much as possible. So the engineers will make decisions that will optimize for feature velocity. They won't optimize for efficiency. And then you're piling that technical debt on for a long period of time. And then, at some point, you know, you have David George or Sarah Wang, and they want to make an investment. And they're like, hey, wait, your margins are terrible, right? And then the startup's like, well, listen, but our, our, you know, Martin Casado, who was our Series A investor, said, go for growth. And now, now you've got this predicament because you don't really know what your COGS are

• 10:23Martin Casado?

  because you weren't actually focused on them. It may be the case that you can reduce them by a lot through smart engineering, but you don't really know. So again, there's this situation where you're in the early stages. Actually, understanding gross margins is difficult. If you're focusing on growth, it's even more difficult. And until the later stages, you don't really know how it's going to pencil out.

• 10:43Speaker 00

  So what do you do? So like, what are the, how do you not get to the point where a David George is giving you the side eye over your margin structure and he looks at your, at your spreadsheets? But how do you, as a founder, start thinking about this earlier? Is there any way to, to get ahead of this?

• 10:58Speaker 02

  I, I just want to say before Martine answers that question, I am way more polite than the side eye guy in the investor meetings.

• 11:04Martin Casado?

  I think this is just starting to hit the boardroom recently.

• 11:08Speaker 00

  Yeah.

• 11:08Martin Casado?

  All the boards I'm sitting on, even at the Series A and Series B, were actually looking at cloud costs and realizing that, you know, double digits of margin are being contributed to them. And so what I think is happening now, very early on, we actually have the discussion. And the discussion is, you know, this is going to be important once you're larger. We don't want you to make architectural decisions that you can't unwork early. We want to understand what is basically a law of margin physics here, which means like you, you're not going to be able to reduce it versus something that you're, you're taking a shortcut now that you can optimize later. Let's document all of those. And so at least you know that that discussion's going to come. Because I think, you know, as of five years ago, there was just an assumption that it was software, you

• 11:52Martin Casado?

  have 80% margins, even though you're doing these very cloud compute intensive things.

• 11:56Speaker 00

  At the later stage, kind of looking back at these companies and as they progress to the next stage in their growth, what do you see or how do you advise companies?

• 12:05Speaker 04

  First of all, I definitely agree with Martine that optimizing too much on margins early on doesn't always make sense. And, you know, by the way, by no means do gross margins have to be in steady state by the time you start raising those later stage rounds. That said, you know, I think the ratio of story to results has to have come down by then. Because at the later stage, this is really when the company transitions to more of a show versus tell.

• 12:32Speaker 04

  At this stage, I think you're going to get a lot of credit for, one, showing that you've demonstrated meaningful margin expansion, regardless of whether it's at steady state, could still be very far from that. But then also, two, having a credible path to what you believe steady state margins will be. Some of this is going to be very well trodden. So cloud cost optimization, economies of scale. And then there's other ones that are more specific to, to just your business. Regardless, having a very tangible roadmap is going to be extremely valuable at this stage.

• 13:01Speaker 00

  Right.

• 13:02Martin Casado?

  From a startup standpoint, it's very difficult to predict how much of the work that you're doing is actually repeatable until you've talked with a whole bunch of customers. It may be the case that after you've serviced 10 customers, you've kind of tempered the product enough that it's fully repeatable and then the amount of implementation required goes down per customer, but it actually may not be the case. You know, notoriously in management, the number of systems you've got to connect to is this very, very long tail and that just ends up being part of your COGS. And actually, we can go through a number of examples in early companies of why you don't really know what the COGS you're going to pencil out. So let's say if you had a company that was basically contract engineering, which means for every new customer, you had your

• 13:43Martin Casado?

  engineers build new features. That's clearly a very low gross margin business because as part of your COGS is engineering, right?

• 13:51Speaker 00

  Right.

• 13:51Martin Casado?

  Now, if you go kind of a little further towards a pure software model, like maybe there is some scripting that needs to be done for like connectors. So like there's a number of companies where they're like, you listen, you know, we've got a core platform, but you may have some special systems we integrate with. That's actually, we'll do that, you know, for you. It turns out in the industry, there's a long tail of those. And so there's, this isn't core engineering, but it is kind of scripting and connectors, but that's part of the COGS, right? And then, you know, you go a little further, maybe there's just some basic configuration and implementation that needs to be done. It's not actually development. It's not actually scripting. It's like, you know, actually just dealing with config files. That's lower COGS. And then, you know, in some models, like maybe there's nothing that you have to do. And that would be kind of

• 14:31Martin Casado?

  the, the highest gross margins. There are many cases where even if it's not core R and D, but it is some sort of scripting or some sort of integration, it's simply something that you're going to have to pay per customer.

• 14:43Speaker 00

  Okay. So we've covered how margins can be messy and how they evolve over the course of a startup's growth. So now let's talk about how modern software companies, particularly SaaS companies, et cetera, are changing how we analyze margins. How did we used to analyze them? And then what is different now?

• 15:01Martin Casado?

  A lot of like original views of gross margin came from the days of shipping software. And the world is very different. So if you ship software, you have an R and D team, they write software, you put it on a CD, for those of us that remember those. And then you ship that CD and you're done, right? And so, and so what are the characteristics of that? One of those is you're done with features the moment the CD ships. So you don't have like, you know, a lot of feature velocity. The second one is you're not running it on your own servers. You're running it on a, you know, the customer's servers. And so in those cases, there's things that you don't really think about. Like you don't think about like, what is the amount of compute relative to dollars? What is the amount of like data munching relative to dollars?

• 15:44Martin Casado?

  You're like, listen, I've kind of put these bits on the CD and I don't have to look at it again. And you've got this, you've got this great business. On the other hand, today, almost all software, not all, but so much software is actually delivered as a service. And so there's a couple of things that implement. So you have operating teams that actually manage the operations, like the care and feeding in that service. Right? In order to deliver it, you need to have those operating teams. That's different too. You know, you are paying for the infrastructure and you know, the efficiency of your software greatly impacts your margin, which you don't really have in the other world.

• 16:20Speaker 00

  Right.

• 16:20Martin Casado?

  Three, there's no reason why you can't have featured velocity that will absolutely increase the, your customer base. Because the amount of features that are dropping will unlock new markets, which you know, is going to have a hand on your growth. And so that's why this is very interesting because we are at such an inflection point with SaaS and with infrastructure.

• 16:41Speaker 02

  The one fundamental business model change that's a huge benefit that's come along with this trend has been recurring revenue versus one-time sales.

• 16:51Martin Casado?

  Hundred percent.

• 16:52Speaker 02

  Right? It's so, I mean, think about that, like the quality of being able to collect your revenue on an ongoing basis as opposed to just selling a disc once and being done with it. You know, yeah, you'd have 20% maintenance per year, but now all of your revenue being in the form of subscription is just such a tremendous advantage. And I think that's part of the reason why software companies have been valued so much higher than they, than they were historically.

• 17:18Speaker 00

  Why is having your revenue in the form of subscription such an advantage when it comes to your margins and valuation?

• 17:24Speaker 04

  So one of the big advantages of subscription business models is that they add predictability to your revenue stream. For every dollar of COGS or sales and marketing that you're spending today, you know how much they're going to bring in not only today or this year, but actually also next year and the year after that. If you think about it, company management are actually the real, you know, ultimate investors because you are deciding at every single moment, what is the highest value-added place that you can invest your resources, your time, your energy, and the money that you've raised? And this predictability of your revenue stream makes it easier to optimize for your margin because, one, you know what you're spending, and then, two, ultimately what will net you from a dollar perspective.

• 18:09Speaker 04

  Obviously, strong retention goes hand in hand with this. And, you know, in terms of valuation, fundamentally, when businesses are trading on high revenue multiples, you hear a lot of that in the press. What does that actually mean? The fundamental underlying reason is that the markets believe the company can grow in the long term with steady cash flows. And when businesses have this high retention in a subscription business, companies can trade at these higher multiples because investors have the conviction in the consistency of both their long-term growth and their long-term margin.

• 18:43Speaker 00

  That's really interesting because, one, you know, external investors can see predictability. But, two, the internal operators, the management team, can also see predictability, and then they can make better decisions. So it has a sort of twofold advantage.

• 18:58Speaker 04

  Absolutely. If you think about it, if you think about a founder or management team as an investor, you know, when you put that engineer or that dollar of, you know, the equity that you've raised into a particular line of the business, does that get you 50 cents one time? Or is that something that's bringing you a dollar and a half, you know, for the next five years? And that's going to influence where you're going to invest, again, your time, resources, and money.

• 19:26Speaker 02

  I actually have one more sort of nuance around gross margins that's emerged from Yet another kind of new type of business, which is businesses with a freemium component.

• 19:36Speaker 00

  Hmm.

• 19:37Speaker 02

  So this has really only emerged probably in the last, I don't know, eight, 10 years or so. But for companies that have a freemium component, so a big user base that they're not actually monetizing, but ostensibly is, is, can in the future be monetized. We've seen varying approaches of how to treat the cost of, of serving those. Um, some people put it up in cost of goods sold. Some people put it in sales and marketing. What we've been telling our companies is, hey, look, this is basically the whole reason you have this freemium line of business is as sort of lead gen for customers that you can monetize. So we would like to see that cost of service go down into the sales and marketing line as opposed to in the COG line.

• 20:23Speaker 02

  So, you know, Slack obviously has this dynamic and, and they keep it in their sales and marketing line. And so they have very high gross margins. In contrast, when Dropbox went public, they actually had it in their cost of goods sold. And it kind of confused investors. So what we've been telling our companies that have this dynamic is, hey, very clearly call out this component, be strict about how you define it, but put it in sales and marketing.

• 20:45Speaker 00

  Right. So another common line item is implementation and services. So how do you think about that component?

• 20:51Speaker 02

  So one of the things that I typically see when a company is in the process of going from kind of early stage and trying to validate that customers actually want what they're selling to making it a repeatable process is there's often gross margin leakage or like inefficient gross, gross margin in the pricing that gets attached to implementation or the services associated with, with selling the software. So, you know, if you've got three months of implementation work that needs to be done, um, you know, oftentimes an early stage company is in the habit of just saying like, okay, these are my first few customers. I have to make sure it works.

• 21:32Speaker 02

  Um, I don't want to charge them for this, or I want to way undercharge them for this just to make sure they're happy. And then as you scale that up. That's sort of a hard cycle to break out of. And so you'll often, at least I've often seen a bunch of companies that are sort of call it like in the 10 to 20 million of ARR range. They have what we would characterize in late-stage world like software, like gross margins for the software piece of what they sell, but they're losing a bunch of money on implementation, which is dragging down their overall cost. To me, that's, you know, something that can be fixed. It also, you know, if you kind of boil it down into simple terms, it's kind of priced also, right?

• 22:17Martin Casado?

  I think this is such an important point that kind of harkens back what we're talking about previously, which is it is not clear to me that implementation can always be fixed. This is where I think a lot of funny games get played in this area, which is there are some products that when they get mapped into a customer environment, they're mapping onto a very, very fat, long-tailed interface, right? And it's, it's integration with systems, it's dealing with legacy systems, it's maybe a bunch of data work that's happening. And it's tempting to be like, oh, we just haven't waved the magic software wand enough to make this repeatable. And in some cases you can do that, but in other cases, that's just too long-tailed and too complex and too eclectic, and you just end up with a lower margin business and it's

• 23:02Martin Casado?

  actually inherent in the product-market fit. And so I, I think it's great that you brought that up, David, because I, I agree 100% with what you're saying. The only thing I would like to point out is that I think we like to will away complexity and you can't always do that.

• 23:19Speaker 02

  Yeah, I totally agree. As you scale up, it makes sense to charge for that complexity, right?

• 23:25Martin Casado?

  100%.

• 23:25Speaker 02

  Charge, charge, charge an amount that like, hey, you're not going to lose a ton of money when you're going through this painful process with the customer. And oh, by the way, if you go through that painful process and it's a really tough services challenge, right? Like, again, this is exactly like Workday that we were talking about earlier, right? Like, it, it is a huge cumbersome. Implementation process, but like that just makes your software so much stickier once it gets in, right? The point I would make is like just charge for it, right? I know that I know that's easy to say.

• 23:56Martin Casado?

  I think you're absolutely right. I think this is a very critical point. And I just think it means that if you're running your business, don't delude yourself to the true nature of the problem. Because sometimes I feel we're like, oh, we just haven't focused on that, which is why it's complex. Oh, if I just put a couple of engineers, that will go away. And I'm telling you that's sometimes not the case, you know? A very, very classic market segment for this is ETL. It just turns out data is messy and you have lots of data sources. And it doesn't matter how magically your product is. It's going to require a lot of implementation. And so you're far better as a founder to just like acknowledge that. To David's point, you know, suck it up and charge for it and make sure that's part of your business.

• 24:40Martin Casado?

  Rather than kind of hold on to this hope that it's somehow you're going to address it in the product. Cause it just may not happen.

• 24:46Speaker 00

  What about emerging new technologies? Like, for example, AI and ML. How should those founders think about managing their margins over time?

• 24:54Martin Casado?

  So, you know, it's not clear to me that if you look at, you know, what we broadly consider AI ML companies, that they even have the same margin structure as SaaS. And I think there's, there's two reasons for this. One of them is fundamentally a technical reason, which is the amount of, of computation and data handling costs actually increase over time. In traditional software development, you know, you do the software and then you can optimize it. And then the cost per compute over time reduces because, you know, the cost of chips gets cheaper. The cost of servers get cheaper, the cost of cloud servers get cheaper, you know, et cetera, et cetera. And there's also a finite amount of work to do that you can optimize.

• 25:39Martin Casado?

  In AI and ML, the accuracy of your solution is dependent on the amount of data. And the amount of data required increases. And so it's not clear you can actually get the same levels of cost efficiencies.

• 25:51Speaker 00

  Right.

• 25:51Martin Casado?

  Because in order to improve your product, you may need to deal with 10 times more data. And then to improve it again, maybe a hundred times more data. So the tail of complexity in the problem domain is just quite a bit different. And, and you actually see this, like if, if you have two companies that are trying to do, let's say some sort of a vision application, let's say there's two companies that are building drones that identify crops. You know, one that's been doing it for a very long time, maybe have ninety six percent accuracy, and one that's been doing it for maybe, you know, far less time will have like ninety percent accuracy. So getting to those extra bits of accuracy requires much more investment, you know, the more accuracy you get, which means that you'll always be kind of fighting this, this kind of data management accuracy game. And that fight

• 26:36Martin Casado?

  will get harder as you get, as you go along. It doesn't actually get better. The second reason is, and this is a bit more of a difficult one to describe, but I'm going to try to do it in a general way, which is sometimes we see AI ML companies basically taking operational costs from their customers and internalizing it. So it's, it's almost, it's almost the exact opposite. So I'll give you an example. Let's say that I'm building my AI ML company that automates, you know, filling in entries, you know, data entry. That's what I say. I go to a customer, I'm like, you have all of these people that do data entry, I'm going to automate it for you, right? So the customer's like, that's fantastic. I got to take this kind of variable cost of data entry off of our books, you do it. And then if you actually look at what the startup is doing, it's actually just, you know, internalize this

• 27:21Martin Casado?

  variable cost, and it does require some humans to do it. Not all of it's automated, and it still has the data handling, the AWS cost, et cetera, et cetera, et cetera. And so, you know, even though they've got like good growth and good revenue, good customers, the reason they have it is they've basically taken somebody else's operational costs and put it on their books. And the, the nature of these problems, to the first point, is you can't always make that go away. It may be an increasing operational cost, even if you have software doing it as opposed to humans or software aided with humans, it still is, you know, maybe, you know, it'll still impact your margins. And so, so we're not sure, like, I'm personally not sure if these AI ML companies have the same margin structure as software. And there's definitely a range of them, but like, I think we're seeing

• 28:06Martin Casado?

  them to be lower. And of the boards that I'm on, we see fifty percent margins to be pretty standard for this class of company that's really reliant on, on data, for example, in order to create a solution.

• 28:16Speaker 00

  And I think that actually gets to one of the underlying anxieties around margins, which is if you have software-like margins, then you also expect that you should get software-like valuations. And I think this is part of why people get hung up on, well, I can't have all these services because I need to be able to have those 70, 80% margins. And so can we talk a little bit about how y'all think about valuations when maybe the traditional 75% software margin, therefore you get software multiples? Does that still hold true in these scenarios?

• 28:46Speaker 04

  I think the first principles way of answering, I think maybe the question you asked Amelia, which is why do gross margins matter? A little bit tongue in cheek here, but I actually think they don't matter per se. What really matters is the return on invested capital that a business generates. For practical purposes, it's often easier to build a higher return on capital if you have a higher gross margin because you have more points to reinvest back in your business, whether that's product or sales and marketing. But there are a lot of low gross margin businesses with a deep moat that can generate very high returns on invested capital. And it really is, at the end of the day, how are you building a durable competitive advantage that makes the particular service

• 29:27Speaker 04

  or product you provide indispensable and irreplicable to the value chain? And by the way, the interesting thing is oftentimes those companies do have pricing power to tie it back.

• 29:38Speaker 00

  Right.

• 29:39Speaker 04

  Not only do you maybe start charging for implementation, but we see a lot of companies as they build that brand or that moat, start to actually charge more for the software. Or if they're a marketplace, maybe it's the take rate, especially in something like vertical software where you can build relatively high market share.

• 29:55Speaker 02

  Yeah, there's maybe one example to bring this to life, right? Because these companies are highly interrelated with one another. So Alphabet or Google is one of the best companies, I would say. If you were to ask just what are the best companies out there in terms of competitive differentiation and moats and defensibility? Google is high on the list. They have 55% gross margins. Contrast it with a company, also a very high-quality company, but I think few would argue a better company, like Booking or Expedia. And those companies have 90% gross margins.

• 30:33Speaker 00

  Right.

• 30:34Speaker 02

  But they actually pay a massive tax to Google themselves. Booking and Expedia are actually the largest customers of Google. So I think that's an easy way to see it of like... In that specific example, like where does the power lie? Like it lies with the business with much lower gross margins than the company with with higher gross margins.

• 30:54Speaker 04

  One thing that that I wanted to just add is I think this conversation has been very focused on enterprise SaaS.

• 30:59Speaker 00

  Yeah.

• 31:00Speaker 04

  And and so a lot of this may be more directly applicable to that. And, you know, I do want to recognize that if you do take something more like a marketplace model, you know, I think we have seen a lot of businesses successfully take gross margins much higher. Because the play there is really to, it's, you know, it's a winner take all market. The play is really to grab market share in the early days. That may be through lower price and and depress gross margins. But then expand as as they do become, you know, the winner in that market. And so I do want to pay homage to that because, you know, I think a lot of this, when we're talking about seventy, eighty, eighty-five percent gross margins, doesn't really apply there. But that by no means means that those are not good businesses or, you know, again, generate high return on invested capital.

• 31:45Speaker 00

  Do do marketplace businesses need to think about the same cost centers as more enterprise companies? Do they have the same considerations for their long term margin structure? Or do you see that they have some very different considerations?

• 31:57Speaker 04

  I think for for both marketplaces and and more of these enterprise software and infrastructure companies, there's definitely a couple of shared cost centers. So if you think about, we mentioned customer support being one of them. That's very much a part of both. But overall, I think that the cost centers can often look very different. And your long term margin structure for both types of companies, but especially in these marketplaces, is a product of the defensibility of the moats that you're building in the business. And I think the the note that I'd add about marketplaces is that we know it takes a really long time to build. And oftentimes, scale is really the name of the game. And so marketplaces can underprice for a long time to build this, which weighs very heavily on their gross margins.

• 32:42Speaker 04

  And so what investors will do is is look through. What your current margin is in a marketplace that is being built? And start looking for the indicators that you've built those network effects. Are your cohorts of customers retaining at higher percentages than the earlier ones? Because you've built that density in your marketplace. Are you creating a dynamic that becomes winner take all? And only when you become winner take all, do those margins actually start to trend towards steady state? You know, there are, I think at that scale, you see examples of what sustainable gross margins look like in similar businesses and marketplaces will often get a lot of credit in between getting to that point. If they've been able to show some of these network effects and other signs of moats

• 33:26Speaker 04

  that they are building along the way.

• 33:28Speaker 00

  Yeah, so it's sort of the same, some of the same points you talked about earlier, where there's a bit of show and tell, just like with the enterprise companies, right?

• 33:37Speaker 04

  Absolutely. Because, you know, if you think about what weighs on, I mean, there's several layers to think about in a marketplace. Oftentimes, you know, if you're, if you are aggregating certain end markets, right? Whether that's houses for rent or a labor marketplace, so it's people. I think the first layer of cost is actually what are you paying out to the Airbnb hooks? Or what are you paying out to the laborers? There's obviously cost buckets that are layered under there. But initially, in early days, to build that scale and to differentiate from the incumbents in the market that maybe don't have a marketplace structure, don't have the technology supporting, you know,

• 34:19Speaker 04

  building a broader marketplace, then, you know, you often do see lower gross margins. In fact, we sometimes see negative gross margins for these companies. Make them bad companies? Absolutely not. And how do you get to, at the later stage, that belief of, hey, you can hit a steady state that is sustainable, that will generate cash in the long term? Well, it's really showing these proof points along the way of the moat that you're building to become the winner in the sector.

• 34:46Speaker 00

  Right. Are there any other, like, very practical pieces of advice that you want to make sure founders know and hear and internalize?

• 34:54Martin Casado?

  I really believe that companies go through three stages. It's the product stage, the sales stage, and then the operation stage. And the product stage is you're finding product market fit, and that can take years. And many companies don't actually do it. And then you have to get to a repeatable sale. So let's say you found that you get to a repeatable sale and you focus on sales. That sales expansion, repeatability, et cetera. And then operations is where you're actually working on, like, really focus on margin and economics, because you know now you're later stages and it really matters, kind of like the hygiene of your business, just so, you know, two Sarah's points earlier, you know, like how much cash you have to invest in the business. So that said, if you're in the product phase or the sales phase, I just wouldn't worry too much about gross margins, A, because I think it is kind of unknowable, like you don't really know where things are going to land.

• 35:39Martin Casado?

  It's much, much more important to David George's point to actually get a business that works than to kind of over-optimize it before it works. And so I wouldn't spend too much time obsessing about it. That said, we've talked a lot about these pitfalls that you can run into later on. So be conscious and cognizant of them. Have the discussion with your technical team. Have the discussion with the business team as far as how you're structuring things. And then don't delude yourself. And I think, you know, all that in, you know, you will not trip up by focusing too much of it, but be prepared once you get to maturity.

• 36:11Speaker 00

  Well, thank you all for joining us on the A16Z Podcast.