SaaS: The Lazy Industry
Big SaaS companies inevitably optimise for increasing cash flow over product improvements. Why are they so hard to compete with?
This post is a bit longer than usual. I hope you find it interesting! If you enjoyed it, be sure to subscribe. And if you’re already subscribed, I love you, forward it to someone who isn’t.
No other industry that I know of enjoys gross margins as good as SaaS. It's like someone came up with the perfect business model for creating big moats. If you're making 75%+ gross margin1, you have so much excess cash to invest in product development that you can create new products and more powerful lock-in effects all the time, right?
In practice, not really. Most SaaS companies spend more on sales & marketing than on product development. A lot more. An audit of public SaaS companies shows that on average almost twice as much is spent on Sales & Marketing (S&M) as on Product Development (R&D), while General Administration (GA) spend is almost equal to R&D2. In other words, over three quarters of operating expenses are not spent on product development. And even that is a broad term; it covers deciding what to build (product management), fixing bugs, tweaking the user interface, minor feature improvements, quality assurance (testing), and a long tail of other tasks. Typically only a small percentage of “R&D” spend goes to what R&D is meant to mean: designing and building new products that create new value.
Because the gross margins are so good, companies have less incentive to innovate.3 GA and S&M are much easier areas to spend money in, and they more directly contribute to revenue growth. As companies mature, it becomes easier and easier to get a junior salesperson doing something cash flow accretive. Sales scripts get better and product quality gets higher, which enable the new rep to hit the ground running. By contrast, it gets harder over time to productively onboard junior developers. This is because the product gets more complex and the ideas for improvement get more niche, both of which make it harder for a new developer to have an impact.
So when companies have a marginal extra dollar to spend, it’s hard to argue against spending it on sales & marketing. And it gets harder the longer the company’s been around.
Almost every company eventually succumbs to this thinking. It takes a very sophisticated R&D organisation to overcome this and continuously prove its investment-worthiness. What mostly happens is that as the product gets more complex and the deals get more lucrative, companies don’t stay product focused. This is bad for customers, because the company inevitably ends up optimizing for how to win the next deal instead of for how to create new wealth and value for customers. When taken to the extremes, it’s common for a company to reach product-market fit and for the founders to never talk to a customer again.
Businesspeople should be shot. They always say, “We are in business to make money.” And I say, “Well, not really, you just want to make a few million or billion.” But the return from PARC is about thirty-five trillion! Count those extra zeros and tell me what they are really doing. They are just trying to be comfortable.
Unfortunately, the SaaS model makes being comfortable look like the right decision. Another word for being comfortable is that the company gets lazy.
The excessive cost of software production
In The Innovator’s Dilemma (or any other business book that clones it), Christensen argues that customer requirements change constantly so there is always an angle for building wedge products that do a lot less than incumbents but are more desirable. His research is based on computer hard drives, but he extrapolates it out to other industries that don’t move as fast.
Given how fast SaaS moves, I’d expect to see incumbents getting replaced constantly by new innovators. But it doesn’t happen that much. Clearly there is some sort of moat that we aren’t seeing. What’s going on?
I struggled to articulate an answer until I read this post:
Kedrosky and Norlin argue that the cost of writing software is artificially high, and always rising. First of all, I think this argument has a lot of merit. After all, the status quo is confusing:
(Software-based) technology is the most profitable industry.
Software development is the fastest growing university degree.
There’s no official credential that makes you a software developer. Thus, there is a strong non-university industry for learning to be a developer, as well as a thriving ecosystem of self-taught developers. Statements like “anyone can be a developer” are exaggerated, but not by much. Compare this to doctors, lawyers, engineers, etc., which have a powerful (but justifiable) gatekeeping mechanism that decreases supply and drives up wages.
The skills you learn as a software developer are 100% transferable from city to city, state to state, or country to country. This means the talent pool is the closest to global of any industry. This should drive supply up and wages down. Compare with a lawyer who must learn new laws, or a doctor who must learn new medicines and maybe a new language, or many fields where relocating for work isn’t financially viable.
The only real cost in developing software is wages. It’s infinitely scalable and there aren’t any raw materials.
Despite all this, the cost of producing software has never gone down. Why? The authors blame a lack of productivity increases, and point to AI and LLMs as a big catalyst for change. (They also blame a concept called Baumol’s Cost Disease; read the article if you’re interested in the economics angle.)
Why do we care? Well, if the cost to produce software always goes up, this answers the question of why B2B moats are so strong even as companies get lazy: the cost to compete is just too high! Even if you know that B2B Company X has stopped talking to their customers, moved most their spending to sales and marketing, and is coasting along, it’s a big investment just to copy their product. How big? Well…
Incumbents have tremendous depth
Let’s say you want to compete with my company, Workforce.com.5 What do you need to build?
You’d need to build a good enough replacement for all of our core features. On an 11 year old monolithic codebase, that’s a lot of features. Not every feature gets equal use, but we’ve also spent 11 years learning exactly what job to be done customers really care about, and making those features extremely good and very comprehensive.
So you need to build a broad set of features reasonably well, and you need to figure out which of those features are really important, and you need to build those features extremely thoroughly.
Then we get to the invisible stuff. We integrate with tens of other software vendors. Some of these integrations are for other products that our customers also use, some are for behind-the-scenes systems. Not all the integrations are publicly documented. You’ll need all of them.
We are GDPR compliant and audited for SOC2. It takes years to get SOC2, and our customers will ask you for it. We have infrastructure in multiple continents for scalability and security.
There’s a lot of software you need to write, just to get to the starting line. Once you’re there, customers will still just pick us because we’re more established. So you need to build differentiators too. But differentiators are extremely hard to get right, particularly if you don’t have many customers. How will you know that your novel idea actually is 10x better than the status quo?
And here’s the worst bit: you have to do all of this while paying more for software developer talent than we did 10 years before you. Developer wages are rising much faster than inflation, so the difference between what you pay and what you can charge is smaller than it was for us.
It’s like climbing up the cliff with a backpack full of weights. Even if we, at the top, have stopped climbing, you still might never catch us.
Competing with lazy companies
AI believers suggest that the cost to write software will plummet in the near future thanks to advances in AI. Perhaps. But writing code isn’t all a business does. If the cost of writing software plummets, the rewards will go to companies that most quickly figure out how to take advantage of that technological wave to ship better products more quickly.
Anyway, that’s in the future. You can get lots of the upside of lower software production costs today, by just choosing to do it more cheaply. For example, we built our product on Ruby on Rails, and we didn’t need to hire a second engineer until we were nearly 3 years old and had lots of customers. By the time we hired another engineer, we had built many of the core features that all the incumbents had. And we did this with no funding at all!
Compare this to someone who chooses a different stack, and needs to hire 10s of engineers before any customer spends money, just to get to a point of parity. They’d need to raise millions of dollars. Investors don’t like investing millions of dollars on such unproven and unexciting bets. So the competing product - which may have one day been much better for customers - might not get built, because it requires someone with deep pockets just to get started.
If you are in the mindset of getting to market very cheaply, then you will resist the temptation to hire too early, and to say yes to feature ideas that you shouldn’t build. So not only will it cost less, you will also get to the starting line faster.
In short: you can disrupt what seem like an impenetrable B2B moats, by rejecting the high costs hypothesis. Just because the cost of software development is high across the industry, doesn’t mean it has to be within your company. Other companies might be climbing cliffs with weights in their backpacks, but that doesn’t mean you have to too.
Avoiding becoming lazy
Despite everything I’ve said so far, I don’t think it’s inevitable that companies become lazy. Despite the insane profitability and business model pressure, it’s possible to keep pushing forward if you really try.
A good starting point is to be disciplined with every expense. The cost of software development only skyrocketed because everyone got drunk on hiring, but most companies would move faster by keeping their teams smaller. This way, each extra hire can make a real impact. This is even more important when lean upstarts are trying to clone your product and compete: a smaller team lets you move faster (and compete more), and a lower cost base makes you harder to beat because you have more flexibility over prices and expenses.
Beyond that, the big challenge becomes splitting focus. It’s very hard to focus on continuing to sell and develop the current product for the current customer base, as well as taking big new bets on big new things that significantly change the product or allow you to enter new markets.
I used to think that you could split this focus by having different teams own different focuses. The output of this was “R&D Teams”, “ Research Labs”, and “Innovation Hours”. None of those worked very well. Inevitably they would get scattered across too many different ideas - they’d start lots, but finish little.
These days I think the right path is to trust the instincts of founders and those who’ve been around for a long time and really know the customer. Rather than trying to create a structure where you can experiment with lots of ideas, just pick 1 that you’re really confident about, and then go build that.
1 person who knows the market very well + 1 person who knows the codebase very well can build a very well done new product in 3 months. Then you can spend the next 3 months with a bigger team polishing, shipping, and iterating based on feedback. This keeps the whole team on their toes. If you do that twice a year, it might feel slow while you’re doing it, but if you take a step back you’ll see it’s a ridiculous pace of innovation.6
If we are less lazy, what happens next?
If you have an incumbent you want to topple, my hope is reading this encourages you to do it faster and leaner, by focusing on better tools and decisions to do things cheaper.7
And if you are an incumbent in a competitive market, I hope you consider how you can be more nimble so you can get better use of your R&D spend and invent new things for your customers.
All these outcomes are good for customers.
Thanks to James, Chris, Joel, and Jill for reading drafts and giving feedback.
As a side note, the original inspiration from this essay came from reading Paper Belt on Fire, which is an excellent book written by the guy who ran the Thiel Fellowships, and then started a VC fund based on the same thesis. The initial goal of the fellowships was to "short the education industry".
Thiel's insight was that universities are overly bureaucratic and not interested in providing quality education, advancing knowledge, or seeding innovation. Instead, they are interested in advancing agendas and collecting donations. They can do this because they are guaranteed an income stream that is not dependent on their educational product being good. Even if lots of individuals inside universities want to advance knowledge, the business model incentives don’t drive that behaviour. His concern was that this results in a lack of innovation, and a stagnating quality of life for all.
A similar argument can be made about the B2B SaaS industry. Most B2B SaaS companies are not interested in providing quality products, making customers more efficient, or enabling new innovation. Instead, they are interested in scaling sales and marketing efforts and raising prices to maximize cash flow. They can do this because they are guaranteed an income stream in the form of MRR, where customers buy the product again each month even if it hasn't changed. Even if people inside the company want to build great products, the business model doesn’t always drive that behaviour.
(Of course there are exceptions to this rule, just like how not all universities are the same.)
Source: https://archive.org/stream/HistoryOfTheoriesAndIdeologiesThatGotUsInTheTurmoil/%5BAdam_Fisher%5D_Valley_of_Genius-_The_Uncensored_History_Of_Silicon_Valley_djvu.txt
If you haven’t read Valley of Genius, it’s much better than this blog.
Nothing specific to us. You could write the next few paragraphs about many SaaS companies.
For a great example of a company returning to innovation when threatened, look at Microsoft under Satya Nadella.
Unless you want to make workforce management software, in which case I suggest raising lots of money and wasting it.