Maybe Thesis - v3

Maybe Thesis - v3
Created with Midjourney - a bright and cheerful optimistic sci fi scene of 2 businessmen in tokyo 200 years from now looking up at the empire and worlds they built. pixar.

Spotify Amazon Apple



We're supposed to be buying boring SaaS businesses that cash flow slowly and consistently. On the small end of the market, that's a little challenging to do. Often, the way that plays out is buying a small B2B enterprise SaaS company that does something pretty "boring." WorkClout was a great example of this. It sold to companies manufacturing things. Not even cool e-commerce things (mostly). When people first hear of our model and the term private equity, this is what they think we do. And they're not wrong, but they're also not quite right.

Keep in mind, I'm mostly talking about the stage of life XO Capital is in. We are still relatively young. After 7 acquisitions and looking to do our 8th here pretty soon, some patterns are starting to emerge. We have, of course, bought boring businesses like WorkClout, but the reality is we've purchased more product-led growth companies that are closer to a prosumer category than enterprise. Enterprise comes with particular strings attached, namely the need to handhold customers while they navigate purchasing and training employees on how to use the software. This process can take months, sometimes up to 12 months. And what you're left with at the end of the day is relative customer concentration, which, for better or for worse, tends to be discounted by the market of buyers and sellers on However, when you have a smaller company with a more diverse customer base and a larger number of customers, that tends to be rewarded by the marketplace and ascribed a higher multiple. This is somewhat anecdotal, but this is what we've seen.

Outside of optimizing for multiples on let's talk a little bit about Alpha.

Hi / Lo

Alpha (α) is a term used in investing to describe an investment strategy's ability to beat the market, or its "edge." Alpha is thus also often referred to as “excess return” or the “abnormal rate of return” in relation to a benchmark, when adjusted for risk. - investopedia

I would describe the playbook of buying "boring" businesses as relatively low alpha. The expected outcome is slow growth but consistent growth and predictable cash flows. However, having done this for a few years now, the reality is, especially in enterprise SaaS, the revenue is lumpy, and when you have only a few customers, you tend to lose sleep at night praying to whatever God you believe in that one of them doesn't churn.

This is a radically different mentality than a PLG company with 100 plus customers. You tend not to have relationships with very many of them. You tend to only hear from them when there's something bad going on, and you really have to measure usage deliberately to get a clear sense of what's going on in the business. Of course, looking at top-level numbers like recurring revenue or monthly cash flows helps to give you a sense, but it will not let you know when a customer might be churning, nor will it indicate buyer intent.

After 3 years of doing this, I overestimated how well we could execute on the enterprise SaaS playbook. We're a bootstrap business, and we have to rely on shared services like engineering or customer support to make margins. This infrastructure falls apart when we have a large enterprise SaaS company with fewer customers that demand more of our time. The surface area of the enterprise SaaS products tends to be much larger than the product-led growth companies, causing us to have to dedicate full-time resources to building just that one app. These larger customers also demand features in a different way than product-led growth companies are asked to build features. The implication on the enterprise SaaS is that you will build what they need, and sometimes those needs are company-specific. When you have a larger enterprise SaaS company, this isn't as much of a problem. It's called non-recurring engineering, and you charge a fee for any engineering work that you have to build for an individual customer that will not translate to other customers. Sometimes, particularly in the early stage of a company, you can charge NRE fees that help subsidize building the rest of the product. Getting an enterprise to sign up for something like this is very difficult without a pre-existing relationship. In fact, this is one of the major benefits of VCs, they often do have these pre-existing relationships and can radically accelerate progress.

But I digress. Enterprise SaaS is hard for us to execute on with shared services. If we had more revenue to hire dedicated resources and staff up the company to run as a standalone entity, many of these headaches would go away. But alas, we are not there yet.

And frankly, there just aren't that many of them that we see coming onto marketplaces like If I had more of a war chest, I would go build relationships with VC companies so that in a year or two when the portfolio starts to mature, we can pick off the distressed assets and try to do our thing. We will get there one day, but until then, we may need a different strategy.

That different strategy looks like a higher alpha strategy where we buy something that just started working. That might mean they have a handful of customers or it might mean that they have zero paying customers. These things are generally under $10,000 MRR, are run by one or two people, and have been bootstrapped. The products themselves tend to be small and single-purpose. They tend to have a single promise to the customer. By definition, these products are easier to execute on for us. Our best buys have been when we purchase something nearly complete. It's okay if there are some bugs or some obvious things that we need to fix, but we should not have to rewrite it from scratch, nor should we be having to deep-dive to figure out what to build next. There's enough traction that customers will start to let us know what they need, and we can be slightly more reactive rather than proactive. These are typically negative associations, but for us, they are paramount to running XO with shared services.

So the first part of this new thesis is that we are saying no more often to "boring" B2B enterprise SaaS companies and saying yes to product-led growth B2B SaaS companies.

If you put me in a corner and ask what we would need to build over the next 12 months to be successful, I would say we should build sales tools. They're easy to sell. They (more or less) directly generate revenue. Founders, CEOs, and salespeople are open to trying new things. Churn tends to be slightly higher because this group tends to be easy come, easy go. We're doing this. We're going to be launching our first product we've built from scratch pretty soon. It's a sales tool. And yes, we built it from scratch. And yes, of course, this may confuse you. XO's charter is to buy SaaS companies. However, I also want to win and don't care how we get there. This is one benefit of not having a fund or any investors. I can wake up one day and decide to build something I want to build and use all the great infrastructure we have at XO to launch it. I'm not saying this will be a new pattern we follow, but it is a fun experiment, and if we are ever to become a fund or ever to raise capital and I lose that ability to do these experiments, then how nice that we get to run them now and have data on it.

Higher alpha will come from taking more risk earlier on in a company's life cycle. We can buy PLG companies with few customers relatively cheaply and, in some cases, cheaper than we'd be able to build it ourselves. This implies we also have the freedom to build stuff ourselves from scratch. This new sales tool we built, I think, will have been around 2 to 3 weeks of development time all in. I built it myself, so it cost us $0. With respect to alpha, this is also an area where we can excel. I could probably build one of these a month without impacting our ability to simultaneously purchase more companies. But, of course, as you well know, building it is not the same thing as having a successful company. Thankfully, we've been starting to run more diligent experiments around growth and documenting them well.

Notice everything above has almost nothing to do with innovation. We're not talking about landing rockets on Mars. We're not looking for a thousand X return with these bets. We're trying to get to profitability and breakeven. We're trying to make sure that XO is unkillable. We're trying to generate excess cash flow to be able to buy or build more businesses or continue to grow the existing portfolio.

Something Danny pointed out several months ago was that most of Constellation Software's growth was through acquisitions. Growth by acquisition is growth. From our vantage point, it has been easier to acquire an additional $10,000 of recurring revenue than it is to grow the existing portfolio by that equivalent amount. Ironically, despite all the negative things I've said about B2B enterprise SaaS, that is one component that was nice. One enterprise customer is generally 5 to 6 figures a year. But again, without the ability to staff that enterprise SaaS company, it ended up eating most of Danny's time, and limiting our ability to focus on the other portfolio companies.

If I had to take this thinking to level 10, I would say that actually we should be making many bets on emerging technology. Despite my recent cease and desist from LinkedIn and run-in with platform risk, that technology is large language models like ChatGPT or the GPT API. This, I believe, would be the case where we are optimistically hoping for a 100x return. Ideally, we would make around 10 bets in various verticals using this technology, and let it play out over a year, either killing or selling off the losers and doubling down on the winners. At this extreme, I care even less about whether we build it or buy it. Emerging technologies create a land grab opportunity. We're in that period now. Billions of dollars of value are going to be created in the next several years due to this technology. I am fully bought in that this will materially increase worker productivity in countless industries. If this starts to sound like a venture thesis, you're absolutely right. The only difference is we will be buying footholds in these markets and sometimes, when it makes sense, building them.

Looking back on some of the older thinking I had in this area, I questioned whether or not our model was subject to power laws. Despite us not seeing them yet in our own portfolio, I believe that we are. I believe that at some point, our revenue distribution between companies will change. Currently, on a relative basis, each of the portfolio companies generates an eerily similar amount of cash each month. Over time, as we increase the number of bets, I believe we should have a small number of outsized winners and a large number of mediocre performers. What I don't know is what number of bets we have to make for us to start seeing that. We've made seven bets and haven't seen it. Will we see it at 15? 30? I don't know yet, but I believe we will see it at some point.

Bringing this back down to reality and into something that we can execute on this coming year, I think that simply looks like continuing to be opportunistic with the opportunities that come across our desk, but with a bias towards making bets on products powered by large language models or that in some way make use of generative AI.