Last week I chatted with a friend who opened a bar some years ago. It became the hippest place in the neighborhood. What's striking is that most of the customers now visit regularly, even as often as every week. They love the place. There is no real need to acquire new customers actively, the bar is packed most of the time. And the more they go, the more they consume. My friend built a solid small business.
The AI craze
This reminded me of the hype cycles of the last years in tech. Instead of focusing on backing and building great businesses, people often get carried away by these cycles. While 2020-2022 seems to have been all about crypto, the ‘Made by AI’ hype is now in full swing… Massive funding rounds into foundation models (we are talking seed rounds with both, 100M+ funding and valuations...) and one larger acquisition (~600M valuation two years after launch) have happened in H1 already. Yes, there are many reasons to be excited about what’s to come in AI land. Especially the long-awaited productivity boost our economies need to drive growth. But most things take longer than expected... Or, as my partner Mark put it: ‘I doubt we will still be talking about LLMs in 12 months…’ They won’t go away, but the hype will.
I am always skeptical of the hype around trends and sectors. While you can and should attempt to envision the future, you can't time it. Usually, the things that are underrated at any given time are the most disruptive. Some of the best AI companies I know were founded years ago. We have seen this over and over again. Facebook, for example, was founded several years after the first social network. Google was by far not the first search engine... The venture craft is to chase long-term trends and entrepreneurs, not buzzwords...
Second, what's striking about these massive rounds is that it is hard to see how they can be great venture investments. Much (or most) of the injected capital does not go into software development but capex, meaning compute power or chips, to train these models... and this is without a proven nor clear model. Seed stage is when a great team has a good idea, but has not yet proven product market fit and a go-to-market model. The bet is to grind until you have early signs of a scalable and sustainable business model that could be worth a billion+ down the road. Once these signs appear, you can inject larger amounts of capital to grow the business and its value. Such ambitious models that work out are rare to come by. The power-law curve is steep. Therefore, seed-stage investing is about maximizing the few bets that work out (more in this post). To make the model work, investors need to get in at attractive prices with relevant ownership, not buy expensive options that require hundreds of millions of burn in a competitive space... It feels like a quick exit is the only exit here. Inshallah...
Back to basics. Building scalable business models.
I can't get enough of tech innovations. This is what drives our industry. But investing we do in business models leveraging technology, not the other way around...
What makes these models so special? Software companies tend to be cash efficient with high margins, have global reach from day one, great working capital dynamics, and relatively high customer retention compared to other models.
If you consider the example given at the beginning of this post, every business model boils down to customer acquisition, retention, and expansion. Whether it’s the bar down the street, a mobile consumer app, or enterprise software. Software companies do this just more efficiently, predictably, and at a bigger scale than any other model. Let’s dive in.
You might think that revenue is the most important metric of any business. But it isn’t. Gross margins, defined as (Revenue - Cost of Goods Sold) / Revenue, matter more. There are exceptions, of course, especially tech businesses handling massive transaction volumes. They are directly tied to a company’s ability to spend in order to grow and achieve profitability. With each sale, they generate much more cash than a low-margin business. Capital that can be reinvested intro growth.
Software companies provide services at low marginal cost, and thus have very high gross margins. Hosting and customer success costs comprise the majority of Costs of Goods Sold (COGS). Gross margins increase over time and can go up to 80/90%. Costs are amortized over a larger customer base. Finally, as the brand and product offering grows, a company can acquire bigger customers with higher contracts.
'Take two startups, both selling products at $1M price points. The first has 5% gross margins, and the second has 95% gross margins. The first company will be able to spend about $50k per sale on Sales & Marketing, R&D, and G&A. The second company will deploy $950k across those departments. That’s quite an advantage for the second business. The second business can afford to spend orders of magnitude more and, as a result, grow much faster.'
Acquisition Velocity and Reach
Customers that love a consumer or enterprise software product tend to stick around, leading to high customer retention and recurring revenues. Many software companies can charge clients a year, sometimes two, in advance. Combine this with high margins, and you get a business that generates a lot of its earned cash to reinvest in marketing and sales.
Suppose you plan to invest 100 Euros in marketing to acquire 10 new customers in a given year. In that case, software businesses can often invest a multiple of that just by recycling the initially invested capital in a given year a couple of times during the year. Let’s say 300 Euros to acquire 30 new customers, for example. This is the second reason software companies can drive non-linear or exponential growth.
Software companies have relatively high customer retention and expansion compared to other business models.
Beyond impacting revenue growth, retention and expansion also greatly impact underlying profitability. Acquiring a new customer is around 5 to 20x more expensive than retaining an existing customer, depending on the industry. It’s often much more profitable to drive growth from an customer than to find a new one.
'A 20% delta in net dollar retention (NDR) may not seem like much, but... it is massive. Tomasz expressed it well. Imagine three companies: one at 120% NDR, one at 140% NDR and the last at 160% NDR. In five years, assuming all else is equal, how much bigger is the last company vs. the first? The answer is 4.2x bigger… And 1.9x bigger than the company with 120% NDR. Each marginal 20% of NDR is a doubling of company ARR in 5 years. That’s big and the third reason software companies can drive higher growth than others.'
This chart gives you a good view of some best-in class benchmarks of publicly listed software companies. Don't forget these companies operate at a massive scale, making those upselling figures even more impressive.
Ultimately, these are the characteristics of software business models that lead to more predictable and exponential growth. And more predictable revenues and profits mean more valuable business models, especially at scale. Great enterprise businesses can usually upsell their clients, and consumer business can recycle their acquisition budgets more quickly. Some can do both. Financing teams to find early signs of those models is what the typical venture capital model is for.
As discussed end of last year in this post, AI will drive many opportunities by reimagining how people interact with software. Just think about all the large tech businesses driven by traditional search engines... 👀 But the core technologies are already being rapidly developed and adopted across the tech world, and traditional service businesses are already being disrupted by new players leveraging these technologies. I can't wait to discover more exciting business models coming out of this. But it won't happen all at once. Back to basics...
I was in Paris last week for the launch of one of our latest investments. They got a small but great venue for this gathering. This view never gets old…
Life is awesome
European VC Europe