Elite business developers design a system of products and operations that naturally lead to a strong defensible moat. They create flywheels that build strength and create clever positioning to capture durable revenues.
The shrewd business developer fears building in excavated lands. Here, all types of developers wander around trying to find ground to settle and develop. Unfortunately for them, their time is most often wasted on fending off others who are looking to compete for their land. While they may achieve temporary success, the fruits of their hard work and persistence will gradually diminish to low or even zero real margins. Because of this, their work has no real opportunity to compound, and they’re forced to either claw and axe unsustainably, or leave.
How Software Changed Product Defensibility
Traditionally defensible technological products relied on high upfront capital expense which would be used to fund technologically complex and patentable inventions and processes. The necessary capital limited the number of new ventures that could start up and compete, and the patents created reassurance that the invested labor and capital would bring durable financial upside.
Software changed this paradigm. Software products had attributes that prevented business developers from relying on these status quo approaches. They had
Zero marginal cost of production
No clear path to intellectual property
Limited reliance on upfront capital
Because of these traits, defensibility evolved to take on new forms. When analyzing this evolution, it’s useful to recall Hamilton Helmer’s framework around the 7 Powers of Business Strategy. In his book, he discusses seven main sources of business moats. They are
Economies of Scale
Network Effects
Switching Costs
Access to a Cornered Resource
Brand Power
Counter Positioning
Process Power
Businesses that create products that have non-zero marginal cost can benefit from economies of scale. Capital intensive businesses create a natural moat by using the cornered resource of money. Meanwhile, intellectual property can both a cornered resource and also process power.
Modern Defensibility
Consumer Apps
With zero marginal costs and no path to intellectual property, profit margins will naturally approach zero. This is because once product value has been demonstrated for a market, there is no stopping countless other developers from building the same experience but cheaper, thereby competing out any marginal profits.
Instead, the most powerful software companies leveraged new paradigms to build business moats. Specifically, consumer social apps extracted value from network effects - that is, their product value increased as their number of users on the product increased. The experience of using the product was driven by the number of other people you knew that were also using the app, and this created a much higher barrier of entry for competitors. Because the network economy requires a high volume of product users, costs still were near zero for the end user, and ultimately, companies were forced to leverage this network for advertising. Winning businesses attracted usage to build a network and efficiently extracted monetary value from that network with advertisers.
Consumer apps that didn’t leverage this defensibility principle were forced to fight amongst cutthroat competition. We can see this as the number of new apps exploded in the early 2010s but then dropped and has since plateaued. Competition in the mobile space is brutal, unless you’re building a product that can defend itself with some level of a network economy.
Saas
After the dust settled in the consumer app landscape, SaaS took its reign as the new lucrative software enterprise.
The strongest SaaS companies leveraged switching costs to create defensibility. They did this by building seamless integrations into existing team workflows. They focused on sales and marketing to attract customer attention and high touch onboarding. This led to a high customer acquisition cost which required more upfront capital and helped fend off customers. It also helps explain why venture funding became so crucial to success in this era.
But once integrated into an office culture, their product became quite sticky. The enterprise customer avoids frequent change. And because these products became the backdrop of employee productivity, they are very hard to switch out. Therefore, companies justified the high customer acquisition cost with equally high lifetime value.
This led to the birth of successful SaaS companies such as Stripe, Slack, Notion, FLexport, Brex, amongst many others. Ultimately though, the power created in these switching costs never quite measured up to that of the network economies in consumer apps, and so the market capitalization of these successful companies never reached those of the consumer app boom.
The New Defensibility Paradigm
A new disruptive technology has emerged, the open AI model.
New generative AI models, such as GPT-3, stable diffusion, etc, have democratized AI like never before. Just like the App Store democratized consumer software and the cloud democratized enterprise software - this new technology will change the business landscape for the next decade.
The companies that were best able to profit form the previous cycles were the ones that managed to best leverage a powerful complementary defensibility principle along with the disruptive technology. So what is this principle for the open AI era?
New publicly available models will launch a new wave of “knowledge apps” - apps built to provide AI experiences to consumer and work use cases. However, these models are trained on general, publicly available data and therefore, will be great for general use cases. This is fantastic for getting PR and for fascinating the general public.
But domain specific use cases will require more than just general models. Customers don’t care about best case outcomes, they often care about worst case outcomes. Companies looking to build durable businesses around this technology will need to master the tuning of these models for customer specific problems. That means collecting domain specific data and enriching models so that worst case output meets a high standard. This will be difficult to do, and that’s why it will be the important factor in creating lasting business value.
Startups will need to collect and curate this data to create their own cornered resource - which will be their significant and lasting competitive advantage. They will engage with users early on and develop model customizations that will delight customers. When these customers look for alternate products, they will be disappointed because the best product experiences will be powered by those companies that have managed to build the best curated datasets.
How is this new? Well prior to open models, most AI experience innovation came from existing tech incumbents because they were the only ones with sufficient data. They had created a vastly defensible cornered resource of user data. They used this to power discovery experiences, recommendation systems, auto complete, etc.
Open AI has leveled the playing field, not my making the data open but going one step forward and democratizing the model itself. But this doesn’t mean that data is now irrelevant - in fact it’s quite the opposite. This innovation represents a dramatic reduction in the activation energy needed to create rich knowledge applications - just like cloud reduced the cost of making web and mobile applications. With a new base case set up, the startups that are the first to create curated datasets to layer on top of this new infrastructure will be the ones that make the best product experiences.
This cornered resource paradigm is the best match for open AI technology and is the critical factor that will differentiate products that build broad and deep moats around their business and those that will be left to fight endlessly in war-torn terrain.