Happy St Patrick’s Day, everyone. Today, we’re going to talk about proof, and two different kinds of proof that startups tend to face yet not fully appreciate. I’ll be discussing two types of startup models – which I’ll call Type 1 Behaviour and Type 2 Behaviour models – and why they bear distinct burdens of proof that are quite different from one another.
A Type 1 behaviour model centers on consumer behaviour that already exists. Companies following this model build a product that makes this behaviour easier, better, cheaper, faster, or improved in some way. They then monetize the difference by selling to whomever it is that’s willing to pay money for this behaviour to be improved. Sometimes it’s the consumer directly (Get a newer faster phone, get a better experience on whatever Apps you’re already using); sometimes it’s another company who makes money off of said behaviour (a lot of M&A in the consumer space is acquisitions like this; think of Facebook buying Instagram, for instance).
A Type 2 behaviour model, on the other hand, is build around consumer behaviour that does not yet exist. Companies following this model build a product that allows this new behaviour to emerge; they then monetize that new behaviour. Many of the giant young companies of the 21st century follow this model: Airbnb, Twitter, and Snapchat are all good examples.
There are clearly many differences between these two models, but the distinction I’d like to focus on in this post is between the different burdens of proof that young companies face when presenting and validating their ideas. Type 2 companies, although they typically have a more difficult challenge ahead of them, actually have a more straightforward task here; we’ll call it Type 2 proof. For a business model built around behaviour that does not yet exist, they need to demonstrate: 1) that given the opportunity, consumers will exhibit this behaviour, and 2) that the behaviour is sticky enough to build a business around. These are certainly challenging things to establish, but at these companies have their work cut out for them: Type 2 proof requires gathering evidence for a positive: “Even though this behaviour does not yet exist, it will emerge under these conditions.”
Type 1 companies, on the other hand, have a subtler task. For a business model built around behaviour that already exists, they need to demonstrate Type 1 proof: 1) Is the Delta large enough to build a business around, and 2) In the longer term, will this behaviour continue to exist at its current level of prominence? That second question actually very tricky to answer: you have to gather evidence that some currently held assumptions will not change. Part of the reason that’s so difficult is because humans tend to naturally hold these assumptions by default, without any actual evidence for it to be so. “People engage in behaviour X all the time; they won’t just all decide to stop all at once!” is very easy thinking, but it’s not how the world works: Behaviour X usually isn’t deliberately abandoned en masse, it’s simply eclipsed by Behaviours Y and Z and loses prominence. Type 1 proof is harder; it requires gathering evidence for a negative: “Even though the world will be different in five years, this behaviour will not have been eclipsed, and out assumptions will not have changed.”
As an illustrative example, I’m going to pick on a company I know well – Algolux, a Montreal-based startup that overlapped with Backtrack at TandemLaunch. (I like the Algolux guys a lot, and think they have a very promising venture – which is precisely why I’m going to give them a hard time.) Algolux is going after a classic first order behaviour problem: people take lots of photos with their smartphones, and have come to expect very high image quality from their phones. However, as phones get thinner it becomes increasingly difficult to fit the necessary optical components into a device. To solve this problem, Algolux builds digital optics solutions that enable better quality pictures from cheaper, thinner cameras. It’s a classic play to improve the experience of an existing behaviour (people taking photos), and monetize the Delta (in this case, through OEM licensing deals.)
Algolux faces the two burdens of Type 1 Proof here: is the Delta large enough, and will the requisite consumer behaviour persist? I won’t address the former here, as I’m hardly an expert in this area (although their demos look very impressive). But the latter, in Algolux’s case, reveals a hidden danger: what if, in five years, we look back and realize in hindsight that ‘Peak Photo’ actually occurred in 2014? If this seems ridiculous, keep in mind that although it’s unlikely that everyone will collectively decide to take fewer pictures, it’s quite possible that as time goes on a lot more of our time and mindshare will be taken up by other things like video and messaging. What if our social landscape five years from now contains a lot more WeChat and Meerkat, and photo sharing takes up less of our attention? OEMs, facing steeper and steeper competition for lowest prices, will only have a limited amount of budget to work with – and may well choose to prioritize faster streaming video over better photo quality. And here’s the thing- there’s really no way to conclusively show one way or the other if this will come true. It’s easy to state an assumption like ‘People take a lot of photos with their phones, and demand high picture quality’, and project that assumption as valid going forward for many years. And most people will agree with you, because it sounds very sensible. But it’s very difficult to establish a confidence interval around that statement: we can make a pretty safe bet that expectations on picture quality from our phones won’t go down, but can we make the same bet on whether or not taking those pictures will occupy the same amount of mindshare? Hard to say; that’s what makes Type 1 proof such a really tricky thing.
It is more straightforward to collect evidence that something not yet true will become so (Type 2 proof) than it is to collect evidence that something currently true will continue to be so (Type 1 proof). For anyone trying to predict the future – i.e. anyone in the startup ecosystem, whether they be founders, VCs, or any other speculators – it may be useful to keep this in mind.