The fraud of product-market fit and addressable markets
Product-market fit has always been a slightly weird concept to me. In my experience the only time you ever get a good fit is when you’re copying someone else (and by definition a budget which already exists).
Usually (and for quite a long time) product market fit means that your customers have got just enough value from your product or service that they’re willing to tolerate the bugs, inefficiencies and brokenness of the other 75% of it.
Peter Levin and Martin Casado put it so brilliantly that I almost got emotional:
Most companies don’t find a magical product that just happens to address a key pain point in the market, thus making their go-to-market (GTM) plan a straightforward one. Rather, they engage in a multi-year-long battle of hammering the shit out of the company and the market simultaneously, trying to get the two to hold together.
An important part of this battle is convincing a selection of companies that they are in fact a category. A label. A cohort. An addressable market.
It’s pretty amazing just how poor the addressable market data is in some sectors (try getting hard data on how much non-UA money is spent advertising to games audiences). I have a loose theory that the more VC-driven sectors have the least resilient market data because the funds can’t afford to diligence it properly. PE firms will spend $1m on a transaction including all sorts of outsourced research. VCs will call people they went to business school with1.
An amusing historical fact is that most of the digital kids ad market sizing which floats around today in pitch decks, earnings reports and all sorts of company presentations actually originated as an investor due diligence report which a PE fund commissioned on SuperAwesome (it was so good we released it as an industry report). But it’s amazing how much that same data gets misattributed.
This is not just true for investor-facing market data. At a certain scale, companies have to move from being product and sales engines to being customer understanding engines. For example, companies which monetize from advertising are fundamentally in the share of wallet business. Early on you can build your projections based on bottoms-up sales team growth targets. But at a certain scale the issue becomes not whether you can secure another 10% of the budget but whether there is any more budget there to secure.
Reading
By far my most abused term is ‘ecosystem’. Although it is slightly slow going at the beginning, a journey through Charles Mann’s excellent 1493 is a worthwhile investment of your time. He describes the Columbian Exchange, essentially the merging of various ecologies across the Americas and Europe after Columbus re-established trading lanes in 1493. While reading it, I was constantly being hammered by thoughts of how the merging of AI and other sectors will play out over the next few decades. What will be the technology/software equivalent of oranges growing in Florida today?
The timeline of human history is being revised almost weekly through a combination of ancient DNA analysis, underwater archaeology, LIDAR analysis and occasionally ancient scroll ML competitions. Many of these outcomes require a displacing of well-seated theories which have either a large volume of evidence behind them or a great number of people who have quoted that evidence over time. A lot of high quality heresy is about reversing social proof and, even if it’s not entirely correct, making people think a bit more deeply about their assumptions. Depending on your interpretation of the evidence, Gunung Padang either is or isn’t a 25,000 year-old man-made pyramid in Indonesia. The authors of a (peer-reviewed) paper on the topic released all the behind-the-scenes correspondence of what happened when their paper got subsequently retracted.
Wardley mapping: a business strategy mapping technique which is helpful to use when thinking about component invention versus commoditisation (and maybe other things too, I’m still working on it)
At some point a universal age API will be a real thing. In the meantime Meta has been investing in its own user age group API so that app developers can sort their offerings into different age groups. Worth reading.
This is not me being mean about VCs! Assuming 2% management fees (at least in the investing period), a $75m seed fund will have ~$1.5m/yr to spend on salaries and transactions whereas a $1b PE fund will have ~$20m/yr for probably the same number of deals. Business school networks are excellent life leverage.