Market Sizing: Numerical Narratives
It is easy to see why Chris' and Fred's "narratives over numbers" market sizing might be appealing. After all, as Brad Feld writes, "Almost every market sizing presentation is incorrect - by a lot. Enough to make it irrelevant." Chris writes, "you should never rely on quantitative analysis to estimate market size. Venture-style startups are bets on broad, secular trends."
This seems to make sense: tell a story because your numbers are going to be wrong anyway. But let's analyze the traditional market definitions to see why traditional quantitative methods are incorrect. Non-quantitative analysis (a story) is not enough. Market sizing has to be quantitative because it is a tool to make an investment (i.e. a quantitative) decision.
What is a market? The traditional definition of an "addressable market" is the problem. VCs and entrepreneurs use either technologies (e.g. search or semiconductors), products (e.g. insurance or printers), or users (e.g. app developers or gamers) to define markets. But these definitions are flawed because this is not they way customers think about markets. This flaw was recognized back in the 1960s by Theodore Levitt who famously noted that industries define themselves by the solutions (railroads, movies, oil), but they should be defined by customer needs (transportation, entertainment, energy).
In contemporary language, markets should be defined by the jobs customers are trying to get done (e.g. store music), not by the solution (CDs, MP3 players). This makes telling a market size "story" much easier because jobs have a beginning, a middle, and an end, just like a narrative. And jobs are independent of any solution idea you may have. With the job as the unit of analysis, the market can be quantified with precision. We call the resulting number the "securable market" to distinguish it from traditional definitions. The inputs into the market size calculation are critical, and, surprisingly, they have nothing to do with the product or solution.
So, yes, market sizing should be based on a narrative, but it should be a numerical narrative. Otherwise, like traditional market sizing techniques, it is just a guess.