You Can Never Size a Market in Excel
medium.com
medium.com
Agree: Early-stage investing is rarely about accurately sizing a market. And nothing can be boiled satisfactorily down to an Excel model.
Disagree: But you still have to do it. For yourselves, if not for VCs. Presumably, VCs want to see that you've thought long and hard about the market opportunity, about the competition, and about your marketing plan and business model. And you should have. Your market sizing will never be pitch-perfect, but you can't just wave your hands in an investor meeting and declare that the entire exercise is bunk.
I would argue that, if you're being asked to "dig deeper" into market sizing, or even if you're being questioned at all about the market size in your meetings, it's really a proxy question for something else. Maybe they find the idea narrow, and they want to be convinced it can broaden to other segments. Maybe they think you're being insufficiently ambitious. Maybe it's the opposite; they think you're being wildly naive. Whatever the case, "What's the size of this market?" is a question their analysts can probably guesstimate in a few days' worth of Excel monkeying. That's not really what they're getting at. What they really want to know is how you think about the business side of your startup (or indeed, whether you do).
But the process of developing the model, working out what the factors are that will impact market size (and your share of it), working out what things you need to monitor more closely or focus on for biggest impact is where the value is.
If your model is logical and flexible, you can work with investors to adjust it and play around with different scenarios. If your model depends on an inflexible degree of supposed precision, you're at high risk of deluding yourself. Savvy investors can probably sniff that out.
All things being equal, it's probably best to present a model that is simple enough to be easily grokked; flexible enough to be adjusted on the proverbial back of an envelope; and dependent on a set of factors that you've thoroughly researched and confidently believe are the factors to focus on. So you steer the conversation to the factors, and less to the outcome of the model under your presented scenario. If you've got the right factors, you can model out any number of scenarios. (And you should probably have done this already prior to the pitch, so that you're prepared to speak to a given range of hypothetical outcomes).
http://www.realdanlyons.com/blog/2011/10/11/a-conversation-w...
He said if the market size is not immediately obvious then you can't figure out market size.
It's not the same thing.
In fact, it is a bit stronger than that: he states that the only knowledge that can be used is instinctual knowledge, but since instincts concerns things like what smells indicate spoiled food and legless that animals are dangerous, you don't have any relevant knowledge at all! The best you can do is baseless guessing, though after having gotten funding and done a bit of work you might have some basis for your estimate (there is a caveat that instincts dominate estimates early in the start up process).
The summary ars provides does seem to be what he strove to say, but since two people can easily disagree on what is to be considered "obvious" even that statement allows for less predictive power that seems reasonable.
His response was that in the early days of the company they had tried doing things without plans, and that bad plans were much better than no plans.
I learned something that day.
"No plans" means exactly "by chance". Bad plans can do considerably worse than chance.
What organizations need to learn how to do (and unfortunately to this day Excel isn't particularly enabling) is sensitivity analysis.
One of my corporate parlor tricks is to render valuation spreadsheets (which are typically cashflows dependent on scenario parameters) into Matlab and giving each unknown parameter a distribution from three-point estimates (min/most likely/max). The difference between good plans and bad plans is the probability that your best valuation exceeds your WACC or the turnkey price to be paid or some such method.
People are often impressed that the difference between good and bad plans even exists.
What's more: Octave (free as in Stallman) is slow but runs Matlab code transparently.
then, what actually happens happens, and you can adjust hypotheses. and you learn from that, and adjust your own guessing for future.
The thought exercise will reveal weaknesses in your strategy.
To quote Robert De Niro from Ronin; "If there's doubt then there is no doubt."
Precision is a moving window in an unknown universe.
Although it doesn't ruin his short essay, "instinct" is unfortunate. Executive knowledge about the kind of nascent industry the article (and HN typically) discusses isn't "natural", or a mere knack like playing music by ear. It's tacit knowledge: it can't be codified.
In Heidegger-speak: market potential can't be reduced to a stockpile of "presence", of something that's stared at and analyzed like a tool on a shelf. Rather, it's always already operational, "ready-at-hand": what's needed is neither in the tool nor in my theory of the tool, but in the operational-skill-in-action.
Moreover: the skill to size a market doesn't live in any particular brain. "Sizing a market" can't be disentangled from the general problem of capitalism -- capital allocation in order to yield more capital. It's not like hammering a nail, it's like building a house. No single person can build a house: house-building, as nascent industry-building, lives in a community of practices. This is the crux of processes like YC "schools" and focal points like the Silicon Valley.
Let "sizing a marked" mean "estimating the number of products that can be sold to a given population of persons". An very precise estimate of this number is currently very hard to provide, but giving an order of magnitude estimate is clearly a Fermi problem. Note that the difficulty is providing an accurate and tight bound, not an accurate but imprecise one and that the latter might be valuable too.
Solving a Fermi problem requires estimating a series of numbers from everyday experience. One then multiplies these number together in a mechanical fashion, which could be done in a "ready-at-hand" way by use of a tool. Evidently the process invariably requires contemplating one's every-day experience, and since ordinarily that experience is simply lived this meta-action is an act filled with "presence".
By the way, the problem of capitalism is solved, and the solution shows that it will allocate capital optimally; this is one of the main arguments in favour of capitalist systems.
The problem that the social reality in which we live only partially approximates capitalism (see: non-profit organizations) is the big problem economists, and anyone else using classical economic theory in the real world, struggle with.
Investors should and do demand help in making those estimates.
A common way to help is to provide a full estimate plus the inputs you used to make that estimate.
Now, I think a BETTER way to help is to provide ranges and discussions of the inputs, rather than point estimates. And it's rare that I've encountered an investor who didn't find that acceptable. But some version or other of the exercise is worthwhile.