Challenging the challenging assumptions about Total Addressable Market

To generate a forecast for your SaaS startup though, the model requires a bunch of assumptions, simplified below:

1. When is product-market-solution fit achieved? (Hypergrowth begins)
2. How long does it take the firm to achieve market share saturation between product-market-solution fit achieved, and approaching the saturation point? (takeover time)

3. What is the Total Addressable Market (TAM) / what is the effective saturation point?

Starting with the first assumption – entrepreneurs are horrendous at estimating when hypergrowth begins. Finding 11 from the Genome Startup Report bears that out – “startups need 2 to 3 times longer to validate their market than most founders expect.” (p. 5). That fact alone isn’t too devastating. The Venture Capital industry manages entrepreneurs in such a way that it doesn’t matter nearly as much.

The second assumption, takeover time, isn’t nearly as dangerous. It’s important. It’s esoteric. It’s high marketing science. And if anybody asks, a post will be written.

The third assumption, TAM and saturation, is important. The concepts are interrelated.

Saturation, the percentage of a given population will adopt the product, is easier to safely estimate than TAM. The model depicted above depicts at 51% saturation point on a TAM of 100,000 customers. Why 51%? Because it means winning in startup scenarios. In an emerging technology area where disruption is rife, reaching 50%+1, in a crowded field of competitors, is winning.

The startup ceases behaving like a startup at 51%. It becomes a business. And with that comes buying out the second place competitor and kicking off harvesting that market and all the other typical anti-competitive behavior that goes with full capture.

Setting saturation at 51%, for SaaS startups, is a terrific target assumption because it’s a goal that is reasonable, and it simulates competitive pressure.

In this example 100,000 customers because it made the chart easier to read.

Note that there isn’t any sort of additional detail about who the customers are, if they’re self-referential when making purchasing decisions, or even what the product is. All of that reality lies outside the model.

Let’s assume that the founding team is rational and they truly want an estimate of TAM. One way would be to consider the social network, count the number of known noses from their non-random sample, and project the total number of customers from there. So, let’s say the founder has a rolodex of 1000, and they can count 100 total people, they might project outward that there are 100 * 1000 total potential customers. That’s fraught with bias.

Working the other way, they may identify a particular market segment as a population, and project downward, the kind of wishful ‘if I only get 1% of all the people in China to buy my toothbrush’ sort of logic.

But it’s also where some solid psycho-demographic targeting can start to come in.

Just a for instance:

There are around 2.1 million Canadian men aged 25 to 34 in Canada, based on all the census data. Using postal code data, about 100,000 of them live alone and earn more than 75K a year. So, single rich men, 25 to 34, constitute a segment of 100,000. Many of them know each other and frequently refer to each other when making purchases about video games. That rises to the level of being a market. Getting to 50,001 of that market, at a monthly ARPU of around $19 dollars…now that’s an interesting beachhead. If you can’t provide enough value for $19/month, then that isn’t an interesting problem to solve then, is it?

There are tends of thousands of beachhead markets – clusters of self-referential people – just waiting for you to solve a problem. If you consider how many of them there are, along with how much they typically spend, or could spend, you got yourself a better set of assumptions against to estimate Total Addressable Market.

Get it right, spend the right way, and the curve responds.