Total Addressable Market Is Useless

A few months ago, I came across a Tweet that clearly enunciated something that had rankled me for a long time as an early-stage investor:

TAM presumes zero future creativity.

Source: Twitter

I believe firmly that, for early-stage startups, TAM is meaningless, and over-focusing on TAM actually ends up being counter-productive in the long-run.

For those who aren’t familiar, TAM is shorthand for Total Addressable Market. It is a metric meant to provide a rough heuristic for how many users a company’s product could ever possibly reach, in the extreme upside case. For instance, one could calculate that the TAM of the internet is currently 7.88 billion people (the population of the world - ie. each person can only ever be one “internet user”), whereas the TAM of a US credit card rewards product would be 166 million (the total number of current credit cardholders in the US).

TAM is meant to be an unrealistically large number, ostensibly to demonstrate to investors and employees that the potential growth of the company - just from selling into its core market - is unbounded. Of the 1,000 or so startup pitch decks I have seen in the last half decade, TAM is almost always present – and it is rare to see a quoted TAM under 100 million users. This metric is usually quickly followed by a claim that if the company can capture just 1% of the market, it can realize [x] in revenue (where x is likewise a very high number).

An incredible amount of consulting and corporate research resources are dedicated specifically to the calculation of TAM for different products. If you have ever seen a news article with a headline like “the market for Artificial Intelligence is predicted to surpass $35 billion by 2027” or “crypto transactions forecasted to be over $100 trillion in the next 5 years,” there is a pretty good chance that there is a big four consulting firm behind the research, using it as marketing to attract business.

So what is the problem?

There are three fundamental shortcomings of TAM - one with the metric itself, and two with how it is frequently used in different corporate contexts.

First, with TAM itself: The calculation of TAM by default always embeds a laundry list of assumptions, most of which are rarely annotated.

Let’s take cryptocurrency market size projections as an example. TAM is usually determined via a bottoms-up analysis (ie: focus on the specifics of crypto users and then extrapolate to the broader population that matches those characteristics) or top-down (ie: focus on the macro drivers that may influence crypto adoption and then drill down into their implications for total market size). These factors are held to be constant, and then forecasted into the future assuming no major changes. Normally, TAM is not built with reference to all applicable user-specific or macro factors, so each estimate references an incomplete set of key drivers.

This is how you end up with projections of the 2030 market size of crypto ranging from $2.4 billion to $4.9 billion to $11.7 billion to $12.1 billion to $347.5 billion - two orders of magnitude between the higher and lower projections. And each additional year into the future that TAM is predicted introduces a significant amount of uncertainty (who could have forecasted a European land war or a banking crisis?) Net-net, it should be illegal to publish TAM without a stated confidence level and annotated assumptions.

The second issue is with how large corporations reference TAM. As previously mentioned, one of the primary reasons that consulting firms publish projected market sizes – with roughly about as much predictive power as guessing – is to sell to clients. The clients are usually large corporates who want to understand the investment case for a new product, business line, or acquisition. For instance: if I am a software developer and I launch a new hardware product, how many people could that hardware product realistically reach? Depending on how I price this product, how much annual revenue could it possibly generate?

For these kind of larger strategic decisions – which have significant downstream implications and imply a major allocation of resources – it is important to have a rough t-shirt sizing of the opportunity. An original equipment manufacturer (OEM) that dominates its niche in office computer manufacturing should have a good sense for whether the size of the mobile phone market ends in millions or billions before deciding to launch a new phone hardware line.

However, other than rough order of magnitude, the specifics of the new product line matter infinitely more than the market size. Which early users will you market to? Could the mobile product line leverage any of the core competencies developed by the computer business? Are the products themselves compliments or substitutes? Can the company generate ecosystem lock-in by rewarding users for buying both products? These details will ultimately be much more predictive of the potential top-line of the new product than whether the total addressable number of users is 10 million, 20 million, or 30 million. A mobile phone built for a market of 1 billion could flop, then add a feature that only appeals to a niche user-set in the millions, and then significantly scale revenue and user numbers as a result. (I would argue that for most products – not all – knowing and focusing on the problems of specific sets of users is a better path to success than one-size-fits-all, but that’s a different conversation.)

The last use case for TAM is one that I interact with personally: early-stage startups quoting TAM as a reason to invest. As the GP of The Fintech Fund, I am very lucky to be the recipient of many illuminating and exciting early-stage fintech startup pitch decks. I love reading about founders’ deep conviction for resolving specific problems, their analysis of why these problems have not yet been solved, their proposed solutions, and the team they’ve built to tackle them.

I always skip the TAM slide.

And the reason is that successful tech products tend to grow in concentric circles over time. To borrow the analogy from Dimitri’s tweet above - when TheFacebook.com originally launched, its addressable market was decidedly niche: an online directory for college students to connect with others at their college. Over time, as Facebook’s userbase grew and the network effects attracted new sets of users, the product broadened - first to an inter-college network, then to non-college students with a .edu address, and then to the world at large. Today, Facebook boasts 2.9 billion users – but nobody would ever have quoted that as a realistic TAM in a pre-seed or seed or Series A or Series B meeting.

The reason that successful products tend to grow in concentric circles is that good teams figure out how to build something that appeals massively to a specific audience, develop learnings from those users, and then leverage those learnings to launch additional products: either those that further monetize (and serve) their core users, or those that build on their initial product’s core competency to reach new users.

The MVP Pyramid: Build a lot of features for a broad market poorly, or a few features for a narrow ... [+] market well?

Source: Brian Pagán

Trying to build something that appeals to everyone at the outset is usually a recipe for disaster, as illustrated above. (The implied x-axis here is addressable users.)

This effect is even more powerful if the initial products are profit-making. Taking Stripe as an example: the company began its life as an online payment processor, before expanding into a massive suite of ancillary products. In aggregate, the transactions that Stripe processes are unit-profitable, ie: it makes money on each transaction, rather than losing money. (On average; chargebacks, fraud, etc. make some specific transactions unit-negative.)

By leveraging a profit-making core product, Stripe had the ability to effectively subsidize its entry into adjacent products like business incorporation, fraud tools, card issuance, etc., even if these products lost money. A cash-flow positive baseline product gives companies the balance sheet and flexibility to invest in R&D and experiment with new products that ultimately end up growing their TAM significantly more than where it was ever projected to be.

So to translate that to investing: if I’m looking at an early-stage pitch deck, it’s not very important to me whether the total number of users that a product could ever have is in the billions, and it’s not very important to me that 1% of the market equates to [x] revenue. (How does the company credibly propose to win over 1% of the market to begin with?)

A much more compelling narrative to me is “we have this specific user set that is crazy about our product, tell us they can’t live without it, aren’t attracted to any competitive products, and are willing to pay us for it.” The number of possible users may be in the hundred thousands, but if the company is that good at serving them, they will quickly win market share in a way that allows them to expand into their next adjacent product and grow their TAM.

At the end of the day, the only meaningful use of TAM that I can think of is as an interview question for undergraduate consulting applicants. Other than that, I think it’s safe to avoid ever overfitting for TAM.

How to Calculate the Total Cost of Any Recipe

Cooking your own meals at home has always been (and likely always will be) the cheaper option. You can opt to replicate your favorite fast food or restaurant meals, seek out simpler, less labor-intensive recipes, or experience the gluttonous glories of the past by digging through grandma’s recipe book. But just because cooking at home is generally cheaper doesn’t mean every recipe is a bargain. And while some recipes will offer a “per serving” cost, that doesn’t help you budget your grocery shopping. If you’re cooking at home more but not necessarily saving a lot of money, you might want to price out the total cost of your recipes to see where the true bargains lie.

The problem with most recipes is that they can obscure their true cost in a variety of ways. Different ingredients are priced and sold differently, and different amounts are used. Many recipes assume you have certain staples on hand at all times, and even if they offer guidance on their cost, it’s typically “per serving” costs, especially if it’s hyped as a budget-friendly recipe. Knowing that a lasagna will cost just $3 per serving is nice, but if you have to spend $50 to make the lasagna in the first place, that’s crucial information for your budgeting.

Pricing out a recipe is both a straightforward and somewhat mysterious process. Here’s a general guide for how to approach the problem.

Once you’ve worked out the cost for each portion of each ingredient used, add it all up—that’s the total cost of your meal.

Of course, there are other costs involved in cooking food. Not counting your time, you probably use water in every recipe, plus cooking gas or electricity (or both). Water is generally pretty cheap—about 3 cents per gallon—so your water cost in most recipes will be a rounding error. While it’s difficult to pin down exactly how much your stove and oven use might contribute to the cost, in general it costs around 10-20 cents an hour to run an oven at 350 degrees, and 10-15 cents an hour to run a burner at full blast. These numbers will vary greatly depending on the energy costs where you live and the age/efficiency of your appliances, but you can make a somewhat educated guess if you crunch a few numbers from your latest utility bill.