The two things worth looking for
When a new concept lands on the table, a strategist is really looking for two things. The first is differentiation. A genuine USP if you’re lucky, a defensible moat if you’re luckier, or at minimum a way of telling an old story differently enough that the market has a reason to listen. The second is the size of the opportunity, which is where the familiar three-letter trio comes in: TAM, SAM and SOM. The total addressable market, the portion you could plausibly serve, and the share you can actually obtain. Profitability sits in the mix too, of course, though the cost of goods is usually an operations question rather than a strategy one. Differentiation and market. If a concept has both, you have something to work with. If it has neither, no spreadsheet will save it.
The trouble starts with which of the three market numbers people fall in love with, and after enough years watching business cases get built, the pattern is painfully consistent. CEOs and CFOs fall for the TAM, or at a stretch the SAM. Almost nobody falls for the SOM. The reasons are understandable. The first two numbers feel factual. They come from analyst reports and industry bodies, they arrive with a source you can cite, and they’re satisfyingly large. The SOM, by contrast, depends on messy things: market conditions, competitive response, distribution, and the effectiveness of your marketing. It’s an argument rather than a fact, and arguments are harder to fall in love with.
One per cent of everything
Then comes the move that turns a preference into a problem. If the honest numbers for your actual product look thin, there’s an easy fix available: broaden the definition. Don’t size the market for what you sell, size the category it sits in. The accounting tool becomes part of the business software market. The protein snack becomes part of the health and wellness market. Each broadening is individually defensible, and each one inflates the TAM and SAM until the plan can support the sentence it was always heading towards. Here it comes. We only need one per cent.
It’s a wonderfully seductive line, because it sounds conservative. One per cent! We’re not even being greedy. Surely marketing can find us one per cent of the market. But that category number was never your market. One per cent of it isn’t a modest target, it’s a fiction with a decimal point, because the customers making up the other definitions in that category were never available to you at all. There is one honest exception. If you’ve invented something truly unique that a mass of people desire or need, the big number might actually be yours to chase. But then you’ve solved the problem the other way, because a product like that barely needs marketing to find its market. If your plan requires marketing to conjure the share, you don’t have that product, and the one per cent isn’t coming.
The machine loves your assumptions
This is an old failure, older than any of the current tools, but AI has given it a new production line. Ask AI to help build a business plan and one of the first things it does is ask for inputs. Market size, expected share, growth rate. It’s a reasonable request, and it’s also the moment the whole exercise gets decided. Feed it the category TAM and the one per cent assumption, and the model won’t push back. As we’ve written before, these tools are trained to be agreeable, and they are never more agreeable than when handed a number to build on. What comes back is the myth, beautifully formatted: revenue projections, hiring plans, a five-year story, all resting on a share of a market you were never in. The plan hasn’t been tested. It’s been typeset.
And then the bubble has to burst somewhere, and it usually bursts on the marketing team. They’re the ones handed the finished plan and asked to deliver the one per cent, which means they’re the ones who have to stand in front of an enthusiastic room and explain that the market in the model isn’t the market that exists. Being the dose of reality is rarely popular, and it’s a role they were written into the moment the first optimistic number went into the machine.
Agree the numbers before the machine gets them
The fix isn’t sharper software and it isn’t a bigger analyst subscription. It’s sequencing. Before anyone builds a case on the numbers, get the team that will have to deliver them into the same room and reach agreement on what they should be. What is the real serviceable market for this product as it actually exists? What share is obtainable, by when, and what has to be true for that to happen? That conversation is cheaper than any failed launch, and it’s the one step AI cannot do for you, because agreement among people is the whole point of it.
AI can help enormously once you’re there, and it helps most when you deliberately point it against your own optimism. Ask it to attack the market definition. Ask what evidence would distinguish your SOM from wishful thinking. Ask it to find the comparable products that chased one per cent of a category and got a fraction of a fraction. Asked that way, it’s a sharp and tireless analyst. But go in the other way, install the rose-coloured glasses at the start by handing it your fondest numbers as facts, and don’t expect anything back except your own enthusiasm, extrapolated to five decimal places. The machine will size any market you describe. Describing the right one was always your job.