The metric that decides if you're a studio or a platform and most founders never name it ( I will not promote )

I've been building in a category (location-based AR for heritage sites) where I watched three well-funded companies build beautiful products and then just... stop growing. For a long time I assumed they had run out of money or gotten acquihired. They didn't. They ran out of a different thing, and it took me too long to see it. Every one of them capped out at around 5 locations. The reason is a number almost nobody names out loud: the marginal cost of your next unit of supply. For us that unit is a monument. For a marketplace it's the next city. For a services-flavored SaaS it's the next enterprise onboarding. For a hardware company it's the next SKU. The label doesn't matter. What matters is the shape of the curve. There are only two shapes. Flat line. Unit 10 costs the same as unit 1. You are a studio. Not a slur, some of the best companies in the world are studios, but the economics are what they are: your growth is linear in headcount, your margins are capped by labor, and investors will price you accordingly no matter how many users you have. Descending curve. Unit 10 costs meaningfully less than unit 1, because unit 1 built something reusable. You are a platform. Every unit you ship makes the next one cheaper, and at some point the thing you built to make units becomes worth more than the units. Here's the uncomfortable part. Most founders in flat-line businesses believe they're in curve businesses. I did. You tell yourself "we'll systematize it later," and then you never do, because shipping unit 6 is always more urgent than making unit 7 cheaper. The tell is simple: can you say, out loud, what your last unit cost versus your first? If you can't produce that number in under a minute, you don't have a curve. You have a hope. What actually bent our curve, in order of impact: Stop researching, start compiling. Most of the expensive "original work" we thought we had to do had already been done by someone else and was sitting in a public archive. Check whether your expensive input already exists somewhere as a cheap one. Build the library, not the artifact. The first version of anything should be built with the assumption that half of it becomes a component. This costs more the first time. It's the only thing that pays. Kill any step a human does twice. Not the steps that are hard. The steps that are repetitive. Those are different lists and founders keep automating from the wrong one. None of this is novel. It's just that in the excitement of proving the thing works at all, the cost-per-unit question gets deferred, and by the time you ask it you've already built five bespoke things and there is no library, just five snowflakes. So: what's your unit? And what did the last one cost compared to the first?