Allometry is the study of how proportions change with size. A mouse scaled to elephant proportions collapses, because strength grows as the square while weight grows as the cube. Growth is not magnification — past a threshold, the structure has to change. Operating businesses work the same way, and nobody names it.
A restaurant at $2M is a restaurant. At $10M it is not five restaurants — it needs a commissary, central purchasing and a brand function, or it breaks. Every operator eventually asks this and answers it on instinct, because no system anywhere answers it with numbers.
| The line it adds | What it reuses | What it does to the valuation |
|---|---|---|
| A second location | brand, playbook, suppliers | More of the same multiple |
| Catering | kitchen capacity in off-peak hours | Slightly better — utilisation |
| Retail pasta sauce | recipe, brand, supplier relationships | A consumer-goods multiple — roughly double |
| A boutique hotel | location, brand, hospitality operations | Asset-backed — higher, and financeable |
The rerate is worth more than the revenue. A restaurant group trading on restaurant economics that becomes a consumer brand earns consumer-brand economics on the whole entity. This is the sharpest form of pricing backwards from enterprise value, because sometimes the honest answer is not raise your prices — it is you are the wrong shape for the valuation you want.
An adjacent line is one that reuses what you already have and needs the least new capital. Every input is something a closed loop already knows — which is why this is the payoff for closing it, and why it cannot be built first.
This page exists because it is the clearest statement of what the substrate is for, not because you can buy it. Two things have to be true first, and neither is a matter of engineering time.
Spare capacity, margin headroom and demand all come from modules writing to the same unit. With three of five event classes populated, every recommendation is computed from a flattering partial ledger. We would rather ship nothing than ship that.
The rerate figures are the entire value of the output, and they are currently marked as assumptions in our own spec. A wrong multiple is worse than no feature when it is the number an operator makes an irreversible decision against.
Close the loop on one operation. The same data that would power this is the data that defends your margin now — which is why the sequence is what it is rather than a matter of preference.
Every input Adjacent Possible needs is something the six modules already produce. A free Margin Scan is the shortest path to finding out whether closing the loop is worth it for you.