We hold more names than we use. On a ledger it reads as an expense — a line item, renewed annually, attached to nothing. Judged as purchases, most of our domains are bad purchases.
They are not purchases. What you acquire when you register a name is the right to build something under it later, at a cost fixed today. The name does nothing on its own. The option does.
The cost of waiting
Every option has a premium, and for a domain the premium is the renewal. You pay it each year for the privilege of deciding later. The arithmetic is unremarkable:
cost = n × r (bounded, known) value = max(0, V - n × r) (unbounded, unknown)
Where n is years held, r the annual renewal, and V whatever the name turns out to be worth to whoever eventually needs it — us, or someone building the thing we once described to ourselves and never built. The left side is a rounding error that compounds slowly. The right side is either zero or large.
That asymmetry is the whole decision. The question is not is this name worth what we are paying but is there a plausible world in which V exceeds a decade of renewals. For a name that is short, plain and unencumbered, the answer is usually yes. For a clever one, usually no.
What a name buys you
The more useful thing a good domain does is constrain. A vague name permits anything, which is another way of saying it directs nothing. A specific one carries a product inside it — a scope, an audience, sometimes a business model — and owning it keeps that direction cheap to walk back to.
We have started projects because we held the name and got tired of looking at it unused. That sounds like a poor reason to build something. In practice it is a reasonable one: the name had survived several years of us not needing it, which is a longer test than most ideas are ever put through.
When we let one go
Most of them never become anything. The model assumes as much, but the assumption has to be acted on, or a portfolio of options quietly becomes a pile of renewals.
So once a year we read the list and ask one question of each name: if it expired tonight, would we buy it back tomorrow at the market price? If not, we let it drop. The names that survive are consistently the ones we cannot describe a replacement for.
We are not domain investors — we do not buy to resell, and we have never commissioned an appraisal. We are not collectors either; a name we would not build under is not an asset, it is a subscription. What we keep is a small set of names we expect to use, and a small annual cost to keep that expectation open.