The number that would actually settle this thread is the one DomainDan raised: what share of last year's renewal bill the portfolio's own sales paid for. I keep that ratio for the ccTLD side I work in, and the pattern there is not kind. A list where sales fund roughly a fifth of renewals is not a portfolio, it is a subscription you are paying to keep a story alive.
Two things I would add to the trimming test.
First, three quiet years is usually not a market statement, it is an exposure statement. A name with no enquiry in three years is often one that nobody outside your own head has ever seen. Before I drop a name on silence alone, I check whether the silence came after any real attempt to reach a buyer, or whether the name has simply been parked in a marketplace search box and left there. Those are different problems. One costs money, the other costs nothing to fix once.
Second, the funded-niche point cuts deeper than it looks. When a sector has money, the companies in it buy exact words, and they buy them early. So the question is not whether a niche is hot but whether a live company already uses that word as its own address. I measured that on a ccTLD recently across a set of single dictionary words, and the split was blunt: the words serving an operating business were a minority, and a comparable number were parked or on a broker lander, most of them on the same parking nameservers. Renewal increases are painful for the second group and almost irrelevant for the first, because a name that answers real traffic is priced by the buyer's business, not by your annual cost.
So my own order of cuts goes: no enquiry after two renewals, then no live user, then a word that a funded sector has already solved with a compound. The extra dollar per name decides almost nothing on its own. It just makes me write the list down earlier in the year, when there is still time to pick which names deserve the next twelve months.
