Domain names are not getting longer

There is a story everyone in this industry tells, and we sell a product that benefits from it being true. The good .com names are running out. Founders are pushed toward worse ones. Names get longer, hyphens creep in, numbers appear.

We went looking for that pattern in 5,538 .com domains registered by Product Hunt launches over two years, and did not find it.

Everything below is .com only, deliberately. Mixing extensions mixes populations, because a .ai label runs shorter than a .com one, so an all-extension median moves whenever the extension mix moves. Across all 11,422 registered labels the median is nine characters; across the .com ones it is ten.

Ten characters, every quarter

The median is exactly ten in all nine quarters. The mean moves within a range of four tenths of a character across the whole period, and its highest value is in the most recent quarter, which is also the partial one.

The distribution behind that median is unremarkable and stays unremarkable.

It is a single unremarkable hump with its peak at nine characters and a tail that thins steadily rather than breaking anywhere. There is no second cluster of longer names growing out to the right, which is what the scarcity story predicts and what two years of this data declines to show.

Hyphens and digits went down

If founders were being squeezed, the tells would show up in punctuation before they showed up in length, because a hyphen is what you accept once the clean version has gone.

Hyphens end lower than they start. Digits end at less than half where they started, though the 5.7% they start from is thirteen domains out of 227, so that particular halving is thin. The larger quarters are the ones to read, and across the four that carry more than a thousand labels each, hyphens sit between 4.5% and 6.3% and digits between 2.4% and 4.6%. Neither series has a direction worth calling a trend, and if either did, it points the wrong way for the scarcity story.

Why we are publishing a null result

This is a finding that costs us something, which is most of why it is worth writing down. We build a tool for finding available domains. A world where names get visibly worse each year is a world where that tool is more obviously necessary.

The honest version is more interesting anyway, because it sits alongside a result we published three days earlier that looks like its opposite. Running live registry checks on words from our own lists, we found new in front of fifteen different nouns registered fifteen times out of fifteen, and digital registered in both positions, thirty out of thirty. Less-used words in the same construction came back mostly available.

Both are true, and holding them together is the actual finding. The obvious names are gone. Founders keep registering ten-character .com domains anyway, at the same length, with the same rate of hyphens, year after year.

What that implies

The scarcity is real but local. It lives in a small set of words that everyone reaches for first, and in the positions they reach for them. It does not extend to the space of ten-character names, which is large enough that two years of sevenfold growth in launches did not visibly compress it.

What founders appear to be doing, and this shows up directly in how they map names onto domains, is moving sideways rather than downward. When the name is taken they shorten it, or they pick a different word, or they name the product to fit a domain they already checked. What they mostly do not do is accept a longer, uglier version of their first choice. The median holds at ten because people would rather change the name than degrade it.

That is a more useful thing to know than the scarcity story we went looking for. It means the search that matters is lateral, across nearby words and shorter forms, rather than a slow slide into get-my-app-2.com.

The launches come from the public Product Hunt API. The full method, the decisions that could have gone another way, and the dataset as a CSV are on how we measured this.

Limits

Length is a crude proxy for quality, and it is the only one we measured. A ten-character domain can be a fine name or a bad one, and nothing here distinguishes dailynote.com from a ten-character string nobody can spell.

This also measures what was registered rather than what was available. If the population of founders shifted over two years toward people better at finding names, that would hold the median steady while availability genuinely worsened underneath. We cannot rule it out from this data.

And Product Hunt is early-stage products launching publicly. A company that spent six figures on a four-letter .com is not in this set. Neither is anyone who wanted a .com, could not get one, and settled for another extension. Just over half of these launches did register something other than .com, but nothing here separates the founders who settled from the ones who wanted .app or .ai all along, and lumping them together would smuggle the scarcity story back in through the side door.