Launch day: what the data says

Everything below comes from this site's own launch archive. It is descriptive, not predictive: it tells you what has tended to happen, not what will happen to your game. Numbers are rounded and deliberately imprecise, because presenting a launch board's internal stats to three decimal places would imply a confidence nobody should have.

With that caveat stapled on, five things show up consistently enough to be worth writing down.

The first hour is most of the day

This one is not really a finding, it is arithmetic. The hot score divides by the age of a launch to the power of 1.5, so a vote at hour one is worth several votes at hour six, and by the next morning the denominator has grown by a factor of over a hundred.

What the archive adds is how sharply it plays out in practice. Launches that finish in the top few of their day almost all have visible early movement; launches that are quiet for the first couple of hours very rarely recover, even when the game is good and the votes eventually arrive. The ones that do recover are almost always the ones that got picked up somewhere else (a subreddit, a newsletter), which is the same mechanism arriving late.

The practical read: your launch is not a day, it is about two hours. Schedule for the two hours you can be at your desk.

Quiet days are underrated

Most makers instinctively launch midweek because that is when the most people are around. That is true, and it is also when the most games go up, and the second effect is usually stronger than the first.

Across the archive, the top slot on a weekend day is regularly held by a game that would have finished mid-table on a Wednesday. The audience is smaller but the field is much smaller, and the front page is a fixed number of visible slots. Being first on a quiet day means an entire day of visibility, while being sixth on a busy day means an afternoon of being below the fold.

This is not a trick and it does not work if the game is not good. It is just an argument for looking at the archive before you pick a date instead of defaulting to Tuesday.

Taglines with a verb do better

The strongest correlation in the whole set is also the least surprising: listings whose tagline contains a concrete action outperform listings whose tagline contains an adjective.

"Park a car using only the handbrake and the horn" is a shape that works. "An atmospheric puzzle experience" is a shape that does not. This holds across categories, across launch days and across whether or not the maker has an audience, which makes it the cheapest improvement available to anybody reading this. It is fifteen minutes of rewriting.

The failure mode underneath it is that makers write taglines for people who already know the game. You have played it a thousand times; the mood is what is left when the mechanics have become invisible to you. The stranger has the opposite problem.

Comments are the cheapest lever nobody pulls

Comments count for about a third of a vote in the ranking, which is not much on its own. What matters more is the second-order effect: a thread with replies in it keeps people on the listing, and people on the listing play the game.

Launches where the maker posted the first comment and answered every reply do measurably better than launches where the maker was absent, and the gap is much larger than a third of a vote per comment can explain. It is not that the comments are worth votes. It is that a live thread reads as a game somebody cares about, and people play those.

The version of this that does not work is the maker posting a comment and then leaving. An unanswered question under a launch is worse than no thread at all.

What does not seem to matter

Some things people worry about that do not show up in the numbers in any consistent way:

• Category. Puzzle games do not systematically beat action games or vice versa. Individual days swing wildly and it averages out.
• Which tools built it. The Built with credits have no visible relationship with performance, which is exactly what you would hope from a field that is meant to be a factual note.
• Screenshot count. Two good screenshots and five good screenshots perform about the same. One bad screenshot is a different story.
• Whether the maker has launched before. Second and third launches do slightly better on average, but the effect is small and mostly explained by those makers writing better taglines.

The honest caveat

All of this is a launch board describing itself. The sample is games that chose to launch here, the votes come from people who chose to be here, and every pattern above is downstream of a ranking formula I wrote and can change. Treat it as a description of one particular system rather than a law about games.

The parts I would actually bet on are the two that are just mechanics: early votes compound because the formula says they do, and a live comment thread keeps people on the page. Everything else is a tendency.