His results showed that the longer a game stayed in early access, the weaker its full-release sales tended to be.

I mean, this seems kinda inevitable, no? The longer you're in early access, the larger the portion of your target audience has likely already bought your game.

There are tons of scenarios where I can see it being useful, and I can often see a clear difference between it and release, but the problem I've got now is that there are so many finished games I could be spending my time and money on right now that it's hard to justify buying an early access game. I think the last one I bought was Palworld, which I played for about 20 hours right when it came out, and now I'm waiting for 1.0 rather than the iterative feedback that early access thrives on. They've still got plenty of people to get that feedback from, but that's the biggest early access release since Minecraft, so it's an outlier.

I'll allow it for smaller studios that have a big, offbeat idea. Games like Factorio, Satisfactory, Kerbal Space Program and Subnautica. All of these games had multi-year early access campaigns that were very successful, Satisfactory in particular. I think it's appropriate for weird games like these that have uncommon mechanics like factory building, space flight or scuba diving.

Thinking about Satisfactory, I imagine their sales weren't spectacular on launch day last year, but a lot of their customer base had already bought the game, so they got their $30. Maybe another way to phrase it is, who cares if it sells before or after launch?

I feel like Abiotic Factor did a really good job with this. It released last summer into EA, then over that next year they released chunks of the game, sort of like as the story was unfolding, until the full game was finally done back in June. Then they upped the price by $10, but not before putting it on sale again for the 1.0 release.