Early Access on Steam: should you use it?

Hades went into Early Access in December 2018 and reached 1.0 in September 2020. Slay the Spire spent November 2017 to January 2019 there, Vampire Survivors December 2021 to October 2022, Baldur's Gate 3 October 2020 to August 2023. These are the stories that make Early Access look like a softer launch: release unfinished, fix it with players, launch properly later.

What actually happens is that the Early Access release is your launch. Steam emails your wishlisters, reviews start counting and stay on the page, and the first impression is set. The games above worked because they were already fun on their first Early Access day; their players were buying a good game that would get bigger, not a promise. A rough Early Access launch rarely gets a second chance at 1.0.

An independent study by the Gamer of Passion blog, published in October 2026, followed every Steam game it could find that entered Early Access between 2013 and 2023. Among the 10,141 with at least ten reviews:

• 61% had reached full release by 26 September 2026; counting games with fewer reviews, 51%.
• 26% were still in Early Access with no update for over a year, and at least 7% had been removed from sale while in Early Access.
• The median time to 1.0 was about 11 to 12 months: 315 days for games that entered from 2013 to 2020 and finished within five years, 352 days for games that reached 1.0 between 2020 and 2025.
• Late finishes are rare. Of games with no 1.0 after three years, 21% reached it in the following five.

Players know this, which is why the answers on an Early Access page are read so closely. A plan you can keep for about a year is the norm; a plan for three years asks players to trust you through the period most games do not survive.

• Do not launch without a playable game. Valve's minimum is a trailer showing gameplay of a game people can play now.
• Do not launch if development is done and you only need final testing.
• Promise nothing specific. Players should buy on the current state, not on future features or a completion date.
• Not crowdfunding, not a pre-order.
• Steam parity. Early Access must not reach Steam later than other stores, or cost more on Steam than elsewhere.
• No permanent discount during Early Access; set a lower base price instead.
• A price increase at 1.0 blocks discounts for 30 days, and cancels the launch discount if it falls within 30 days of the transition.
• Keep the page current. Steam shows a notice on Early Access pages with no build and no update event for more than 12 months.

• Two launches. Steam emails your wishlisters at the Early Access release and again when the game moves to 1.0, and Valve gives full launch visibility only at the full release.
• Revenue during development, from players who want the game now, rather than from an advance with a revenue share attached.
• Thousands of playtesters who balance, break and suggest at a scale no closed test reaches.
• Time to find the game's audience before the launch that counts most.

Each comes with its cost: two launches means the first one has to be good; revenue means paying customers with expectations; playtesters write public reviews.

Early Access fits games whose fun is in systems that get richer with content and tuning, and fits badly with games whose fun is a story told once:

1. Is it fun now? Would a stranger pay today's price for today's game, with no promises? If not, it is a playtest, not Early Access; see Steam Playtest.
2. What will players change? If you cannot name it, you do not need Early Access.
3. Can you update for a year? A regular cadence, every two to six weeks, is what players read as alive.
4. Can you answer the six questions honestly? Valve's Early Access questions sit above the buy button; our guide to answering them shows what a strong answer looks like.
5. What will 1.0 be? If you cannot describe the finished game, the Early Access period has no end.

• Story games. Players finish them once; an unfinished story spends your best first impression and invites spoilers.
• A first hour that is not ready. Reviews written in Early Access stay on the page at 1.0.
• No capacity to update. An Early Access game that goes quiet collects reviews saying so, and after a year Steam's own notice says it too.
• Money as the only reason. Valve is explicit that Early Access is not a way to crowdfund; a game in Early Access for revenue alone tends to read that way.

Plans change. Valve asks developers to contact it if Early Access development ends. A game that is playable and well reviewed can stay on sale without the Early Access label, which cannot be reversed and brings no launch visibility; otherwise it can be removed from Steam, and Valve may offer refunds to buyers. Neither is a good outcome, which is the best argument for entering Early Access only with a scope you can finish.

Flockbay is a free AI game builder on the official Godot engine. Its playtest links put builds in front of testers before you decide on Early Access, and bring back their errors and crashes for your AI to replay and fix. Its Steam page drafts the Early Access answers with the rest of the page, for you to review, as a .zip you upload yourself. Hosted playtests and the drafting are free in beta until 1 December 2026.