What a Game Demo Can and Cannot Tell You Before Buying

During the last Steam Next Fest I downloaded nine demos, played all of them on stream to a chat of two people, and bought three of the games. One of those three I refunded at 95 minutes, because the thing the demo was good at turned out to be the only thing the game did. That result is roughly my long term average, and it is a lot better than the years when I bought on trailers.

The reason demos trip people up is that nobody tells you what you are holding. A demo is not a slice of the finished game cut at random. It is a separate build, chosen and tuned to sell, usually branched off the main project months before release. Treat it as a sample and you will misjudge games in both directions, buying ones that fall apart at hour six and skipping ones whose first twenty minutes happen to be their weakest.

What a demo build really is

Studios build demos from a branch, lock it, and then keep working on the main game. By launch day, a demo that went out for a June festival can be four to six months behind the shipping code. That gap explains most of the confusion people have afterwards: the stutter you felt is often already fixed, and the ability you unlocked at minute ten might be a tenth hour unlock in the real game.

The content is picked rather than sampled. Developers choose a section that works without context, which usually means an early area or a self-contained encounter with the difficulty flattened so a stranger does not bounce off in the first five minutes. Festival demos also sit under practical constraints: they need to be understandable to someone who will give them fifteen minutes, so tutorials get compressed and the interesting systems arrive early.

None of that is dishonest. It does mean the demo is answering the question "can this game make a good first impression", which is not the question you are asking.

The things a demo answers reliably

Here is what I take from a demo as near enough fact, in the order I check them.

How it feels to hold. Input latency, camera speed, whether aiming or platforming is satisfying at the millisecond level. This is engine level and almost never changes between demo and release. If a character feels floaty in the demo, it will feel floaty in the game.
Whether your machine can run it. Not the exact frame rate, but the shape of it. Open the frame counter and look for the stutters rather than the average. A demo that dips below 40 on your hardware at medium settings is unlikely to become comfortable.
Accessibility and comfort options. Subtitle size, colourblind modes, remappable keys, motion blur and field of view sliders, separate volume channels. Demos ship the real options menu more often than not, and a menu with three sliders in the demo rarely grows a dozen by launch.
Art direction and sound. Whether the style holds up on your screen, whether the music loops annoy you, whether the UI is legible at your viewing distance. All of this is established long before a demo exists.
The core loop, once. You will learn what the game asks you to do minute to minute. You will not learn whether doing it 400 more times is fun, which is the next section.

What a demo cannot tell you

Pacing is the big one. A demo is 20 to 60 minutes and a game might be 30 hours. The question that decides whether you finish a game is how it behaves in hours eight through fourteen, when the novelty has worn off and the systems have to carry it. No demo touches that window.

Difficulty curve is the second. Demo encounters are usually tuned softer so they do not stop a newcomer, and the version in the finished game may well be harder. I have been caught out by this twice, most memorably by a tactics game whose demo mission I cleared on the first try and whose equivalent real mission took me nine attempts.

Then there is everything that only appears at scale: the late game economy, whether the crafting system collapses once you have money, how much backtracking the map design forces, save system annoyances, and bugs that need a long session to surface. Multiplayer population is in the same bucket, and it is the one that burns people on smaller releases. A demo tells you nothing about whether a lobby will fill in six weeks.

A demo shows you the first impression a team wanted to make. The review question is what the game is like once impressions stop mattering.

The no demo red flag argument is wrong

You will see this line a lot: if a studio will not put out a demo, they are hiding something. I thought that for a while and I now think it is backwards often enough to be bad advice.

A demo is a product. It needs its own build pipeline, its own QA pass, its own store page, its own save handling, and it needs maintaining if the game patches during the demo's life. For a small team that can be weeks of work taken directly out of finishing the actual game. I would rather a four person studio spend that time on the last act than on a polished twenty minutes.

There is also a genre problem. Narrative games, mysteries and horror lose real value by showing their opening, because the opening is the hook and you only get it once. Several developers have said publicly that a demo measurably reduced wishlist conversion for their story driven game, and that pulling it helped. For a puzzle box, playing the demo can cost you the best part of the full experience.

So the absence of a demo is information about the type of game and the size of the team, not about quality. What I do treat as a warning sign is a demo that exists but is locked behind an email signup, or one that is clearly a trailer with button prompts. If a team shipped a demo and would not let you touch the actual systems, that is a choice worth noticing.

The ninety minute demo protocol

Years of editing stream footage taught me that I remember how a game felt wrongly, almost every time. So I stopped trusting my memory of a demo and started recording them. The whole routine takes about an hour and a half across two sittings.

First sitting, 25 minutes. Play with capture running, default settings, no tinkering. Go through the options menu once at the start and screenshot it. Then play badly on purpose for the last five minutes, meaning take fights you should lose and poke at the edges of the map, because the failure states tell you more about a design than the success states.

Second sitting, the next day, 20 minutes of playing plus 15 minutes of watching your own capture at double speed. This is the step people skip and it is the valuable one. Watching your hands on footage shows you how much time went into menus, how often you waited on an animation, and whether you were actually having fun or just being competent. I have talked myself out of two preorders this way.

The gap between sittings matters as much as the playing. If you have no interest in loading it again on day two, that is a real signal, and it is the closest thing a demo gives you to an answer about hour ten. Pair it with the approach in what to watch in a streamer's VOD before you buy a game, since a long VOD covers the stretch a demo cannot, and if the game is a console port check whether the PC port is worth buying at launch as well. More of this kind of pre-purchase work sits in the game reviews section.

FAQ

Are game demos the same as the final game?

Rarely. Most are branched builds locked two to six months before release and tuned to be approachable, so performance and difficulty often differ from the shipped version.

Does no demo mean a game is bad?

No. Demos cost a small team weeks of work, and story driven games can lose their hook by showing the opening. Absence tells you about team size and genre, not quality.

How long should I play a demo before deciding?

About 45 minutes split across two days, with 15 minutes spent reviewing a recording. The second session is what reveals whether you actually want to return to it.

When do Steam demos usually appear?

Next Fest runs three times a year, around February, June and October, and many developers keep the demo on the store page afterwards rather than pulling it.