How Many Players Actually Finish Games?

This isn’t something I’ve really done before, but I was nosing through the Steam achievements for some different games – I wanted to find out whether the completion stats for Wolfenstein: Youngblood were unusual or not. I ended up on a bit of a data spiral, and – well, I found some interesting things. I’ll give you the table off the bat, and then we can talk through it. Currently they’re not ranked highest to lowest – I’ve gone roughly alphabetical by series, because I think it’s easier to look at for some of the stuff we’ll discuss. Here’s the table:

GameTutorial completedStory completed
Arkham Asylum69.230.7
Arkham City84.940.8
Arkham Origins81.840
Arkham Knight8838.6
CoD MW264.648.5
CoD MW385.655.1
CoD BlOps69.942.3
CoD WWII78.344.9
Sniper Elite 4 73.329.5
Dark Souls 392.923.2
Dishonored85.441.1
Dishonored 2 91.137.9
Hitman57.422.5
Hitman 242.516
Overcooked79.98
Overcooked 28619.8
Shadow of Mordor88.536.3
Shadow of War65.714.3
Wolfenstein: New Order7646.8
Wolfenstein: Old Blood9457.4
Wolfenstein: New Colossus93.950.6
Wolfenstein: Youngblood77.628.4

So some of the initial things to note here – on average, from this list, 35% of players finish a game on PC. There are some very low numbers that are throwing us off there – for instance, Hitman and Overcooked both have exceptionally low completion rates. In the case of Hitman, it’s possibly people who have the games but never played them – the middle column lists the number of people who’ve essentially finished the tutorial missions in each game, and both Hitman games have the highest number of players who clearly never started. The Overcooked games are the opposite – they start strong, but drop off quite quickly. For the first game, only 52% completed the first level after the tutorial – so already they’d lost nearly 30% of their players between starting the tutorial and finishing the first level. That’s probably a result of those games being party games – you get your mates round and play it together. It’s obviously more difficult to get them round on multiple occasions to play it through to the end. There are also difficult or longer games dragging that score down – Dark Souls 3 has a 23% completion rate, and Sniper Elite 4 took me 25 hours to complete a single play-through, which no doubt contributed to its 30% completion rate – and then there are games that people just didn’t like. Wolfenstein: Youngblood has a 28% completion rate, and a 2.3 user score on Metacritic. The critical rating is just under 70, so it’s not that everybody panned the game – but the combination of players review-bombing and a low completion rate makes sense.

So we might suggest that, unless there’s an issue, maybe 30-40% of players normally finish a triple-A game. That seems like a fair estimate when you exclude the games that have very specific reasons for low finish rates. And obviously, you know, we’re dealing with a sample size of 22 games, so it’s not statistically significant. We’re just chucking around numbers. Academic studies have shown, for instance, that completion rates could be as low as 10% or 15%, although – this is tricky, right. For one, I can’t even read that article, so I’m just quoting from the highlights. At best we can use the database that they link in the article description, which does at first glance show a very low completion rate. Some of those will be the glut of low-level indie games dragging down the count (for instance, Legends of the Universe – Starcore, which costs $3AUD and has been completed by 4.7% of players), but even if you limit the database to games that are big enough to have a Metacritic score (240 games total), you still end up with a 16% completion rate. There are some gaps and complexities in this data – for instance, the article is from March 2019, so some games naturally have different completion rates now (they list Arkham City as having a 47% completion rate, but at time of writing it’s 40.8%). Some games they incorrectly list as not having a completion achievement (including Arkham Knight, Dishonored, and Sniper Elite 4), meaning they aren’t counted, and sometimes I’ve used a best-fit achievement where they’ve been more circumspect – for instance, Dark Souls III and Dishonored 2 have multiple endings, and no specific achievement for finishing the game, so I’ve used the achievement stats from the most common ending, meaning the real number will be higher. Use the data at your own risk, basically.

Anyway, let’s get back to our tiny dataset of twenty games. I think it’s important to talk about the number who completed a game in terms of the number who started it. There are a bunch of games in bundles that might have been redeemed but never played, which would throw off the count, and there are also a bunch that have a strong multiplayer component, such as the Call of Duty games, where we can assume that some players do not play the single player campaign at all. In Modern Warfare 2, for example, 65% of players have the achievement for completing the first campaign mission, but then 71% of players have the achievement ‘Two Birds with One Stone’, which you get for killing two enemies with one bullet in either the single player or in Special Ops, a series of co-op shooter missions. You don’t shoot anyone in the first campaign mission – so clearly, 6% of players had played through one or multiple Special Ops missions, picked up the ‘Two Birds’ achievement, and then never started the single player campaign. So it’s worth comparing start rates against finish rates – that gives you a sense of how many people dropped out of the game, rather than the number of people who never started. Again, I’ve taken a best-fit approach, taking the earliest possible story achievement and comparing it against the completion achievement. There will be a bit of wiggle room in there – but it’s essentially comparing the number of people who finished the tutorial against the number of people who finished the game. When we run those numbers, we see quite a shift:

GameTutorial completed Story completedStory completed as % of Tutorial completed
Overcooked79.9810.0
Shadow of War65.714.321.8
Overcooked 2 8619.823.0
Dark Souls 392.923.225.0
Wolfenstein: Youngblood77.628.436.6
Hitman 242.51637.6
Hitman57.422.539.2
Sniper Elite 4 73.329.540.2
Shadow of Mordor88.536.341.0
Dishonored 2 91.137.941.6
Arkham Knight8838.643.9
Arkham Asylum69.230.744.4
Arkham City84.940.848.1
Dishonored85.441.148.1
Arkham Origins81.84048.9
Wolfenstein New Colossus93.950.653.9
CoD WWII78.344.957.3
CoD BlOps69.942.360.5
Wolfenstein: Old Blood9457.461.1
Wolfenstein: New Order7646.861.6
CoD MW385.655.164.4
CoD MW264.648.575.1

Key shifts here include Arkham Asylum, which has quite a low 31% completion rate. When that’s taken as a percentage of people who finished the tutorial (in this case, people who knocked out Zsasz, as that’s the earliest story achievement), it shoots up to 44%. I wonder here whether people bought this GOTY edition as part of a package deal after having bought and played the original, and just didn’t bother to replay it. I know I got this game as a hard copy, originally – it came out in 2009, when Steam wasn’t as much of a thing. I only bought and played the Steam version this April, eleven years after the original release. That might account for the relatively low start rate of 70%, which is at least 10% lower than every other Arkham game on this list. This approach also suggests broader attitudes towards the Call of Duty games. A lot of people don’t bother to play the single player campaign, but when they do, they generally finish it. At least 50% of players complete it, and up to as much as 75% – you can see how they’re all clustered at the bottom of the list. The Wolfenstein games also generally have quite a high completion rate, which – well, that’s the surprising thing about Youngblood, isn’t it. A base completion rate of 28% is very low. Weighting it against people who completed the tutorial doesn’t improve matters much, raising it to a middling 37%. New Order does have a similarly low tutorial completion rate, although in that instance the earliest story achievement was in Chapter 4, so that’s what the 76% is drawn from. About as many people made it to the end of Youngblood‘s tutorial as made it a quarter of the way through New Order.

And I understand why Youngblood had such a high drop-off so early in the game. It’s very different in style to its predecessors, which gives it a high bar to entry. There are damage types, which are new; there’s a weird RPG system set up around weapon upgrades; stealth is presented as a valid option at character creation but it’s not really viable until the mid-game; if you’re playing by yourself, you have to manage your dumb-ass AI sister; your enemies have levels and health bars, which feels fussy and micromanaged – it’s a lot to get used to. It’s very fiddly compared to the rip-and-tear DOOM-esque gameplay of the previous three instalments. And – like, I really enjoyed the previous Wolfenstein games. I’ve written literally over two dozen articles about them. And I found Youngblood utterly unplayable after maybe two hours. It’s only several months later that I’ve come back and tried it again – I managed to finish it this time, but, yeah, 28% completion sounds about right.

So there you go. Other writers have carried out their own half-assed analysis – for instance, Ungeek estimates 34% average completion rate for popular PS4 games, based on an analysis of ten titles, and a Gamasutra article offers a graph based on nine Xbox games. None of us are really doing anything other than scratching the surface, but it’s still interesting. It gives us some of the tools to start thinking about why games get finished, and maybe why they don’t. That in turn gives us an opportunity to think about how we talk about game structure. Only 8% of people finish Overcooked? Sure – so what should the developers be trying to do with the time that’s available to them? Who are they making their game for, and how do they treat the attention that they have? It’s not to say that playing for longer is necessarily better, or that players have to finish a game in order to find value in it – and it’s also not to say that developers should kowtow to every trend in audience behaviour. It’s just worth knowing what those trends are, so you know what you’re getting yourself into with what you make.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s