Steam sales estimates: why game popularity matters!

[The GameDiscoverCo game discovery newsletter is written by ‘how people find your game’ expert & company founder Simon Carless, and is a regular look at how people discover and buy video games in the 2020s.]

The weekend? No more. The working week? Very much upon us. Once again, thanks for supporting myself and my colleagues at GameDiscoverCo (there’s four and a half of us nowadays!) by reading the newsletter, using our consulting & all that good stuff.

Before we start, shout-out to Griffin Aerotech, Dodge Roll & Devolver for shipping an original arcade lightgun game - Enter The Gungeon: House Of The Gundead - as a tie-in to the popular indie game. (This was originally revealed in 2019, and is finally now through production, post-pandemic.) We love it.

[Heads up: only a coupla weeks left on our 33% off dealfor new GameDiscoverCo Plus subs. You get an extra weekly newsletter (example) that goes deep on games & trends, a Steam data back end (video) with much useful info, & GameDiscoverCo Plus Discord access. Sign up..]

(Yes, yes, it’s an old meme, we know.)

How do you know how many people have paid for - or own - a game on Steam? Right now, there are two methods: either a) do a multiplier of a game’s Steam reviews, as GameDiscoverCo has previously surveyed for, or b) try to scan individual Steam user profiles to calculate overall owners, as sites like SteamSpy and GamingAnalytics do.

Both have their pluses and minuses. The ‘reviews multiplier’ - also called the Boxleiter number - is by far the easiest to work out. But… why do we have to keep contradicting our own Plus back-end estimates recently for new, popular games using it?

We realized one reason might be because most people want to know about medium and high-selling game multiples, not low ones. And most popular methods of estimating are based only around the year of release - for example 20-55x reviews as a paid owner estimate for a Steam game released in 2021, with a median of 30.

And in our Friday Plus-exclusive newsletters, what often say - based on our clients’ data and some public metrics - is ‘actually the multiplayer for a hot new game is closer to 60x over time’. (And starts higher than that, since reviews take about 5-7 days to ‘catch up’ after initial release.)

But can we get any data to back up this hunch? Yes we can, and here’s the ‘recent years’ (2020-2022) results from our recent anonymous survey of 200-ish Steam games - which sold anything from 20 units to 16.8 million units (?!):

As you can see, it’s games that have sold less than 1,000 units that have the low multiple, with a 20x median. But it quickly steps up, with a 36x median multiplier for 1,000-9,999 units sold, a 49x median for 10,000-49,999 units sold, and 59x for 50,000-99,999 units sold. (It comes down slightly to 48x for games with >100k units sold!)

Why would this be, though? We can think of a couple of reasons:

BTW, our limited data seems to imply that Early Access games do get 20-30% less reviews per sale than 1.0 games, as you might expect. So if you see a game is in EA, add 25% to your previous multiplier estimates.

And yep, this ties in with data like The Wandering Village, which had 100,000 copies sold in its first week and got 1,148 Steam reviews during that time - an 87x multiplier. So sure, maybe it was a 60x for a popular game, plus a bit for Early Access?

So that’s an interesting starting point using this data. We’ve also made the raw survey results available (Google Drive link) if anybody wants to do anything fancier with it - or refute any of the above. But we think we’ve proven some causation here.

And yes, we think the median for all games released in 2022 could still be 30 across all games, and review ratios have come down over time. Here’s the data showing this - 2022 median was 29x in our data set:

But it’s just a question of what you want to look at. And if the games with low multipliers are majority those which sold less than 1,000 copies, it’s important to acknowledge that. We’ll be changing our Plus back-end calculation soon to reflect this.

Having said all of this, of course, there are still going to be obvious outliers. For example, Elden Ring just got to 17.5 million units LTD across all platforms, and somehow has over 400,000 Steam reviews, wow.

It’s likely this is because Elden Ring is a ‘game that people want to talk about’ - with >90% Positive reviews. And so 1 in 20-25 players leave a review, compared to the 1 in 50-60 that is more normal with a large-scale game. (We don’t think it’s sold >10 million units on Steam.)

So this is why all estimates can and should be taken with a grain of salt. In theory, you might be able to combine the ‘review multiplier’ estimate with the ‘look at Steam profiles’ estimate* to make something better than both. But that’s for another day…

[*There are a couple of disadvantages to the ‘scan public Steam profiles’ method - primarily that Steam profiles are private as default from April 2018, which can skew things, and ‘owners’ also includes things like Humble Choice key redemptions, which can be inflationary.]

Your browser can't play this video.

Learn more

Something we don’t talk about much is how game design can reinforce hook. I was part of an interesting discussion on a Discord about why Peglin (Peggle x RPG) ended up doing well on Steam - maybe 200k copies sold? But Roundguard (also a Peggle RPG!) hasn't broken out in quite the same way on PC - perhaps 15k units LTD on PC?

Quoting her with permission, Kitfox’s co-founder Tanya Short identified something smart, which is that Roundguard has ‘subtle power-ups’ for you to gradually progress through the game. This is still super fun to play* - but it’s less exciting to stream.

On the other hand, Peglin has power-ups which stack and upgrade in more radical ways, and can majorly change the meta - see the above ‘4 million damage’ Peglin video. (And here’s another 600k damage one from yesterday with 22k YouTube views.)

This reminded me of the fact that The Binding Of Isaac has done well in the long-term in part because it has such large, complex and buildable power-ups which completely change ‘runs’. It can be genuinely surprising every time you play it.

So while this may not be possible for all games you work on, it’s a good idea to consider much more aggressive/unexpected upgrades in a replayable game. If properly balanced, it’s potentially viral - and helps people see something different every time.

Two other great examples of this factor at work: Tanya adds: “My other example of crazy-different synergistic runs would be Luck Be a Landlord.” And duh, Marvel Snap has a whole heap of these ‘unexpected card combos/locations cause interesting results’ game mechanics, which the devs have obviously spent a ton of time balancing.

[Just a final shout-out for Roundguard, which is also on Apple Arcade and console, and I’ve personally played a lot. I love the ‘Peggle hybrid’ microgenre, and y’all should check it out too! And a gradual power-ups approach can still work fine - Spelunky is a good example of this.]

Finishing up the newsletter for this fine Monday, here’s some of the notable game discovery and platform news which turned up - while you were nappin’ or workin’:

Finally, since we’re occasionally allowed to succumb to people saying nice things about us, author Robin Sloan had some nice, outsider-y views re: the space we cover:

[We’re GameDiscoverCo, an agency based around one simple issue: how do players find, buy and enjoy your premium PC or console game? We run the newsletter you’re reading, and provide consulting services for publishers, funds, and other smart game industry folks.]