Hey, all! Turns out that I’m accumulating enough interesting & relevant game discoverability tidbits to do a monthly update - ‘here’s some fun stuff I found on the Internet, and my comments on them’.

So this is the second one of these compendium pieces - here’s the first one. Let’s get it started!

Steam Labs & machine learning to help pick existing games to play?

Funnily enough, I was just talking at the DICE Summit in Las Vegas with other attendees about game discoverability - and their impression that platform companies had underinvested in it.

Well, I think that’s true for a lot of the ‘big guys’ like Sony & Microsoft whose game featuring is still primarily - though not solely - led by commercial concerns (do we own the studio in question? do we have a pre-existing retail relationship with them? Are we pushing a subscription service? Is their game a big hit already?) And Nintendo ain’t that sophisticated yet.

But that leaves Steam, and they’ve rolled out a new Steam Labs experiment, ‘Play Next’, which intriguingly works on titles that you already own. As they explain:

“Problem: You have a bunch of unplayed games in your Steam library, and you can't decide what to play next. Solution: Our machine learning system helps you to choose, by suggesting the games it thinks you'll enjoy most, among games you already own.

The Play Next experiment uses the same underlying technology that powers the Interactive Recommender, and applies it to your existing Steam library. Up to three selections are shown at a time. If the first set doesn't quite grab your attention, you can cycle through to see other suggestions (when available).”

I’ve made no secret of the fact that I think Steam has made big strides in discoverability over the past year or so. It’s one of the fairest systems out there for ‘make the right kind of game, & you can have a hit!’

Of course, you still have overwhelming numbers of games out there. But I particularly like this innovation because - sure, it’s post-purchase. But if people feel like they are managing their backlog of already-purchased games better, they may be more inclined to buy more.

Even more numbercrunching on the state of Steam!

One cool thing that happened recently is that Jake Birkett (Shadowhand/Regency Solitaire), probably one of the prime discoverability numbercrunchers, got hold of Sergio Garces’ amazing stats dump on Steam in 2019, and made some interesting observations based on it.

Wanted to particularly highlight these two Tweets:

#1 - "In 2013 median indie games made approx. $110K and top 25% made $590K. - In 2019 median indie games made $1400, and everything below that basically $0. Top 25% made $12K, which is obviously not sustainable. - You have to be in the top 5% of indie games these days to make $413K."

#2 - "@sergiogarces Some other quick observations from the dataset: - There are 30K games on Steam - New indie releases have levelled off. But still pretty high. - Median revenue per game has been dropping since 2013 peak but is maybe levelling out now new indie games are levelling out."

So, wow, 30,000 games on Steam now? That’s… quite a few. The Twitter replies on this thread were also interesting, with Charles Goatley noting, in relation to the scarily diminishing median revenue: “25% of 499 games released in 2013 is 124. But 5% of 8384 in 2019 is 419. That's 3x more games being "hits" in 2019 than in 2013.”

So maybe that might make some of you feel a little bit better? But my favorite response was from RimWorld (smash hit, folks!) dev Tynan Sylvester, who sagely noted:

“It used to be that people were mostly prevented from succeeding by gatekeepers like Steam. Now they're fail[ing] because customers just aren't interested in their work. The second one is more psychologically challenging, but not necessarily worse.”

I feel like that’s… probably true.

Is it worth doing a demo for your game on Steam?

My No More Robots compatriot Mike Rose wrote an interesting Twitter thread about the demo for Yes, Your Grace which NMR put out just before the holidays. Click through and read on for the full details:

"Thread about whether it's worth doing a demo for your Steam game *pre-launch*: Devs talk a lot about doing demos for your game after it has launched, but not so much about pre-launch demos. I decided to find out for myself whether it was worth doing one 📈GET READY FOR GRAPHS📈."

But if I was to TL;DR to you, I would say here’s Mike’s best attempt at a takeaway: “I learnt that demos [may] heavily reduce the number of people who wishlist your game, which seems detrimental... But then, maybe people who play the demo are more likely to buy it? I didn't learn anything, did I”

But he concluded that he would not try a similar experiment - free Steam demo on the game’s main page - again, because it was definitely not a clear win, statistically.

The general takeaway here is - it’s not even possible to A/B test this type of thing, unlike in some other areas of F2P games or websites. And Yes, Your Grace already had buzz before this experiment - its now up to #75-ish in most wishlisted pre-release games now. So difficult to split out the demo effects from all the other variables.

But Mike did note an interesting (not necessarily Valve-endorsed) trend of making a separate game entry for a free prologue to your game - as done by Backbone - and then trying to get people to hop across to the other app to additionally wishlist your game there.

I think this CAN work, but also is a bit of a hack and will probably only work for some VERY specific types of game. We’ll see! (It’s good - and fair - that Valve doesn’t explicitly prevent this, though.)

Bonus links!

To finish out, here’s some other neat things you might have missed:

- Daniel Sanchez-Crespo, who has had a decent success in Early Access with Killsquad, wrote a good Twitter thread about lessons learned from his first foray onto Steam Early Access. (His studio previously worked closely with Sony for a number of years.)

- For anyone who loves fetishizing Steam traffic breakdowns to work out where your players come from, Lottie Bevan has a super-detailed breakdown of post-launch traffic sources for Cultist Simulator that you should probably read.

- I know, this newsletter talks too much about Steam and not enough about other platforms. So here’s an interesting tidbit from Ars Technica about how one particular Google Stadia ‘free’ game for Stadia Pro seems to be faring. (Not great, but it’s very, very early days for Stadia, and the dev presumably got paid a lump sum anyway for inclusion.)

- Lastly, I enjoyed this Chris Zukowski article on actually talking to your customers, one on one. There’s an argument you can also or additionally do this well via surveys & Discord. But the general sentiment is correct - you need to identify (& size!) your player base and engage them, both pre and post-release.

[You’ve been reading Game Discoverability Now!, a regular look at how people find - and buy - your video games. Or don’t. You may know Simon from helping to run GDC & the Independent Games Festival, and advising indie publisher No More Robots, or from his other newsletter Video Game Deep Cuts.]