Deep dive: how Shapez.io went from web game to $1 million Steam hit

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

As you probably know, we love transparent data sharing here at GameDiscoverCo. Which is why we were very excited when Shapez.io creator Tobias Springer approached us, offering to explain how his title’s success actually happened.

The title is described on its Steam page as “a game about building factories to automate the creation and processing of increasingly complex shapes across an infinitely expanding map.” As Tobias tells us: “I really loved Factorio [but] I found it too stressful… also, I wanted to make a game which is as approachable as possible, since I found Factorio hard to get into back then.”

So yes, Shapez.io is Factorio-inspired, but with much more of an abstract and simplified graphical style, with replay hooks galore. If you’d like to understand the game better, I recommend watching this Brothgar (sponsored) YouTube playthrough:

Origin story: the free web!

But let’s concentrate on the story of how this game got started. Intriguingly, it was as a free web game, back in May 2020 - Tobias explains: “From my previous games like Yorg.io, I had a lot of experience with releasing games on the web. My marketing concept works like this:

The great thing with this is that you get a lot of free traffic when you have your game running on the web. There are numerous game portals (iogames.space, Armorgames, Kongregate in earlier times, Miniclip, Crazygames) who will list the game, and if you have a track record with them, it’s even easier.

The other great thing is that you can get feedback, so at the launch day you already have 100,000 players or more who playtested it - for free. Additionally, you also can start building up a community on the Discord, as well as collecting wishlists.” Here were the results for Shapez.io (the free web version) in unique daily players, post-launch:

The game was also released on Itch.io on the same day as the web version. The Steam version was still under review, so the page was up collecting wishlists, but that was it. Tobias notes that Itch sales were muted compared to later Steam sales, even when streamers mentioned it: “My guess is that Itch.io as a platform is much less recognized, so if a YouTuber links to the game there, most people won’t buy it.”

Constant gameplay iteration? Vital!

We’re talking a lot about discovery, of course. But there’s another notable advantage to having lots of users early. Using data from all of the free and early paid users, a lot of data-driven game design iteration on Shapez.io was possible.

This improves retention, and therefore recommendations/discovery. Tobias explains: “I track how long the player takes to complete each level, when he unlocks upgrades. I did quite a lot of tweaks to make sure you have something ‘new’ every 15 minutes, on average.” Here’s his internal data about the game’s first 20 levels, for example:

(The game is also open source, by the way, which helped people get excited about the concept, especially on Hackernews!)

Steam launch and take-off!

So the Steam version of Shapez.io launched on June 7th, 2020 at $3.99 USD - relatively inexpensive! (Its base price got increased to $4.99 in October 2020 alongside the Wires update, and to $6.99 in March 2021, and Tobias agrees that it’s still very modestly priced for the value it gives.)

What’s interesting is that its pre-launch Steam wishlists (the game had only been up for a few days) were very poor - only 958 wishlists at launch! Tobias notes: “People actually knew the game already from Itch.io. Actually, a lot of them bought the game again on Steam because they liked it so much, which also gave it a nice push.” So that’s a good reason to offer early versions of the game elsewhere.

But here’s where the virality of the gameplay really helped Shapez.io take off. The game was variously featured in the ‘New & Trending’ section on Steam, ‘The Community Recommends’, and got featured in the ‘similar games’ section for Factorio, which Tobias thinks helped a lot.

Because of all the game’s natural hook/attractiveness, this visibility, and streamers (which we’ll discuss shortly), “the sales were exploding”, as Tobias says:

Another post-launch win for Shapez.io was Reddit - around the time of the June 11th, 2020 spike, Tobias:

The pitch for Reddit - “I made a "Zen Mode" Edition of Factorio” - is a great one, and then “…the increased visibility also was noticed by streamers, and Giantwaffle, a Twitch streamer with >800k Twitch followers started to pick up the game. His streams had around 3,000-4,700 peak watchers, which is quite a lot on Twitch.”

From there, a lot of the YouTubers who normally play Factorio started to pick up the game. For example, KatherineOfSky (146k subs on YouTube) started a Let’s Play series on the game, starting on June 19th. And when sales started to dip in early July 2020, Tobias says: “I wrote a lot of emails about sponsorship opportunities (I think 500-1000 in total). Luckily that worked out, and I got quite some big YouTubers to cover the game, which was really worth it for me.”

Some of the best-performing paid YouTubers included DangerouslyFunny, whose October 9th video has over 1.25 million views and performed very well for Shapez.io sales-wise. I will say that the reason this works is that the game has a hook, and DF built a further hook (“I Was Paid To Try And Break Shapez.io”) on top of it.

Another major win was from Aliensrock - and also a great example of how paid ads can occasionally lead to organic videos, as Tobias explains: “This one was a really nice audience match - the initial video was really well received. From this point he even started to play the game in his streams and make other videos for free, since he liked it so much.” 

Sales for the second half of 2020 for Shapez.io kept being driven by streamers and a Steam Daily Deal in November 2020 (big spike to the right, below.) There were also overall revenue improvements, thanks to an October change to a $4.99 USD base price:

But going into this year, things were a little less sunny. Tobias notes: “There were a few scheduled sales in the first half of 2021, as well as another Daily Deal, but besides that, it looked like we’d reached the long tail period for the game. I decided I wanted to give it another boost, and started work on a DLC for the game.”

But what, precisely? He says: “People really enjoy the puzzle-centric optimization part of the game, so I decided I’d make a Puzzle DLC, where you have limited space and complete predefined or user-submitted puzzles.” Makes sense.

While working on that, Tobias and a second programmer also added Steam Achievements to the game. This was just in time for the second price rise to $6.99 USD in March of this year, which was right after the Daily Deal hit.

Around this time, Tobias says, “…a Chinese publisher, Doyoyo Games contacted me. They were offering me a MG fee, promotions in China and a release on the Wegame platform. Since the China sales were at just 3-6% at that point, I happily worked out a deal with them”, starting with the Puzzle DLC release (pictured below).

In a traditional ‘you make multiple bets, and not all of them pay off’ move, the build up to the puzzle DLC was timed to some traditional PR/press releases Tobias had paid for, which just didn’t return sales-wise.

But the Chinese partnership was a major success: “They arranged a partnership with some key selling sites, and they also managed to sponsor one of the biggest influencers on their YouTube-platform equivalent (BiliBili).”

Here’s the worldwide sales units for DLC & China launch in June 2021, up to now:

The success of this launch bled into the Steam Summer Sale, which started in late June, and Tobias notes: “After the Summer Sale, somehow Steam seemed to like the game more. The game got more traffic, and the sales almost doubled since then. I believe this is due to a lot of external traffic, which makes Steam believe the game is more popular.”

By the way, the Puzzle DLC has sold around 17,000 units ($62,000 in gross revenue), with around 30% of them being from China. This shows the strength of the Doyoyo Games partnership there, and the amount of new Chinese buyers timed for that DLC launch.

But the real prize in this revitalization has been in boosting long-term interest in Shapez.io itself, increasing yield, and extending sales more globally. Seems like the franchise has a while to run yet!

What didn’t work?

Before we get to the grand finale (the full Steam revenue and player charts!), I had a chance to ask Tobias what didn’t work for him, paid advertising & discovery-wise. He identified the following things:

Tobias’ summary? “The game is simply too cheap to advertise for on the regular advertising providers.” We completely agree. If your game costs $30-$40 and you are being very targeted, you may be able to do profitable paid advertising for Steam games. But at $7 gross and $4 net per copy sold in the U.S., trad advertising is too pricey for it.

Conclusion: lots more beautiful data!

So, the web version of Shapez.io is still important - Tobias says it’s “driving a lot of sales from the 100-200k monthly users - around 2.500 to 3.000 units are being sold solely through via web version monthly.”

And Itch.io is a lovely platform which has helped out beneficially. According to Tobias: “In total, the game made roughly $8.000 on itch.io. However, getting the game on itch.io was really easy, they also were really friendly and also gave me a spot on the landing page when I released.”

But the Steam version is really driving revenue for Shapez.io. So let’s finish up by looking at some high-level indications of what’s been going on. Firstly, here’s Shapez.io Steam units since its launch, with annotations on the source of sales spikes:

Next up, here’s lifetime gross/net revenue on Steam for Shapez.io - not including DLC revenue - via the SteamWorks back end. Reminder that the ‘net’ here is before Valve takes its 30% cut. (BTW, that refund rate is impressively low, presumably due to game quality and lower price?)

And here’s the regional split for Shapez.io nowadays. You may remember that the Chinese global unit share was 3-6% before the partnership and localization, so you can see how that’s boosted things:

Finally, here’s the play time stats for the game - 21 hours, 48 minutes average, and 7 hours, 11 minutes median time played? Those are some monster stats:

We’ve literally run out of space in Gmail for any other images. But also wanted to highlight that the game has had 138,000 wishlist additions and 18,000 deletions, with 54,000 of those becoming purchases. This made for a lifetime wishlist to purchase conversion rate of 39.9% - way, way above average.

Thanks again to Tobias for being so transparent - we received 23 pages of notes (!) to put together this newsletter. And we hope you all find this helpful as you pursue a hit of your own, as part of the wild world of game discovery. Until next time…

[Thanks for reading this newsletter! If you loved it, our GameDiscoverCo Plus paid subscription includes the ‘Complete Game Discovery Toolkit’ eBook for members, a Steam Hype back-end for unreleased games, a neato Plus member Discord - where I arranged this article with Tobias, haha - & lots more. You can upgrade today!]