Glass Cannon Release Log
This post is going to be a log of how my recent videogame release, Glass Cannon, did on Steam. My expectations for the game are as follows:
- It will sell 1.5k copies at $5 over a year.
- Approximately 1k-1.2k copies in the first 6 months and 300-500 copies in the next 6.
- The first month will probably only see up to 100 sales.
- In the second and third months the game should reach peak concurrent players and therefore sales as it will slowly get discovered.
- Playtime distributions will remain largely the same, with the edges getting more pronounced. So, we will see more players with very low playtime (<30 minutes) and more players with very high playtime (>10 hours, which is top 5 percentile for this game)
Let's get into it.
Day 1
As expected, a very modest reception, much like Gnomber, my first release. I'll be comparing the game to Gnomber as that's my only other game that had any significant sales. It should follow my 1.5 estimate for sales even at the start.
Day 2
Definite improvement over Gnomber: 11 vs 2. Ignore the 1 retail activation, that's a free key I gave out. Glass Cannon keeps people interested for longer, so the higher number makes sense. Doing more daily comparisons after this doesn't make sense because Gnomber's numbers basically oscillated between 2-3 sales a day until May 12th 2024 where a small youtuber covered it and it got 123 sales on that day alone.
So I'll do a weekly comparison, then a monthly one and in between compare performances of the two games when they receive streamer/youtuber coverage. Because the numbers are going to be most certainly different, I'll compare the fatness of the tails from the sales boosts. In other words, what was the average daily % of sales loss after the boost.
Day 9
Now we're up to 120 sales total. Day 3 had 11 more sales and day 4 had 40, which can be seen in the peak below:
The sales have levelled off as I had 7 sales yesterday and only 2 today. The major issue with the game is that it's too hard at the start and too easy once you start a few effective combos going. Players keep reporting that the earlygame is too tough and it's too easy to die unexpectedly. This causes them to quit the game earlier than intended, thus diminishing the value of the game in Steam's eyes, so it stops showing my game to people.
Steam did indeed stop showing my game as much, but a different major thing happened on the 3rd release day to have caused this bump in sales. gamingonlinux.com covered my game with an article . This is reflected in the store page traffic as well:
The traffic from the article probably accounts for at least 1/3 of all direct navigation (the green line) on May 3rd and 4th. It's unidentified traffic probably because linux people are particularily privacy-minded so they're probably using browsers or plugins which strip referrer information, so Steam has no clue where they came from.
The initial bump in direct navigation on May 1st is definitely mostly bots that steam couldn't determine were bots. It bumped right after release, so all the game aggregators and bots quickly picked up on it.
The discovery queue can be seen doing a lot of work and mirroring external page hits. Gnomber's main source of new players was also the discovery queue. Along with the game being a new release, it also gets a bump in the discovery queue, so it makes sense that it stayed so high and then had a dampened and slow reduction as Steam slowly lost confidence in the game's ability to sell.
Problems with the game as of Day 9
The game is difficult at the start and easier at the end because the player's strength scales exponentially whereas enemy health scales linearly. Once you get a few combos going, they play off of each other combinatorically which scales it even faster and you start to outpace enemies really quickly, usually towards round 19 or so. Unless you get cocky, surviving round 19 basically means you'll win round 24 too, even though at round 24 the average enemy health is like 8 points higher than at round 19.
When you have no combos, you're relying on the base weapons and stats, which scale slower than the enemies, so the early-mid game becomes hard to bypass. The healths here should be lower so you actually get a chance to get to the part of the game where even weak earlygame combos get a chance to start popping off.
The solution it to just scale the enemy health exponentially. This will slow the ramp-up in the early game and increase it towards the end where the player feels invincible and stops having fun because there is no more uncertaintiy of winning.
The second issue, which I've patched just now, was insta-deaths. There were just too many cases in the game where you'd instantly die without you even knowing you made a mistake.
The third issue is content. I thought 120 mods, 16 weapons and 8 enemy types with 24 randomly generated rounds (for each run) were enough content for the game, but turns out it's not. Now I'm actively working on the 'coming soon' stuff I snuck into the game at release. These are challenges, arenas and selectable cannons which appear in the game's menu with their previews and text saying that it's coming soon, but can be unlocked by doing whatever.
The previews would give the players a sneak peek at upcoming content and tell them that I intend to update the game. Had I known that the game would pique the interest of so many people from the get-go I would've capitalized better on the launch and had come prepared with content and better balancing.
The first and third issues could've been avoided given 1 more month of work. I should've played my own game more and should've added the content I'm working on now. That definitely would've made the initial wave bigger, but I think it would've died out eventually after a slightly longer time. Still, given the information I had at the time, launching on May 1st was the best decision I could've made. Delaying a game that already took triple the initial estimated time to make would've been hell.
The insta-deaths would've probably still made it into the game because I was so blatantly blind to such an obvious flaw. I've played my game for up to a hundred hours, I've watched people play the game for hours as well. People have told me it's too hard at the start. I myself could only beat the game 4-5 times. The fact that I was so blind to this flaw despite all the feedback means that no amount of beta playtesting would've revealed that to me - people actually buying the game and venting to me in frustration is probably the only thing that would've communicated it to me.
I've patched the game with a mid-sized update that addresses issue #3. I don't expect it to start selling until another wave of players is brought to the game though, possibly through a streamer. I know eventually some small-time content creator will find and play the game for an audience, so I'm now preparing for that 2nd wave of players. The game must have dozens of hours of content when that happens so it can ride that wave and not fall on its face like it did with the launch.
Despite this flaccid initial showing, it's still better than what I had anticipated. I expected to reach 100 sales in the first month, but that happened in the first week. I'm feeling more positive on this, the game seems to have some potential. So much so that I think if sales start becoming literally 0 like with Man Pit, I'll probably try to advertise it to get the ball rolling again because I think the money and time spent on the marketing will be easily made back by the resulting sales - up to a point, of course.
I reckon it won't go down to 0 sales though. The game is much more appealing than Man Pit, so Steam will probably keep giving a small trickle of players who over time should bring in more if the game is good. Remains to be seen.
Day 18
To describe the last few days, sales quickly went down to 1 or 2 a day (May 08 - May 14 on the graph). During this time I pushed out a bunch of updates addressing the issues stated above. Also put out new content, like the hexagon arena and a new challenge level. After that the game got to 10 reviews (all positive!) and the game just now started selling again.
The traffic numbers explain the surge:
Most of the traffic is coming from steam itself through the discovery queue, a whopping 92%. Here is a breakdown of the traffic from May 15th to May 17th (the latest date I can specify):
Steam is doing a lot of heavy lifting, obviously. There are a two possible reasons for this. One - the game passed 10 reviews and they're all positive, so the game gets a small boost in Steam's ranking algorithm. Two - the game's playtimes are seemingly improving.
Steam won't show me a zoomed-out bucket graph by hours when the median playtime is lower than 60 minutes, so it's a bit harder to tell, but since the most significant 'minimum time played' brackets went up, Steam has taken notice. I'm especially thrilled that the 5h bracket increased by 7%.
We're so back? Not so fast
The boost in traffic was so sudden that it seems to me that Steam's algorithm rewarded the game for reaching a milestone. The game shows up as 'positive' after all and before that it didn't have a rating, so Steam probably just did a oneshot boost which will likely be short-lived because the game has meiocre retention. In contrast, for a good game like Nubby's, any small wave can jumpstart a chain reaction which not only makes sales persist for a long time, but begets more and larger waves. For the really good games like Minecraft or Stardew, the never-ending wave only dies down when the product has reached complete market saturation.
That domino image from the popular Chris Zukowski article is how I believe virality on every platform works, not just Steam. This principle applies to any viral product or idea in general. More users beget more word-of-mouth marketing which begets more users.
Unfortunately, I won't be able to follow this new small wave of players with consistent updates in the next week because I'm attending a (so far overhyped) gamedev conference in Krakow. I'm very skeptical of these kinds of networking and schmoozing events in general, but I still went because I had a free pass. I'll write about it once I experience it in entirety, but I'm not expecting much.