ZOMBS.io: 0 to 1 Million in 7 days
I thought I’d write this so I can document my thoughts at this moment, as well as to just write about why we — my partners and I — decided to make a online browser game out of the blue.
In February, I went to GDC (Game Developers Conference) in San Francisco for the first time. It took going there to realize most people don’t actually go inside — I spent a total of 2 minutes in there (apologies to my sponsor who paid for the ticket if you’re reading this). Instead, everyone just chills at the hotels.
As someone who plays games relatively often, I find myself spending 90% of my free time trying to find a good game to play, rather than actually playing. Despite all the top titles and popular games out there, I was never hooked on anything for more than a day. After going to GDC and seeing passionate indie game developers, I decided — hey why not make our own game?
So we did.
We had no idea what the hell we were doing at first. We had no tech, no game idea, and definitely not creative artists on our team. Finally, we decided on the first thing — we wanted it to be multiplayer. Single player games are absolutely boring. Social connectivity fuels my personal desire to play games, so there was no way we were ever going to make a single player game.
We wanted to “innovate,” but it’s tough when gaming is such a big industry and everyone’s probably done everything already. Plus, breaking into mobile app store on the first try is probably not a good idea. Despite “io games” being “2016,” we settled on browsers as our platform.
The first game we made was lasersharks.io, which was a 2d shooting game where you’re a shark……..with lasers. It took us a week, but let’s not talk about this game anymore. It eventually got over 100k users, but the game play was terrible.
Next up was “zombs.” We all love tower defense, like Clash of Clans, Bloons, etc. But those games were really “AFK experiences.” You open your phone, upgrade a few things, then you close it and wait. It wasn’t. Clearly SUPERCELL is doing everything right, but for me, it seemed a bit boring.
So what we decided on was real-time massive (as in…more than 2 people), multiplayer tower defense game. FYI — still have no idea what direction Zombs.io is going towards. We felt that there was some way we could combine the fun tower defense aspect of those games, add in a multiplayer/social aspect and create something that we could play around with.
This was what Zombs.io became. It was never meant to be anything serious because we just wanted to fiddle around with game dev tech and see what people liked.
Minecraft, Roblox, and other sandbox multiplayer games have been inspirations for this, but they were too lax. In Minecraft, you just kind of run around do whatever you want, there was no pressure. That was the reason we added Zombies that force-spawned at night (every 2 minutes). With that, it pressured players to work fast in building their base.
Zombs.io, as it is right now, is a multiplayer 2D online base building, tower defense game — with a party system. (I’ll probably expand more on why we chose what we did in a subsequent post).
For now, I want to go over our metrics from the first 7 days.
When we launched the game, we expected to get maybe 30 people, turns out we got over 88,000 in the first 24 hours due to simply organic traffic by submitting our games on “io game sites.” As tempted as we were to use the 1.1M follower Twitter account, we felt that’d be against the whole purpose — which was to experiment if our game is actually sticky.
Similar to Pokevision, we couldn’t have been more ill prepared with our server optimization (our AWS bill is crazy). We had to scale to 500 servers (c4 larges) amidst the panic. After slow realization that we were using only <50% of the capacity on average, we’ve since downsized.
We hit our peak concurrent (14,000 players) on the third day due to a popular Youtuber streaming it on Youtube (our average at that time was actually around 3,000. On day 7, today, our average is just over 8,000 concurrent players at any given time. Quite exciting!
On the first day, we’ve had an Overwatch professional player sit there for 12 hours playing the game. We only have like, 30 minutes of content, so what the heck? He started sketching out base designs and we were even like “WTF? People like the game?” Anyways, him and other day users (and everyone else) has been amazing help in terms of giving feedback.
As this is essentially our first time launching a game of any sort — we had no idea of the costs…let alone bandwidth. I was warned about bandwidth costs by Stanislav Vishnevskiy — let’s just say RIP $. (sidenote: we only learned of the 3–30x multiplier on bandwidth costs in South America YESTERDAY). RIP, again. Within two days, our operating expenses skyrocketed to 5 figures. We could have spent day 2 and 3 optimizing, but we figured improving user experience and focusing on growth was far more important. We finally optimized on day 4, which reduced our cost by 4x.
Keeping this short — we had someone reach wave 453, of course he could've AFK’ed the last 200 waves, but that’s still 15 hours in the game…Wow.
At the time of writing this, we have seen over 1,130,000 players total in the first 7 days (which as of now, hasn’t ended yet). This is crazy! Despite Pokevision having more 75M users in the first 10 days, getting 1.13M in 7 days for Zombs is still mind blowing to me.
Moving forward — we have a few things we’d like to try out since we’re lucky enough to have a big player base to run some large scale case studies. Tonight, we’ll launch our first major update, but for the next few days we will definitely be improving our infrastructure; we haven’t even set up an autoscaling script for deploying/shutting down servers yet, haha.
If there’s anyone that have any advice for us, or just want to chat, definitely don’t hesitate to message me on Twitter Yang Liu
My DM’s are always open!
Thanks for reading!
Yang
Updates:
Day 8: Avg concurrent (at noon EST)
Day 9: Avg concurrent (at noon): 12.5K
Alexa (rankings are logarithmic, but still fun to look at)
- 20k global rank, 5k US since day before
