What Went Wrong

1. Skimping on the Metagame

One of the corners we cut to get the game out was the world map that you play in. In the early days of War Commander, you were presented with a list of targets (AI- and player-controlled bases). You could scout and attack anyone on the list, and killing them would remove them from your list and replace them with a higher-level target. The system worked well, but it was a poor substitute for a real world map, which is a game in itself. We didn't get around to adding a map until April 2012, eight months after launch. As soon as we released it we saw a 25 percent jump in engagement, retention, and monetization -- and realized how important the world map meta-game layer was (and how stupid we had been in not delivering it sooner).

2. Not Going Wide Enough, Soon Enough

For too long we went down the path of adding more buildings and units to the game to satisfy the needs of the late-game players. Instead we should have added features that create sandboxes they can play in. Instead of offering them new things to build, we should have given the players new ways to tweak existing content to make the game unique to each player and his or her army/base. This changes the game from "I have everything built and at max level" to "I have everything at max level, but maybe this isn't the best combination of things; I need to experiment!" This attitude has the benefit of creating a wider variety in base designs and attack strategies, making it more varied for the attackers and defenders.

3. Poor Time Estimates

We had a solid RTS game; you had your base, your army, and a world to dominate, but there was one bit missing: real-time battles. For many technical reasons we didn't allow attacking between two players that were online at the same time. We fixed this in October 2012 with the launch of Live Battles, so now players who were online at the same time could take part in the same battle and interact with them in real time. We knew this was not going to be something that would drive monetization in a big way, but we did it anyway because it was cool and we were not happy putting our name to an RTS that didn't have synchronous PvP battles.

We knew it would be hard to retrofit synchronous multiplayer into a live game while continuing to develop and release new features, but we had no idea how long it would actually take. We had to rewrite a very large portion of the game in a completely different language (a language none of the existing dev team was familiar with) so it could run on both the client and server. This took us a painful, drawn-out six months or so to develop, test, and release. If we had done it before we launched, we might have been able to save ourselves a few months of work, and if we had taken the time to sit down and correctly break out all the tasks and estimate them, instead of just diving into production, we could have better planned our time. The one silver lining to come out of it all was a new process on the team that has helped us hit every deadline since the launch of Live Battles with a pretty high level of accuracy and minimal crunch.