Having just finished the main development cycle of Dojo Storm, there are plenty of lessons that we learned along the way. We took 3 of the most important ones to share with the world, hoping that someone else can benefit from our experiences.
1. Insufficient QA Time
From the outset, we decided that the final 2 weeks of development would be enough to test, fix bugs, and polish the entire game. For a year-long project, this was not nearly enough time. We were able to fix all of the game-breaking bugs in time, but had to scramble to fix the rest. We could have also used more time to polish the game and optimize it.
In order to remedy this issue in the future, we intend on building QA/polish time directly into our production schedule. For instance, if we have a 4 month development schedule, we will be sure to only plan enough work for 3 of the months, so we can spend the final month polishing and testing.
2. Disorganized Documents
Dojo Storm's world has a lot of areas in it, and even more dialogue. As a result, we had dozens of documents to keep track of everything, from NPCs to student descriptions to items. Unfortunately, some of them ended up containing redundant information, and when something was updated in one document it may not have gotten updated in all of them. This led to the incorrect info being used in the game on several occasions. Fortunately it was always caught internally, but it still resulted in additional work to fix it.
In the future, rather than having multiple documents that might end up containing the same info, we will be utilizing a wiki to serve as a centralized repository for all of our content. This will prevent the same info from being copied to several different areas, and we will only have to update it once in a single place.
3. Late Distribution
For Dojo Storm, we had a hard deadline: the commercial was airing on September 12th, so we needed to make sure the game was available for download on both of the app stores in time. We were able to get both up in time, but we had to expedite the iOS version because we underestimated how long it would take to be reviewed. On top of that, we did not realize that we wouldn't be able to update the screenshots without going through the review process all over again, so for the first few days we had only a single screenshot of the game on the iOS version.
Fortunately, this is the sort of issue where once you encounter it, you learn from the mistake moving forward. In the future, we will be sure to upload an iOS version 2 weeks before any deadlines, and will include all of our screenshots with the initial submission. Android is much more lax with its rules, so we are not as concerned with submitting to it in time.
So there it is. We learned a ton during this project, and plan on taking as many of the lessons forward as possible. Our total post mortem consisted of a lot more points, but these 3 are what we feel are the most important. We hope others can learn from our experience, so they don't have to make the same mistakes we did.