One tool I do not see enough developers use is a content burn map. This is one of the most powerful pre-production tools for any Producer or Product Manager and can answer questions from scoping and progression to drop rates and content burn rate. Have you ever pondered: How powerful will player characters be after 3-months? How many levels do I need at launch? How much content do I need to make and at what rate do I need to release it? Then I have the tool for you!
A content burn map, which is a name I unceremoniously came up with for this article, is a tool that models how different cohorts of users interact with your game over a period of time, and the state of those players during that interaction. We’re going to use back-of-the-napkin math to model average engagement with our game then model different cohort of players and their interactions with it. From there we can answer so many of the pre-production questions that end up sinking otherwise great game launches.
The Match-3 genre is such an easy example to start with because these games are generally very linear. Let’s start by modeling the engagement with our simple Match-3 we’ll call, well, “Simple Match”. Simple Match is a simple game, it is string of linear levels where player attempts their highest level (leading edge level) and if they beat it, they unlock the next level. For this example, we are going to try to answer the questions “How many levels should Simple Match have at launch?” and “How many levels at what release rate should Simple Match run while live?”
Let’s start by modelling Simple Match from it’s lowest level interaction up to highest level concept. At its lowest level, Simple Match requires the player to swap two pieces on the board. It’s simple. At the next tier of interaction, we want the container that holds these swaps and dictates the rules of whether our swapping was successful or not. Simple Match has level container that those swaps take place in, Simple Match calls the level container well… a Level, and a level ends by either the player completing the objectives or the player running out of available swaps. And that’s pretty much all there is. It’s called Simple Match for a reason people.
Since we are worried about level-based content cadence, we can focus on level attempt times instead of swap times and swaps per level .
“But where am I getting this data?” you might ask. If your game is live, aggregate your user data through a query. If your game hasn’t launched yet have no fear, remember this is back of the napkin math to find major flaws in Simple Match’s content runway at launch and long-term content cadence, this isn’t fuel calculations for a mission to Mars. Simply have players of different skill levels play available levels multiple times and take the averages.
So now we have the base interaction:
|Average Level Attempt||2 minutes 39 seconds|
Now let’s model our average player. For modeling the game, we started with the lowest level interaction and worked our way upward. To model our average player, I find it’s easier to work backwards and go from highest level down to the lowest level. How high level do you think about your game? Outside of MMORPGs, or games with heavy monetization through monthly subscriptions, I’d argue most games are better served thinking about their engagement in terms of weeks. Match-3’s, like Simple Match, roll out content on a weekly or bi-weekly cadence, have event schedules on a weekly schedule, and see engagement patterns that repeat on a weekly basis so it makes perfect sense to think about your player engagement on a weekly basis.
For Simple Match, we want to know what the average number of days per week a player plays. For any day, we want to know the average number of sessions a player has on each of those days. Finally, we want to know how long, on average, those sessions last for our Simple Match players.
Base player model:
|Days per Week||3.6 days|
|Sessions per Day||1.6|
|Avg Session Length||8 minutes 27 seconds|
Here is a graph of the cumulative hours our average players will put into Simple Match, pretending the game is launching on January 1st and measuring through the end of April.
With that, you can get important things like level attempts per day or week. Level attempts per day is calculated by simply dividing the average session length by the average level attempt (for this calculation I floored the result to get whole level attempts). Multiply level attempts per day by days per week to get the level attempts per week. However level attempts does not equal level progression, we need to take into consideration Simple Match’s average win-rate, 80%, to get the overall level progression our average player will make off of those level attempts.
|Level Attempts per Day||4|
|Level Attempts per Week||14|
|Level Wins per Week||11|
On the flip side, Simple Match only has so many levels available to be played at a given time. A general content cadence, or how often new content is released, is every 2-weeks for the Match-3 genre. Let’s see what our average player looks like if we launch Simple Match with 50 levels available on the launch-day and release 10 levels every 2-weeks.
|Levels at Launch||50|
|Content Release Interval||2-weeks|
|New Content Released||10|
It’s probably not a good idea to have our average player run out of content and continue to be out of content as the game goes on. Remember this is only the average player, so you will definitely have a large population of more highly engaged players that run out of content even faster. We could increase the levels at launch, lower the interval between content releases or up the levels per content release. This is where manipulating the content burn map, while considering your liveops production cost, can answer critical content cadence questions. For Simple match, the easiest solution is to increase the levels per content release. Here is what our Out of Content graph looks like now.
|New Content Released||30|
Now that your average player looks well… content, what about other players? Here is where player cohorts can be extremely useful. If you lined up all your players based on an engagement metric, what would your 25th percentile player look like? What about your 75th percentile player? What does your most dedicated player look like?
Looking at Simple Match’s players, the 75th percentile player will straddle end of content while the 99th percentile player will engage so hard they can’t be satisfied (the green line is hidden behind the red line). Here is where other goals such as target audience and monetization strategy come into play to determine whether Simple Match has successfully proved out its content runway.
What does it all mean?
You can see how a content burn map can be used to help gauge how much content Simple Match needed at launch and what content cadence is needed to support different levels of player engagement. What else can we do with a content burn map?
This tool can be applied to a lot more than levels in a Match-3:
- In an RPG, what level will the player’s characters be at after 40 hours of gameplay?
- What is the fastest car unlocked by any player during the launch week of a Racing title?
- What will the economic inflation will be in an MMORPG after 2 years?
- If a player plays an FPS for 2 hours every day, how long will it take for them to unlock every rare skin?
- What can a player accomplish during an average game session?
Content burn maps can easily be extended to forecast Product Manager concerns like revenue and player churn.
The real power of this tool is from the ability to manipulate the models and weigh production cost against potential upsides across a number of game performance spectrums. I believe each of your games can and will benefit from this effective pre-production tool.