Discounted Bundle Design -- A Design Document Example
Baseline game
Call of Duty Black Ops Cold War
Design goal
Encouraging players to do more in-game purchases with personalized discounted Store Bundles.
Problems need to be solved
How often does the bundle show up; How long does the discount last; What might be in the bundle; How much is the discounted price; How does the bundle work with current store; How to prevent players from reverse engineering the system.
Overview
The forming and pricing of the recurrent discounted bundle takes the player’s most recent playing and purchasing behavior into consideration to maximize the sales.
Time limits
• The discounted bundle lasts for one week.
For most people one week is a complete cycle of working/ studying and playing. We don’t want players to have too many discounted bundles at one time.
• A new discounted bundle is provided to the player every 100 levels.
The leveling usually takes 20-30 hours, and that is longer than weekly playtime for most players. For example, there is a limit-time bundle adding to player’s store when player’s level reach 100, 200, 300 etc.
• More than one discounted bundle might coexist in the store, if some players play a lot and reach two 100 level milestones in a week.
Contents:
• Every discounted bundle includes two weapon blueprints, two to four other accessories.
• Items in the bundle chosen from all single items (weapon blueprints, calling cards, emblems, charms, etc) in existing bundles excluding the following:
o Items player has already owned.
o Items in the player’s current store.
o Items appeared in last 3 discounted bundle of the player.
o Items from the bundles that recently released.
Bundle scores:
• Every bundle has a bundle score estimating how desirable the bundle is to specific player, positively related to the rate of discount, sales of every single item in the bundle.
• High coefficients are applied on the blueprints of the weapons we think the player would be interested in purchasing ( for example, weapon the player used most in the latest 30 games, weapon the player had the best K/D ratio in the latest 200 games
among all weapons the player regularly uses).
Target scores and bundle generation:
• A target score is set up for every player, describing the minimum bundle score the player might want to buy.
• Target score decides the desirable level of bundles the shop is providing. A bundle is generated by program to have a bundle score close to the target score of the player by selecting contents and discount level. An upper limit is applied to the discount rate.
• The target score can be reached only if the bundle has two of the weapon blueprints that are eligible for the love-to-use high coefficients I described above. When there are multiple suitable combinations, a random combination will be selected.
• The target score will start from a default value. If the player does not buy the discounted bundle, the target score will slightly go up, making the player more likely
to get more desirable items or better price next time the discounted bundle appears. If the player buys the discounted bundle this time, the target score will go down.
Price Sensitivity:
• Price sensitivity is a value set up for player describing their purchasing habits. Higher price sensitivity means the player cares more about the price.
• If there is not any kind of purchase activity in 2 months, the sensitivity will increase. Every purchase will decrease price sensitivity.
• When we need to increase the target score of a player. Player with higher price sensitivity is more likely to get a better price while player with lower price sensitivity is more likely to get more valuable items. When we need to decrease the target score, player with high sensitivity is more likely to get less favorable items while players with low sensitivity is more likely to get higher price.
Coordination with current store:
• The personalized discounted bundle might include items from existing bundles. When a player tries to buy a bundle that he/she already owned part of it, corresponded price would be deducted from the total price.
About Reverse engineering:
• To prevent Reverse engineering, following designs can be used:
o Adding an upper limit to the target score.
o Adding a random term to the target score.
