Overview
This was an assignment from Scopely for the economy product manager role. It involved analyzing and retuning card pack drop rates and rarity structure for a card album that ran for a 50 day event.
- I used Python to generate 10k player samples since I needed to track card duplicates for each player over the event.
- I uploaded that dataset to Google BigQuery since I find it easy to upload to and to work with.
- I ran SQL queries to analyze the dataset, finding that few players complete the album but they get hundreds of duplicates—which would reduce monetization, retention, and player satisfaction. Most interestingly, the rarer the set, the more likely players are to complete it, which is the opposite of how that should be tuned.
- After finding the weaknesses in monetization design, I made recommendations to adjust the tuning which would result in rarer albums being harder to complete and reducing the number of duplicates.
- I reran the simulation with my new tuning numbers to provide estimates for exactly how many players would complete each set and how many duplicates they would get. This could help inform exactly the impact on monetization and retention that the retuning would provide.
My Analysis & Suggestions
Assignment (for reference)
- Model player card collection during a 50 day event.
- Track how players are receiving cards of different rarities, and analyze how they are distributed.
- Analyze and provide recommendations.
Current Tuning
- Common Packs:
- 100 cards
- 4 packs earned per day
- 2 cards per pack (70% chance of common, 25% chance of uncommon, 5% chance of rare cards)
- Uncommon Packs:
- 60 cards
- 2 packs earned per day
- 3 cards per pack (55% chance of common, 35% chance of uncommon, 10% chance of rare cards)
- Rare Packs:
- 25 cards
- 1 pack earned per day
- 4 cards per pack (30% chance of common, 40% chance of uncommon, 30% chance of rare cards)