The Game Life-Cycle and Game Analytics: What metrics matter when? Data Science Day Berlin Mark Gazecki (Chairman)
Introduction HoneyTracks: Web-based game analytics solution Deep analytical capability Cohort analysis, funnels, A/B-testing For all types of games Social games, browser-games, client games, mobile games Real-time Custom metrics & funnels Easy-to-use graphical interface Avoiding data-graveyards (happens if people can t use it) For everyone in the company Information at everyone s fingertips: Game design, product mgmt, marketing, management, 2
The 5 most important metrics The never-ending quest for the most important 5 metrics 3
The 5 most important metrics The never-ending quest for the most important 5 metrics... is indeed a never-ending quest 4
The 5 most important metrics there is no such thing as the universal 5 most important metrics Games are unique & different Games have a life-cycle To generate actionable insight differences in each game must be considered. This has an implication for the metrics you want to monitor. What is important changes over the life-time of a game. This must be reflected in the metrics / KPIs 5
Moore s lifecycle adoption model applied to games Prototypical Technology Product Lifecycle (taken from Crossing the Chasm ) Growth Maturity & Revenues Like any other technology-product, games have a product lifecycle (may be more or less pronounced for certain game-types and individual games) First focus is on growth then on managing maturity and maximizing revenues 6
Your re launching your game: Virality vs retention What would you rather have? Double the virality? Half the churn-rate? 7
Why retention comes first Number of active users (conceptual) 3000 Assumptions Viral game Ret. game 2500 Viral game Viral invites / user 2.5 1.25 2000 Churn-rate 80% 40% 1500 1000 500 Game with better retention 0 Month 1 Month 2 Month 3 Month 4 Month 5 Month 6 Month 7 Month 8 Month 9 Game with better retention has higher number of average monthly users No retention no sustainable growth no hit and since users tend to monetize better as they progress in the game, higher retention lays the basis for strong monetization 8
Game life-cycle KPI framework Game Life-Cycle (time / age of game) User acquisition Retention Monetization Bring initial users into the game (x-promotion, limited launch ) Virality Engagement metrics Acquisition & virality metrics Monetization metrics Start out by making sure that retention is good enough with an initial flow of users, i.e. not all users you acquire churn out immediately Then move onto optimizing user acquisistion, virality, and monetization but of course this is an additive view!!! 9
Retention metrics: What to start with Retention / Engagement Metrics 1-7 day retention Tutorial steps funnel Drop-off rates (by level) Visits / DAU Session times Optimize tutorial (to get users effectively into the game) A/B-test user funnels Optimize user drop-off events (make it less difficult, more fun, ) Give user more / less stuff to do / more energy (-> session length. engagement) Track feature-usages (also for mid- / end-game) A/B-test game mechanics (esp. mid- / end-game) Churn-rate (monthly) 1 / monthly churn-rate = Player lifetime (in months) 10
Acquisition metrics: What to start with Acquisition Metrics Conversion rates (CTR) User acquisition cost (CPC, CPI / PAC) Metrics by marketing channel / ad (cohort analysis) Metrics by demographics (cohort analysis) Metrics by geography (cohort analysis) Test different marketing channels A/B-test different creatives A/B-test different targeting (demographics, geographies) Monitor PLTV > PAC (for channel cohorts, demographies etc) Metrics by user source (e.g. player lifetime value) (ads, viral, x-promotion) Start tracking monetization metrics by user cohorts early on (channels, demography, ) 11
Example: Marketing channels Screenshot: Channel profitability Segmenting users by marketing channel...... shows that Channel 1 has 50% of Channel 32 revenues despite having 2.5x in DAU 1 32 Marketing Channel 1 2 3 4 5 6 7 8 9 10 11 32 33 Marketing Channel We could improve the game: Focus on user aquisition from ch32 Double check payment type (SMS) and charge backs in ch1 Switch off certain payment methods at special times Comparing payouts to revenues shows that Channel 1 has more lost revenue, i.e. issues in the payment process Marketing Channel 1 32 12
Virality metrics: What to start with Virality Metrics k-factor Number of sent invites / DAU Acceptance rate (by type of invite) % of virally acquired users (last 30 days cohort) A/B-test content for viral message (how should buttons look, images, etc) A/B-test different viral triggers (in the game) A/B-test different acceptance mechanisms Number of viral users by viral source 13
Monetization metrics: What to start with Monetization Metrics ARPU ARPPU Payment conversion rate Avg. transaction value First purchase trigger A/B-test alternatives to improve first-time buyer conversion (e.g. specials, variants of that particular virtual good) Optimize user-flow towards first purchase trigger (-> get more users there) A/B-test different virtual goods & packages Optimize payment process (conversion steps) A/B-test pricing Paying user cohort (by marketing channel, by geography) Player life-time value (PLTV) 14
Custom Metrics Game life-cycle KPI framework: Introducing custom-metrics User acquisition Retention Monetization Virality Standard metrics Custom metrics Standard metrics are great for detecting issues on a high level To derive actionable insight need to drill deeper and look at custom metrics 15
Custom Metrics Drill-down capability & custom metrics to derive actionable insight Peeling the onion Observe slight decrease in aggregate ARPU in month of July Payment conversion rate is decreasing ARPPU remains constant Payment conversion for existing users stays constant Payment conversion for the user cohort acquired in June is very low Users acquired in June from marketing channel SuperDuperAds have a significantly lower conversion rate Mix of users in June shifted towards countries with generally lower conversion rates The pricing for a virtual good, which typically was the first virtual good purchased by users, was changed 16
Example: ARPU cohort analysis Screenshot: ARPU cohort analysis Aggregate ARPU is 2 Euro... and we see that ARPU improved from April to May cohort Monthly cohorts show that ARPU actually becomes 4 Euro! Aggregate numbers don t tell the truth As a next step we would dig deeper into the May-cohort to understand why it generated better ARPU 17
Game Analytics Examples Peel the onion : Payment conversion (1) Screenshot: Revenue analysis by level Majority of revenues achieved in levels 20-30 Pretty effective at monetizing advanced users... 0 10 20 30 40 What are virtual goods that are useful at earlier levels? 0 10 20 30 40 50 Can we push users into making purchases earlier?... but what about users in earlier levels? 0 10 20 30 40 50 18
Game Analytics Examples Peel the onion : Payment conversion (2) At lower levels food is being purchased relatively higher... so this may be the virtual good, which converts users into first time buyers... even though food doesn t play a major role in revenues We could improve the game Offering food specials to users at lower levels Try lower prices for food to generate more first time buyers 19
Game Analytics Examples Peel the onion : Whales Analysis Who are my whales? What is her profile We could improve the game What, when and how much of each item works best for her? See what works best for whales and offer higher variety of same type Increase prices step-wise for new items and monitor closely Try out special offers for items that work for other whales Optimize payment options What payment does she use Different colors indicate different feature/item types - mouse over shows details 20
Game Analytics & Game Life-Cycle How to approach it right Start with retention metrics. Then move to user acquisition-, virality-, and monetization metrics. Start with standard metrics. Then move to custom metrics to generate actionable insight Peel the onion to derive actionable insight (cohort analysis etc) Understand it is an ongoing effort, which involves multiple functions / departments in your company (not all which are tech-people) Make sure you have the right game analytics system (it should support all of the above) 21
Game Analytics & Game Life-Cycle Read more! Casual Connect Magazine (summer 2012) 22
Contact information Want to see HoneyTracks in action? Check out: www.honeytracks.com @HoneyTracks Mark Gazecki mg@honeytracks.com 23