Gleb Mashchenko, RoasUp

Mathematical Approach to Advertising on Facebook

(White Nights Conference Moscow 2018)
The official conference website — http://wnconf.com

Transcript

1. Mathematical Approach to Advertising on Facebook Gleb Mashchenko Business Development Director, RoasUp, Inc. gleb@roasup.com
2. ROASUP is a performance marketing agency for mobile games focused on user acquisition from Facebook gleb@roasup.com 2 Our Clients
3. Facebook is a Powerful UA Platform for Mobile Games gleb@roasup.com 3 * The AppsFlyer Perfomance Index - Edition VII, H1 2018 https://www.appsflyer.com/2018indexpage/ Power Ranking*
4. Facebook Ad Account Structure gleb@roasup.com 4 Facebook Account Ad Campaign Ad Set Ad Set Ad Set Ad Ad Ad Ad Campaign
5. Ad Set Parameters gleb@roasup.com 5 to show WHAT to show WHERE to optimize HOW Creative Headline Text CTA Audience Placement Device Optimization Conv. window CBO BUDGET Now you have to set 25+ parameters for an ad set!
6. Conversion Funnel and ROAS gleb@roasup.com 6 Impressions Clicks Installs Payers 10000 100 50 3 $$$ Spend $$$ Revenue ROAS = Revenue Spend
7. ROASUP Data-driven Mathematical Approach gleb@roasup.com 7 • Collect data from Facebook • Automated data analysis
8. How Do Ad Set Parameters Influence ROAS? gleb@roasup.com 8 # Parameter H* 1 Creative 847 2 Optimization 199 3 Optimization conversion window 54 4 Location 40 5 Gender 28 * Kruskal W. H., Wallis W. A. Use of ranks in one-criterion variance analysis. // Journal of the American Statistical Association. — 1952, 47 № 260. — pp. 583–621. Asymptotic significance (p << 0.001)
9. Creative Performance gleb@roasup.com 9 Most of creatives do not receive enough spent for valid A/B tests. Facebook automatically optimizes creatives in an ad set
10. Creative Performance gleb@roasup.com 10 Performance of creatives dynamically changes
11. Creative Performance gleb@roasup.com 11 ROAS is not distributed normally among launches of any creative Standard methods of classic statistics are not applicable! K-S d=0.27405, p<0.01, Shapiro-Wilk W=0.59, p=0.0000
12. Creative Performance gleb@roasup.com 12 • ROAS distribution estimation • Bayesian Statistics • Automated analysis of creatives performance dynamics ROASUP Approach
13. Optimization and Conv. window Performance gleb@roasup.com 13 LTV of users from different optimizations varies LTV Days
14. Optimization and Conv. window Performance gleb@roasup.com 14 LTV is high, but CPI is high too! Cohort ROAS Days
15. Optimization and Conv. window Performance gleb@roasup.com 15 CPI ROAS CPI and ROAS depend on optimization and targeting High CPI is not an issue if ROAS is high!
16. Optimization and Conv. window Performance gleb@roasup.com 16 • Analysis of cohort ROAS for each optimization and targeting • Maximization of ROAS through selection of optimization and conversion window for each targeting • Performance dynamics tracking ROASUP Approach
17. Ad Set Control Problem gleb@roasup.com 17 Ad set dynamics is described by non-stationary stochastic process
18. Ad Set Control Problem gleb@roasup.com 18 Lifetime of the most of ad sets is less than 7 days
19. Ad Set Control Problem gleb@roasup.com 19 N of Ad Sets N of Ad Sets Opt. day Opt. day Data imbalance: +90% of ad sets do not meet KPI The first day is optimal for turning off for the most of ad sets
20. Ad Set Control Problem gleb@roasup.com 20 Simple methods of machine learning lead to the issues: • Unstable classification of ad sets (classes imbalance, sparse data) • Complex dynamic of an ad set depends on the fast changing market and changing performance of creatives • Low amount of data
21. Ad Set Control Problem gleb@roasup.com 21 • Deep Reinforcement Learning • Generative models (GAN, VAE) • Autoencoders ROASUP Approach
22. Conclusion gleb@roasup.com 22 Efficient user-acquisition on Facebook requires automated mathematical analysis
23. gleb@roasup.com 23 Thank you! Gleb Mashchenko Business Development Director, RoasUp, Inc. Email: gleb@roasup.com Facebook: gleb.mashchenko Skype: gleb.mashenko Telegram: @mgleb