Machine learning for digital advertising
2 likes5,250 views
The document discusses the role of machine learning in digital advertising, focusing on how e-commerce businesses can optimize their marketing strategies to increase profits. It explores various aspects such as user data, ad bidding, conversion rate optimization, and challenges faced with acquiring and analyzing data. Practical techniques like gradient boosting trees and active learning are mentioned as effective for improving advertising outcomes.
Machine learning for digital advertising
- 1. Machine Learning for Digital Advertising Marc Garcia PyData London - May 8th, 2016 1 / 41 Machine Learning for Digital Advertising
- 2. What is the Internet? 2 / 41 Machine Learning for Digital Advertising
- 3. What is the Internet? A lot of bored people... 3 / 41 Machine Learning for Digital Advertising
- 4. What is the Internet? ...who use a gadget... 4 / 41 Machine Learning for Digital Advertising
- 5. What is the Internet? ...to access useless sites. 5 / 41 Machine Learning for Digital Advertising
- 6. What is our goal? 6 / 41 Machine Learning for Digital Advertising
- 7. What is our goal? Get their money! 7 / 41 Machine Learning for Digital Advertising
- 8. What is our goal? That’s the truth, but let’s say it in a more subtle way... 8 / 41 Machine Learning for Digital Advertising
- 9. What is our goal? We want to attract them to our site... 9 / 41 Machine Learning for Digital Advertising
- 10. What is our goal? ...and sell our product. 10 / 41 Machine Learning for Digital Advertising
- 11. How does it work? 11 / 41 Machine Learning for Digital Advertising
- 12. How does it work? The user journey 12 / 41 Machine Learning for Digital Advertising
- 13. How does it work? The user journey 13 / 41 Machine Learning for Digital Advertising
- 14. How does it work? The user journey 14 / 41 Machine Learning for Digital Advertising
- 15. How does it work? The user journey 15 / 41 Machine Learning for Digital Advertising
- 16. How does it work? The user journey 16 / 41 Machine Learning for Digital Advertising
- 17. Ad exchange point of view 17 / 41 Machine Learning for Digital Advertising
- 18. Ad exchange point of view How the ad exchange works? A publisher has an available ad space. 18 / 41 Machine Learning for Digital Advertising
- 19. Ad exchange point of view How the ad exchange works? Advertisers bid for their ads on the cost for a click. bid: £0.35 bid: £0.20 19 / 41 Machine Learning for Digital Advertising
- 20. Ad exchange point of view How the ad exchange make money? income = bid · clicks impressions (1) Impressions are constant regardless of the ad Bids are known for each ad Clicks are unknown, but can be predicted 20 / 41 Machine Learning for Digital Advertising
- 21. Ad exchange point of view What is the best option? bid: £0.35 expected clicks (for 1000 impressions): 25 expected profit (for 1000 impressions): £8.75 bid: £0.20 expected clicks (for 1000 impressions): 47 expected profit (for 1000 impressions): £9.40 21 / 41 Machine Learning for Digital Advertising
- 22. E-commerce point of view 22 / 41 Machine Learning for Digital Advertising
- 23. E-commerce point of view T-shirt e-commerce example 23 / 41 Machine Learning for Digital Advertising
- 24. E-commerce point of view Income statement1 Sales £23.89 100% Cost of goods sold £11.72 60% Delivery £4.36 18.25% Payment gateway £0.84 3.5% Total direct costs £19.53 81.75% Available for marketing £4.36 18.25% Average cost-per-click £0.20 Conversion rate (break-even) 22 1 Excludes many costs (product returns, freight-in, development, support, hardware, salaries, etc.) 24 / 41 Machine Learning for Digital Advertising
- 25. E-commerce point of view Not so easy to make money... 25 / 41 Machine Learning for Digital Advertising
- 26. An optimization problem 26 / 41 Machine Learning for Digital Advertising
- 27. An optimization problem What can be optimized? Higher prices (added value, branding...) Lower production costs Leverage marketing costs Virality (share on social networks...) Recurrence (newsletter to customers...) Conversion rate optimization Targeting / market segmentation Ad optimization Website optimization (A/B testing...) 27 / 41 Machine Learning for Digital Advertising
- 28. An optimization problem Conversion rate optimization: the numbers income = average_price · margin · num_sales − average_cpc · num_clicks (2) 28 / 41 Machine Learning for Digital Advertising
- 29. An optimization problem Conversion rate optimization: Features 29 / 41 Machine Learning for Digital Advertising
- 30. An optimization problem Features: Users What Facebook, Google... know about our users? Age Gender Location Interests / Pages visited Friends / Interactions Sociocultural status ... mostly everything What do we know about them? Not much individually We can access aggregated data 30 / 41 Machine Learning for Digital Advertising
- 31. An optimization problem Features: Ads 31 / 41 Machine Learning for Digital Advertising
- 32. An optimization problem Features: Website 32 / 41 Machine Learning for Digital Advertising
- 33. An optimization problem Features: Summary Users Age Gender Location Interests Friends Sociocultural status ... Ads Source Message Type (image, text,...) Placement Site ... Website Layout Colours, font size... Content density Offers, reviews... Conditions (delivery, returns) Steps to purchase ... Other User behaviour Day of week Time Weather Social environment ... 33 / 41 Machine Learning for Digital Advertising
- 34. An optimization problem An amazing dataset Age 28 Gender female Location Lewisham Occupation receptionist Interests hiking, salsa Education university Device smartphone Source Google search Placement top Ad type image Message type sales offer Web layout basic Web main colour green Conditions free delivery Returns free Product type shoes Quality high Price £79 Day business day Time evening Purchase true 34 / 41 Machine Learning for Digital Advertising
- 35. An optimization problem What could we get from a dataset like this? Women aged 25 to 35 buy shoes on rainy Sundays Users leave pages with more than 800 characters on the layout quite fast People from city A buy more product X, while people from city B, prefer product Y Ads with red colours have more clicks, but we get less conversions from them And all this means higher profits. 35 / 41 Machine Learning for Digital Advertising
- 36. An optimization problem Amazed? 36 / 41 Machine Learning for Digital Advertising
- 37. An optimization problem Some problems Data is expensive (average of £0.20 per row) :( No public data So, we need to obtain it all ourselves... ...and it’s expensive :( Unbalanced / sparse data (1 purchase for 100 clicks) So, we need 100x the data... ...and it’s expensive :( High (or very high) dimensionality So, we need a huge number of rows to avoid the curse of dimensionality ...and it’s expensive :( 37 / 41 Machine Learning for Digital Advertising
- 38. An optimization problem More problems We bias the data when we know who is more likely to buy Exploration an exploitation problem We have to buy "bad" clicks to avoid it Many categorical variables They don’t play well with many models Data seasonality Black Friday, sales, Christmas... Access to data Google, Facebook do not give access to raw data We have to use aggregated data And they do not let you cross variables We need to assume independence among them 38 / 41 Machine Learning for Digital Advertising
- 39. An optimization problem He is very sad now 39 / 41 Machine Learning for Digital Advertising
- 40. An optimization problem What models can be used? Some techniques that gave good results: Gradient boosting trees Active learning (semisupervised learning) Online learning (for data freshness) 40 / 41 Machine Learning for Digital Advertising
- 41. An optimization problem Questions? @datapythonista 41 / 41 Machine Learning for Digital Advertising