liveBook · Manning

Ecpm

As explained in Machine Learning System Design

A revenue metric normalized to a quantity of 1,000 interaction opportunities, calculated in this section as revenue per click multiplied by 1,000.

eCPM is revenue normalized to 1,000 interaction opportunities. Here, calculate it as revenue per click multiplied by 1,000. In ad recommendation, it can serve as the reward an algorithm seeks to maximize when choosing among banners. section 12.1.3

Ad-banner selection

A recommender can choose an ad banner using user context and optimize eCPM, alongside alternative objectives such as click-through rate or conversion rate. section 12.1.3

How Machine Learning System Design covers it

How deep it goes
Defines eCPM as revenue per click multiplied by 1,000, but does not develop its calculation beyond that formula.
Examples
Uses ad-banner selection as the concrete setting, alongside CTR and CVR as alternative reward objectives.
What you can do after
Evaluate/understand ad-selection rewards using revenue normalized to 1,000 interaction opportunities.
How it connects
Connects eCPM to ad recommendation, click-through rate, conversion rate, contextual bandits, and reinforcement learning.
Cover of Machine Learning System Design
liveBookmanning.com
Read Machine Learning System Design on liveBook Get the book at manning.com

This explainer was generated by Manning's knowledge graph from the text of Machine Learning System Design. The section links point to the parts of the book it draws on.