Auction Determinants & Quality Score
When Criteo's Engine tries to determine what ads to show for a given impression opportunity, it looks at a few elements to optimize for both retailer yield & campaign performance. Generally, these can be broken down into 3 buckets: contextual information, campaign information & user information.
Contextual information takes into account information about the page where the impression opportunity would appear (category, search terms, etc.).
Campaign information examines the past performance of the products under the line item and looks at the types of users, page categories, and other features where the product has gotten the most engagements.
User information looks at user browsing behavior, what they searched for, what they clicked on, and what they eventually bought. Criteo then takes that information and uses it to help inform which brands and sponsored listings should appear when a shopper searches using a specific term. Criteo only uses premium 1st party retailer data.
For auctions (SP), contextual & campaign information ends up driving most of our predictions, given we have less room for user-level optimization. For offsite, user information is much more important and plays a larger role in determining ad placements
Quality score is the prediction output that determines the likelihood of a user to click on a particular product ad (pCTR, stands for predicted Click Through Rate). Quality Score is not something that an advertiser can control or influence. It stems from past campaign performance, brand name recognition, product performance, and user information.