Ad bids

Meta’s patents describe an ad bid as contributing to the compensation a social networking system expects when an ad is displayed, receives an interaction, or meets another specified condition.23 In Meta’s proposed scrollable format, advertiser bids and people’s past interactions with content in such units both help determine the order of the sponsored groups.23 Meta’s patents also describe choosing and ordering the individual sponsored posts within those groups, so comparing groups and arranging their contents are distinct parts of the proposal.23

Where the scrollable-feed patents describe using bids, Meta’s October 2018 patent proposes a way for merchants to specify them for different audience segments—groups of people divided by selected criteria.234 Meta’s patent lists criteria including age, gender, location, time of day, preferences, expected budget, loyalty and affiliations.4 The dashboard proposed in Meta’s patent can break a relevance score—a measure of how closely content matches a person—into individual values that merchants can use when setting bids.4 In one example, Meta’s patent requires a higher price for a relevance score above 0.7.4

Meta’s March 2022 patent describes calculating an initial presentation value partly from a bid associated with a candidate content item.5 In Meta’s patented approach, the system communicates its initial estimate to the person providing the content.5 Meta’s patent then describes receiving information from the viewer’s device about where candidate content items actually appeared in the feed.5 Using that placement information, Meta’s patented approach calculates the detrimental effect on the viewer’s engagement and adjusts the presentation value accordingly.5

A different reason for adjusting bid-based comparisons appears in Meta’s June 2023 research: making the demographics of people who see ads more closely match those of the audience allowed by the advertiser’s targeting, while protecting individual privacy.1 Meta’s research describes changing an ad’s position in an auction—the competition among ads for a placement—through a comparison score, rather than changing the advertiser’s monetary bid.1 Meta’s research calculates a multiplier, a number used to scale the bid-derived part of that score, while accounting for a separate quality-related part, so the result reaches the highest score among competing ads.1 Meta’s research also describes using the lowest competing score to move an ad to the bottom, with the comparison including both housing and non-housing ads in the same auction.1