Publishers have spent years stuffing bid requests with more signals without knowing which ones actually move the needle.
Amazon Publisher Services wants to change that.
At its annual summit on Thursday, APS unveiled an expansion of Signal IQ, its bidstream testing tool first launched in 2024, that measures how signals passed through OpenRTB requests influence advertiser demand and publisher revenue.
The Signal IQ expansion pushes APS beyond evaluating the impact of alternative audience ID signals like LiveRamp’s RampID or The Trade Desk’s UID2 that are meant to replace third-party cookies. It now weighs broader bidstream data about the ad impressions themselves, including global placement IDs (GPIDs), transaction IDs and video classification parameters.
“The biggest impact comes from signals that improve bidder confidence, and we’ve found that extends way beyond ID signals,” Scott Siegler, director at Amazon Publisher Services, told AdExchanger. “Some of the signals we see ‘pop’ are contextual, specific to the device or even the genre of streaming TV content.”
To complement the Signal IQ update, APS also introduced new AI-powered optimization tools, expanded its commerce-driven Shopping Insights deals and added new outcome-based planning capabilities tied to advertiser consideration metrics.
What signals really matter
APS argues publishers still lack visibility into which bidstream signals actually influence DSP bidding behavior.
Many publishers continue investing in identity tools and supply-path optimization strategies without clear insight into whether buyers actually value those signals and factor them into their bidding decisions. Plus, even if publishers have confidence that a given signal is prized by buyers in general, they don’t usually have transparency into what motivates individual buyers.
“What Signal IQ shows us is that different bidders respond differently to different signals,” Siegler said. “A signal that drives significant lift with one bidder may have minimal impact with another, because each bidder has its own models and priorities for how it values inventory.”
In addition to providing better insights at the individual buyer level, Signal IQ also introduced a new signal coverage report for publishers that benchmarks how consistently they pass key OpenRTB signals compared to their peers.
Later this summer, APS plans to bring its expanded signal reporting capabilities to Signal IQ’s A/B testing framework, allowing publishers to estimate the projected revenue lift associated with specific signal investments.
Meanwhile, Signal IQ’s addition of GPID support is a response to buyers pushing for greater transparency around placement-level data and inventory quality. GPIDs are universal identifiers for individual pieces of ad inventory, and passing them in bid requests lets buyers know exactly what specific inventory they’re buying. Now, publishers will be able to tell if providing that data actually translates to higher revenue yield.
AI revenue optimization
Nailing down which signals yield higher revenue is key for publishers. Rather than simply encouraging publishers to pass more data into auctions, APS said it wants to quantify which signals actually improve bidder participation and monetization outcomes.
APS is also trying to provide more context around why certain signals matter to buyers.
“We are working on the ‘why,’ though it often requires additional context only available to the bidders,” Siegler said. “Teams are regularly speaking with bidding partners, and we use that market insight together with Signal IQ insights to help publishers understand what the results mean and how to act on them.”
APS also announced that it’s further embedding AI into its revenue optimization workflows.
Its new Publisher Supply AI assistant, launching in open beta later this summer, offers a chatbot interface for analyzing bidstream and performance data. The assistant allows publishers to troubleshoot issues and surface optimization recommendations without manually pulling reports.
The company also plans to introduce AI-powered monitoring agents that proactively flag changes in publishers’ ad performance and operational issues.
Siegler said APS intentionally focused its recent AI efforts on operational and data workflows rather than directly influencing auction mechanics.
“We see AI becoming a powerful co-pilot for publisher monetization teams because of the scale and complexity of the data they manage,” Siegler said. “The ideal model isn’t to automate everything; it is to surface properly governed agents that can make recommendations, explain impact and allow teams to approve, test or set guardrails.”
Outcome-based buying
APS also unveiled new product updates with a broader focus on helping publishers package inventory around purchase intent signals and measurable advertiser outcomes.
For example, it released a new Mobile SDK bridge designed to help third-party bidders compete more effectively for publishers’ app inventory by enabling shared auction transparency signals between the buy and sell sides.
APS is also expanding its Shopping Insights-powered campaigns. The offering enriches publisher inventory with Amazon shopping and browsing data to help advertisers target audiences based on purchase intent in Amazon DSP. Now, Shopping Insights can enrich web and mobile inventory, in addition to its initial focus on CTV inventory.
At the same time, the Amazon Publisher Cloud clean room platform will begin supporting audience matching for consideration-based campaign objectives, such as branded searches and detailed page views.
“Advertisers are looking for audiences with real shopping intent,” Siegler said, “and publishers are looking for ways to make those signals easier to package and activate across formats.”
