Protected Audience 입찰은 시각적인 방법 또는 Perfetto를 사용한 SQL 쿼리 중 하나로 분석할 수 있습니다. 광고 기술은 Perfetto를 통한 프로파일링 방식을 사용하여 다음과 같은 Protected Audience 입찰의 실적을 측정할 수 있습니다.
입찰 및 점수 스크립트 CPU 시간
HTTP 요청의 지연 시간(예: 키/값 서비스)
콜드 캐시와 핫 캐시의 성능 영향 비교
더 많은 맞춤 잠재고객과 더 적은 맞춤 잠재고객 비교
더 큰 신호 집합과 더 작은 신호 집합
모든 입찰에 동일한 스크립트를 사용하는 것과 맞춤 잠재고객별로 다른 입찰 로직 스크립트를 사용하는 것 비교
[null,null,["최종 업데이트: 2024-09-03(UTC)"],[[["Perfetto can be used to analyze the performance of Protected Audience auctions, including bidding and scoring script CPU time, latency of HTTP requests, and the impact of various factors like cache and custom audiences."],["To analyze auctions with Perfetto, developers need to set up a test environment, record a trace during an auction using the provided trace configuration file, and then open the trace in the Perfetto UI."],["The Perfetto UI allows visual exploration of traces, highlighting details like execution latency for different auction stages, such as bidding, scoring, and ad selection."],["Specific trace segments within Perfetto provide insights into various aspects of the auction process, such as filtering contextual ads, loading buyer and seller signals, running bidding and scoring logic, and persisting auction results."],["Perfetto also supports SQL queries for precise measurements, allowing developers to calculate metrics like average execution latency for specific operations, like fetching JavaScript payloads."]]],["Profiling Protected Audience auctions with Perfetto allows ad techs to measure bidding/scoring script CPU time, HTTP request latency, and the impact of cache, audience size, and signal sets. To analyze, clone Perfetto and Privacy Sandbox repos, and build the sample app. Run an auction on a test device via adb, initiating a Perfetto trace with `trace_config.textproto`. The trace data is visually explorable or can be queried using SQL to determine latency of segments. Key trace segments include bidding, scoring, and auction processes.\n"]]