Yahoo подключила Amazon Bedrock для поискового ретаргетинга в рекламной платформе DSP

Yahoo has connected Amazon Bedrock to the Search Retargeting (SRT) system within its Yahoo DSP advertising platform. This was reported by the AWS Machine Learning Blog: the integration expands advertisers' reach through AI-driven analysis of search intent — both on Yahoo's own properties and through partner networks.

What is Search Retargeting SRT

SRT is a core audience targeting tool in Yahoo DSP that allows advertisers to reach users based on their search query history. The technology links search intent to display, video, and native advertising: if a user searched for "laptops" on Yahoo Search, the advertiser can show them an ad in a completely different context — on a partner site or within a video.

Before the Amazon Bedrock integration, SRT worked primarily with keywords that users typed directly into the Yahoo Search bar. The updated version goes beyond that limitation: AI models identify search intent not only on Yahoo's own properties, but also through connected partner systems, capturing a broader spectrum of user signals.

Why Yahoo Chose Amazon Bedrock

Amazon Bedrock is a managed AWS service that provides access to foundational language models from Anthropic, Meta, Mistral, and other providers through a single API. Its key advantage for enterprise adtech is the ability to deploy powerful LLMs without needing to build proprietary training infrastructure or maintain models manually.

Yahoo DSP operates under strict performance requirements: the platform participates in real-time advertising auctions where millisecond latency affects revenue. Bedrock enables connecting the right models for specific tasks — search intent classification, semantic query expansion, audience cluster construction — without rebuilding the entire advertising infrastructure from scratch. This reduces operational risks when deploying LLMs in production systems with high response-time requirements.

What Changed for Advertisers

The Amazon Bedrock integration allows SRT to move beyond exact keyword matching. AI models analyze a user's search behavior holistically: what queries they entered, what categories interest them, how intent has shifted over time. The result is audiences built on semantic intent rather than literal phrase matching.

According to the service description in the AWS Machine Learning Blog, the updated SRT provides advertisers with:

• Reaching users with relevant search intent, even if they did not visit Yahoo Search directly
• Cross-channel targeting — display, video, and native advertising based on a unified user search profile
• Aggregation of partner signals: SRT collects search activity from integrated external systems
• A direct bridge between search and display channels — intent is converted into targeting without intermediate steps

"SRT is a core audience targeting tool that helps advertisers reach

users based on their historical search behavior, connecting search intent with display, video, and native advertising," states the official AWS Machine Learning Blog.

What This Means

Yahoo DSP is one of the major independent players in the advertising market, and its move to Amazon Bedrock reflects a broad trend: large adtech platforms are transitioning to LLM infrastructure for interpreting user signals in real time. Competition for audience targeting quality is increasingly determined not by the scale of a cookie database, but by AI's ability to accurately read intent — and quickly translate it into a relevant ad impression.

*Meta has been recognized as an extremist organization and is banned in Russia.