How AI in Google Search Ads transforms ad ranking and personalization

How Google uses AI models like LLMs and LEMs to optimize ad ranking, bidding, and user experience in its Search Ads ecosystem.

Illustration by Febrina Tiara

By Alana Salsabila and Widya Putri

Artificial intelligence (AI) in Google Search Ads plays a transformative role in determining which ads appear, how they are ranked, and how advertisers engage with users. This sophisticated integration of AI — specifically through Large Language Models (LLMs) and Learning-to-Earn Models (LEMs) — enables Google to process massive volumes of user queries, predict ad performance, and optimize ad delivery for both user satisfaction and advertiser revenue.

This capability is at the heart of what makes Google Ads one of the most influential digital advertising platforms. Based on internal documentation revealed during the U.S. Department of Justice’s (DOJ) ongoing antitrust case against Google, we now have a deeper look into how AI drives every stage of Google’s ad engine. The revelations show just how much AI in Google Search Ads reshapes the digital ad ecosystem.

Two AI systems power Google's ad pipeline

Google's reliance on AI in Search Ads is built on two foundational systems: Large Language Models (LLMs) and Learning-to-Earn Models (LEMs). Each serves a distinct yet complementary function in the ad ranking process, moving from understanding a user's search intent to finalizing the order in which ads appear.

LLMs: Understanding users through natural language

Large Language Models are primarily responsible for interpreting the meaning behind user queries. Their tasks include:

• Natural Language Understanding (NLU): Decoding what a user wants from a search.
• Mapping user intent to relevant categories of products or services.
• Estimating ad relevance by analyzing semantic context.

Through these tasks, LLMs ensure that only the most pertinent ads are retrieved in the early stage of a search ad cycle. This means when someone types a vague or complex search query, LLMs work behind the scenes to interpret it and retrieve ads that match the user's possible needs.

LEMs: Predictive power for conversions and value

Once LLMs retrieve potentially relevant ads, Learning-to-Earn Models step in to optimize outcomes:

pCTR (predicted Click-Through Rate): Forecasting how likely a user is to click on the ad.

• pCVR (predicted Conversion Rate): Estimating the probability of a purchase or conversion.
• Predicted Conversion Value: Calculating how much revenue a click or conversion could generate.
• Auto-bidding decisions: Assigning bid values based on machine learning predictions.
• The LEMs essentially ensure that ads with the highest potential return — both for Google and the advertiser — are prioritized.

How AI in Google Search Ads determines ranking

A core output of Google's AI ad systems is the ad ranking, which is governed by a metric called LTV or Lifetime Value. This metric blends three essential factors:

1. Revenue Potential: The monetary value Google expects from showing a particular ad.
2. User Impact: How helpful or relevant the ad is expected to be for the user.Advertiser Value: The bid and predicted ROI for the advertiser.

Step-by-step AI-driven ad ranking

The process, powered by LLMs and LEMs, unfolds in several stages:

• Query Interpretation: LLMs interpret the user query and fetch relevant ads.
• Ad Filtering: Low-quality or irrelevant ads are filtered out based on AI assessments of user experience and trustworthiness.
• Creative Personalization: Ad formats are tailored, especially for verticals like retail or travel, to increase user engagement.
• Bid Optimization: LEMs calculate the optimal bid, factoring in predicted performance and competition.
• Final Ranking and Auction: Ads are ranked using the LTV metric, and winners are displayed in search results.

The seamless interplay between these AI models makes ad delivery more targeted, strategic, and profitable.

Performance benefits and industry impact

Google’s internal data highlights how essential AI has become. According to trial documents, LEMs alone contribute 85–90% of Google’s incremental long-term revenue per thousand impressions (LT-RPM). In simpler terms, AI drives most of Google’s ad profits.

Beyond revenue: Enhancing user trust and safety

AI in Google Search Ads isn’t only about money. It’s also used to:

• Detect and prevent spammy or fraudulent ad clicks.
• Predict “Goodclick” outcomes — whether a user will stay on the landing page, indicating a positive user experience.
• Deliver privacy-safe personalization using anonymized behavioral embeddings (e.g., “X-MEN” signals) instead of relying on cookies or direct tracking.

These capabilities help maintain trust among both advertisers and users.

What former Googlers are saying

Jyll Saskin Gales, a former Google employee and now a Google Ads Coach, shared her take on the AI-driven ad infrastructure revealed in the DOJ documents:

“Depending on who you ask, Quality Score is either an immensely important metric or completely irrelevant.
My interpretation of this doc is that Ad Quality and the user experience are more important than ever.”

She pointed out that while Google does not always explicitly show how Quality Score is calculated, AI metrics such as pCTR, landing page experience, and relevance filters make up its real foundation.

The evolving role of AI in digital advertising

The increasing dependence on AI in Google Search Ads suggests broader implications for the digital ad market:

• Consolidation of power: Advertisers who can invest in AI tools or understand Google’s system have a competitive advantage.
• Limited transparency: Advertisers and agencies must rely on signals from Google rather than having direct control or clarity over how ads are ranked.
• Regulatory interest: As AI plays a more dominant role, regulators may begin questioning the fairness and accountability of AI-based ranking systems.

At its core, this AI-driven transformation reveals a tension: more efficiency and personalization, but less transparency and control for users and advertisers alike.

Final thoughts

AI in Google Search Ads is not a distant future — it’s already here, driving nearly every decision behind the ads users see. With LLMs decoding what users want and LEMs predicting which ads will perform best, Google has built a robust, automated ecosystem that maximizes value across the board.

While this delivers remarkable performance gains, it also highlights challenges of fairness, data privacy, and market competition. As regulators and advertisers delve deeper into these systems, the influence of AI on digital advertising will only grow — reshaping how we search, shop, and connect online.

More from PPC coverage

• New AI-powered features for Google Search ads promise better targeting, improved reporting, and increased conversions.
• Temu’s sudden U.S. ad shutdown disrupts Google and Meta ecosystems, with major retailers gaining ground.