Google Tests Gemini‑Powered Ad Formats in Search

Google Search is moving beyond simple keyword matches toward a more conversational experience. To keep advertising relevant in that environment, the company announced a series of experiments that use its Gemini large‑language model to generate ad content on the fly. The effort is part of a broader push to make paid placements feel like helpful answers rather than interruptive banners.

The problem: ads get lost in AI‑driven results

When users enable AI Mode in Search, the interface surfaces synthesized answers, lists, and recommendations. Traditional text ads often sit at the bottom of the page or appear in a separate carousel, making them easy to overlook. Google’s data shows that 75 % of users say AI Mode helps them decide faster, but the same data indicates a gap: users still want explicit product details and trustworthy guidance that a generic ad can’t provide.

What’s being tested

1. Conversational Discovery ads

These ads are triggered by a specific user question. Instead of showing a static headline and description, Gemini drafts a short, contextual response that directly addresses the query. For example, a search for “low‑maintenance ways to make my home smell like a spa” could generate an ad that highlights a brand’s automatic diffuser, cites its quiet operation, and links to a product page. The ad is labeled Sponsored and includes an independent AI explainer that summarizes the product’s key points, separate from the advertiser’s copy.

2. Highlighted Answers

When AI Mode returns a list of recommendations—such as “best language‑learning apps for a short trip”—high‑quality ads can appear as a Highlighted Answer at the top of that list. The Gemini model evaluates the advertiser’s assets, pulls the most relevant product data, and presents a concise recommendation that sits alongside the organic answers.

3. AI‑powered Shopping ads

For high‑consideration purchases like appliances or electronics, Gemini assembles a set of the advertiser’s most relevant products and writes a brief explainer that compares features, price, and fit for the user’s stated need. The result is a shopping ad that feels more like a curated recommendation than a generic product feed.

4. Business Agent for Leads

In sectors where lead capture is critical—higher education, automotive, professional services—Google is testing a chat‑style agent embedded in the ad. Clicking Chat launches a Gemini‑driven conversation that pulls answers from the brand’s website in real time, allowing the user to get instant clarification without leaving the search page. The interaction still ends with a lead capture form, but the conversation replaces the static questionnaire.

Direct Offers pilot gets an upgrade

The Direct Offers pilot, launched earlier this year with brands such as Chewy, Gap, and L’Oréal, now supports:

• Promotion bundling – advertisers can upload multiple discount types (percentage off, giveaways, local coupons) and let Gemini assemble the most compelling bundle for a given query.
• Native checkout – integration with the Universal Commerce Protocol (UCP) lets shoppers complete a purchase without leaving the search results, reducing friction for high‑intent users.
• Travel expansion – partners like Booking.com and Expedia will be able to surface special travel deals directly within AI‑assisted trip planning flows.

These enhancements keep the offers visible inside the AI Mode response, making them discoverable at the moment a shopper is evaluating options.

How advertisers can prepare

Google advises marketers to build a solid foundation with its AI Max for Search, AI Max for Shopping, and Performance Max campaign types. Those products already incorporate Gemini signals for bidding and creative generation, so they serve as a natural stepping stone toward the new ad formats.

Why it matters

The experiments illustrate a shift from “ad placement” to “ad participation” in the AI conversation. By generating context‑aware copy and attaching an independent explainer, Google hopes to reduce the perception of bias that often accompanies sponsored content. If the tests show higher click‑through rates and better post‑click conversion, the formats could become a permanent part of the Search ad inventory, reshaping how brands reach users who are already in a decision‑making mindset.

---

The article is based on Google’s announcement dated May 20 2026. For further details, see the official Google Ads blog post.