B2B SaaS: 1.8x ROAS with AI Ads in 2026

The emergence of AI search fundamentally reshapes how users discover information and, critically, how advertisers capture attention. Traditional ad formats, designed for a keyword-centric world, often fall short in environments prioritizing synthesized answers and conversational queries. This case study details a recent campaign where we rigorously tested new ad formats to establish dominance in an evolving AI search field, illustrating the critical role of continuous campaign testing.

Campaign Teardown: Pioneering AI-Native Ad Formats for B2B SaaS

Our client, a B2B SaaS provider specializing in supply chain optimization software, faced increasing competition in a mature market. Their existing Google Ads campaigns, while consistent, showed plateauing performance metrics. The objective was clear: develop and test ad formats that could thrive within AI-powered search interfaces, specifically targeting decision-makers in large enterprises.

Strategy and Hypothesis: Beyond Keywords

The core strategy revolved around moving beyond traditional keyword-to-ad mapping. We hypothesized that AI search environments would favor ads that:

• Directly addressed user intent expressed in longer, conversational queries.
• Offered immediate, value-driven solutions rather than generic product descriptions.
• Integrated smoothly into summarized AI responses, potentially through new ad unit types.

Our approach involved segmenting the campaign into two primary testing tracks: one focused on adapting existing ad formats with AI-centric copy, and another exploring nascent ad unit types on early-access platforms. We allocated a total budget of $120,000 over a six-week duration for this experimental phase, with a target Cost Per Lead (CPL) of $150 and a Return on Ad Spend (ROAS) of 1.5x.

Creative Approach: The Conversational Imperative

For the first track, we developed several creative variations for Responsive Search Ads (RSAs) and Dynamic Search Ads (DSAs) on Google Ads. The key difference was the copy: instead of focusing on features, we crafted headlines and descriptions that anticipated common user questions in an AI context. For instance, headlines like “How to Reduce Supply Chain Costs by 15%?” or “Automate Inventory Management: See Our Solution” directly mimicked conversational search queries. We used AI-powered copywriting tools to generate hundreds of headline and description permutations, focusing on clarity, directness, and problem-solving language. This allowed for rapid iteration and testing of message resonance.

The second track involved experimenting with what Google was calling “Action Snippets” within its Search Generative Experience (SGE) previews. These were highly condensed, often single-line calls to action or solution statements designed to appear within or directly below an AI-generated summary. The creative challenge here was extreme brevity and impact, requiring a different approach to value proposition communication.

Targeting: Contextual and Intent-Based

While traditional keyword targeting remained a baseline, we augmented it with advanced contextual and audience targeting. We focused on intent signals derived from long-tail, question-based keywords and competitor brand searches. On the audience front, we targeted custom segments built from LinkedIn data, focusing on job titles like “Head of Operations,” “Supply Chain Director,” and “Procurement Manager.” The goal was to ensure our experimental ad formats were exposed to the most relevant, high-value decision-makers.

Results and Analysis: What Worked, What Didn’t

The experimental phase yielded significant insights, demonstrating both the potential and the challenges of advertising in AI search environments.

Ad Format Performance Comparison

We observed a clear divergence in performance between traditional and AI-optimized formats:

The AI-Optimized RSAs, specifically those with question-based headlines, showed a 15% higher Click-Through Rate (CTR) compared to their feature-focused counterparts. This directly translated to a 22% lower Cost Per Lead (CPL) for these formats. It suggests that ads directly answering potential user queries resonate more effectively in a search environment increasingly driven by conversational input.

The most promising results came from the nascent SGE Action Snippets. While impressions were limited due to the early-access nature of the platform, these snippets achieved a remarkable 2.1% CTR and an average CPL of $110. This format, designed for extreme conciseness and direct action, integrated smoothly into AI-generated answers, often appearing as a direct solution recommendation. The eMarketer report on SGE’s impact supports this, highlighting the need for highly relevant, succinct ad content that complements generative AI responses.

Creative Learnings

• Question-based headlines drive engagement: Ads that posed a direct question related to a pain point (e.g., “Struggling with Supply Chain Delays?”) consistently outperformed those stating a feature (e.g., “Advanced Supply Chain Analytics”). The conversational tone mirrors how users interact with AI search.
• Solution-oriented copy wins: Within the ad descriptions, detailing the immediate benefit or solution proved more effective than listing product capabilities. For example, “Reduce inventory waste by 20% with our platform” garnered more clicks than “Our platform offers inventory management features.”
• Brevity is paramount for AI-native units: The SGE Action Snippets reinforced that ultra-short, high-impact messages are essential. Every word counts when an ad might appear as a single line within a generated summary.

What Didn’t Work as Expected

Dynamic Search Ads (DSAs), while generally efficient for broad keyword coverage, performed less optimally in this AI-focused test. Their reliance on crawling website content for ad generation sometimes resulted in less precise, less conversational ad copy compared to our manually crafted, AI-optimized RSAs. The CPL for DSAs in this experiment was approximately $195, significantly higher than the AI-optimized RSAs.

Another challenge was the limited availability of strong analytics within the early-access SGE platform. While performance looked promising, attributing conversions directly to specific snippet variations required more manual tracking and correlation than with standard Google Ads reporting. This is a common hurdle with new ad environments, and platform providers are still developing complete measurement tools.

Optimization Steps and Future Outlook

Based on these findings, we implemented several optimization steps:

1. Increased budget allocation to AI-optimized RSAs: We shifted 40% of the overall campaign budget towards these higher-performing ad groups, reducing spend on traditional formats.
2. Refined creative for conversational flow: We continued to iterate on ad copy, using A/B testing to refine question phrasing and solution statements, aiming for even higher CTRs.
3. Explored more AI-native ad opportunities: We maintained close communication with platform representatives to gain earlier access to new ad unit types as they become available. This proactive approach is critical for staying ahead in a rapidly evolving search field.
4. Enhanced landing page alignment: We ensured that landing pages directly addressed the questions posed in our AI-optimized ads, providing immediate, relevant information to users who clicked through. This improved conversion rates across the board.

The early success with SGE Action Snippets, despite their limited scale, points to a future where ad formats are deeply integrated into AI-generated content. Advertisers will need to think less about keywords and more about intent, context, and how their message can become part of a synthesized answer. This demands a strategic shift towards hyper-relevant, concise, and value-driven communication.

This campaign demonstrated that proactive campaign testing of new ad formats is not merely an option but a necessity for achieving AI search dominance. The field is shifting, and those who adapt their creative and targeting strategies now will reap the benefits as AI search becomes the default.