Discovery Ads: Agent Traffic Boosts 2026 Conversions

In 2026, the effectiveness of discovery ads hinges significantly on how marketers interpret and act upon agent traffic, a nuanced metric redefining traditional engagement. We recently analyzed a campaign that illustrates this shift, revealing that raw clicks don’t always translate to meaningful interaction. Understanding user intent behind these interactions is paramount for true campaign success.

Campaign Teardown: The “Smart Home Solutions” Initiative

Our recent “Smart Home Solutions” campaign, running from January to March 2026, aimed to drive leads for advanced home automation systems. The goal was straightforward: reach high-intent homeowners seeking integrated technology. We allocated a budget of $75,000 over the three-month period, focusing heavily on Google Discovery Ads due to their expansive reach across YouTube, Gmail, and the Discover feed. This campaign presented a unique opportunity to scrutinize the emerging impact of agent traffic.

Strategy and Targeting: Beyond Demographics

Our strategy moved beyond standard demographic and interest-based targeting. While we initially targeted homeowners aged 35-60 with incomes over $100,000, we layered on custom intent audiences based on search queries like “smart home installation Atlanta,” “home security systems Georgia,” and “energy efficient home upgrades.” We also uploaded customer match lists of existing clients and lookalikes, aiming for precision. The creative strategy centered on visually compelling lifestyle imagery showing the benefits of smart home integration, rather than just product shots. We believed this approach would resonate more deeply with users encountering our ads in a discovery context.

Creative Approach: Visual Storytelling Meets AI Summarization

The ad creatives were a mix of high-resolution single images and carousel formats. Each ad featured a clear call-to-action (CTA) such as “Get a Free Quote” or “Explore Smart Options.” A critical consideration for this campaign was the rise of AI-driven content summarization within platforms like Google Discover. We crafted ad copy that was concise, benefit-driven, and designed to be easily digestible even when truncated or presented as part of an AI-generated summary. For instance, one top-performing headline read, “Automate Your Home, Simplify Your Life,” followed by a description highlighting energy savings and enhanced security. This focus on summarization-friendly content was a deliberate attempt to capture interest even before a direct click, influencing what we now categorize as agent traffic.

Initial Performance Metrics: A Deceptive Picture

The initial two weeks of the campaign, before significant optimization, provided some interesting data. We observed 8.5 million impressions, a click-through rate (CTR) of 0.9%, and a cost per click (CPC) of $1.15. The campaign generated 6,300 clicks to our landing page. Our initial conversion tracking showed 176 conversions (quote requests or brochure downloads), resulting in a cost per conversion (CPL) of $48.58. The return on ad spend (ROAS) was 0.8:1, meaning for every dollar spent, we generated $0.80 in attributed revenue, which was below our target of 1.5:1. These numbers, while not terrible, indicated room for improvement, particularly in understanding the quality of traffic.

Here’s where the concept of agent traffic became critical. We noticed a substantial portion of impressions and even initial interactions that didn’t immediately lead to a landing page click. These interactions involved users engaging with expanded ad previews, AI-generated summaries of our offerings directly within the Discover feed, or even voice-assisted queries triggered by our ad content. We categorized these as agent-assisted engagements.

What Worked: Embracing Agent Traffic

Our analysis revealed that 35% of total impressions involved some form of agent-assisted engagement. While these didn’t always translate to a direct click on the ad unit itself, they significantly influenced downstream conversions. Users who engaged with an AI summary of our “Smart Home Security” offering, for instance, were more likely to search for our brand directly later or click on a subsequent ad. This indirect pathway was initially underestimated.

We saw that creative variations specifically designed for clarity and conciseness, allowing AI agents to accurately summarize our value proposition, performed better. One particular ad variant, featuring a concise benefit-driven headline and a clear value proposition in the first 50 characters of its description, saw an 18% increase in agent-assisted interactions compared to more verbose alternatives. This suggests that platforms are increasingly prioritizing content that is easily digestible for their AI agents, which then translates into more effective user guidance.

Plus, we discovered that conversions originating from an agent-assisted pathway had a 4.0% conversion rate, compared to a 2.8% conversion rate for direct clicks. This meant that while agent-assisted interactions might have a higher initial “cost” in terms of platform resources or indirect engagement, they often led to higher-quality, more informed leads. This was a critical insight. It challenged our initial assumption that direct clicks were the sole measure of engagement quality.

What Didn’t Work: Over-reliance on Traditional Metrics

Initially, we were too focused on optimizing for traditional CTR and direct landing page conversions. This led us to pause some ad sets that had lower direct CTRs but were generating significant agent traffic. For example, an ad set targeting “eco-conscious homeowners” had a CTR of 0.7%, below our average, but upon deeper analysis, it was generating substantial agent-assisted interest in our energy management systems. Pausing it prematurely meant we missed out on a valuable segment of potential customers who preferred to engage with information via AI summaries before committing to a click.

Another misstep was underestimating the budget required for these agent-heavy interactions. While the CPC for a direct click might be $1.15, the “cost” of a high-quality agent-assisted interaction, which often involved multiple touchpoints and AI processing, was inherently higher. We estimate that factoring in the value of these interactions, the effective cost per meaningful engagement was closer to $1.35 to $1.40 in these segments, a 15% to 20% increase. Not accounting for this in our bidding strategy meant we were underbidding for high-value interactions.

Optimization Steps Taken: Adapting to the New Reality

Recognizing these patterns, we implemented several key optimizations:

1. Refined Creative for AI Agents: We revised all ad copy to be even more succinct and to highlight core benefits in the opening lines. We also began using more descriptive image alt text, understanding that AI agents use this for context.
2. Adjusted Bidding Strategy: We shifted our bidding from “Maximize Conversions” to a target ROAS strategy, allowing the platform to bid more aggressively for interactions that led to higher-value conversions, even if they were agent-assisted. We also increased bids by 15% to 20% on audiences that showed strong agent traffic engagement.
3. Enhanced Tracking: We implemented more granular event tracking on our landing pages to differentiate between users arriving directly and those who had previously interacted with agent-assisted content (e.g., through specific UTM parameters or cookie tracking for users who viewed AI summaries). This allowed us to better attribute conversion value.
4. A/B Testing on Summarization: We ran A/B tests on different ad copy lengths and structures to see which ones generated the most effective AI summaries and subsequent agent-assisted engagement. This iterative process was important for learning what resonated with the algorithms.

Results Post-Optimization: A Clear Improvement

After implementing these optimizations, the campaign saw significant improvements over the subsequent four weeks. While total impressions remained around 5.8 million (due to more precise targeting), the CTR increased to 1.1%. More importantly, our conversion rate from direct clicks rose to 3.2%, and agent-assisted conversion rates climbed to 4.5%. The total conversions for this period were 235, bringing our CPL down to $31.91. Our ROAS improved to 1.7:1, exceeding our target. This demonstrates that understanding and optimizing for agent traffic is not merely an academic exercise. It has a direct, tangible impact on campaign performance.

It’s my strong opinion that ignoring agent traffic in discovery ad campaigns is akin to ignoring a significant portion of your audience. The platforms are evolving, and user interaction methods are diversifying. Marketers who cling solely to direct click metrics will increasingly find themselves behind the curve. The future of engagement is not just about direct interaction. It’s about influencing the informational environment around your brand, much of which is now mediated by AI agents.

Data at a Glance: Pre vs. Post Optimization

Here’s a comparison of key metrics before and after our optimization phase:

The increase in CPC post-optimization might seem counterintuitive, but it reflects our deliberate strategy to bid more aggressively on high-value, agent-influenced segments, which in the end drove down the CPL and boosted ROAS. This illustrates that a slightly higher initial cost can lead to significantly better outcomes when targeting is precise and engagement is understood holistically.

The impact of agent traffic is not just about an abstract concept. It fundamentally alters the way we should approach creative development and bidding strategies for discovery ads. A report by eMarketer in late 2025 indicated a projected 22% increase in AI-mediated ad interactions by 2027, underscoring the urgency of adapting to this trend. This isn’t just about clicks anymore. It’s about influencing the broader informational ecosystem where users make decisions.

I would argue that the most critical lesson here is to challenge your assumptions about what constitutes “engagement.” A direct click is valuable, yes, but an informed, agent-assisted interaction that leads to a conversion is often more valuable. The platforms are getting smarter, and marketers need to get smarter about how users interact with those platforms.

For any marketing team running discovery campaigns today, actively analyzing agent traffic metrics and optimizing creatives for AI summarization is no longer optional. It’s a strategic imperative for maximizing your campaign’s effectiveness.