The way consumers find information is changing fast because of generative AI search, and marketing departments are scrambling to keep up. By 2026, people are having conversations with AI to get answers, a huge departure from just typing keywords, which means we all have to rethink our content and where we buy ads. For CMOs trying to stay visible, this means you can’t just rely on ranking #1 anymore. You have to be the source the AI trusts to build its answer.
Key Takeaways
- Your content has to be conversational. That means structuring it to answer questions directly, because generative AI doesn’t just match keywords, it interprets intent and synthesizes a complete answer.
- You need to own your first-party data. Collecting and using it is the only way you’ll be able to personalize experiences and make your ads relevant when an AI sits between you and the customer.
- By Q4 2026, CMOs need to put 25% of their digital ad budget into testing new generative AI ad formats. If you’re not experimenting, you’re already falling behind.
- Forget impression-based metrics. You have to shift to engagement and outcome-based reporting because AI synthesizes information from ten sources before a user ever sees your site, so a click is no longer the whole story.
- Get an internal AI ethics committee in place by mid-2026. This group’s job is to create guardrails that prevent your AI-powered ads and content from becoming biased or inaccurate.
We just ran a campaign for “InnovateTech Solutions,” a B2B SaaS company in supply chain optimization, that shows what it’s like to work in this new reality. The goal was straightforward: get 20% more qualified leads for their AI inventory management platform in six months, on a $750,000 budget. Our core purpose was to capture the attention of high-level decision-makers who are now using generative AI for their early vendor research.
Simply bidding on exact-match keywords wouldn’t have worked. Generative AI like Google’s Search Generative Experience (SGE) or Microsoft’s Copilot reads multiple sources to construct a single, consolidated answer, often making a click to a specific page unnecessary. That meant our content had to be so authoritative and complete that the AI would have no choice but to digest and cite it.
We built our whole strategy around answer-centric content creation. We stopped targeting “best inventory software” and instead created content answering questions like “how AI predicts supply chain disruptions” or “optimizing warehouse logistics with predictive analytics.” We produced deep-dive whitepapers, long-form articles, and case studies that solved real business problems. We also built a content cluster with a main pillar page on “AI in Supply Chain Management” that linked out to 15 granular articles on specific technologies. This kind of clear information architecture is absolutely essential for an AI to recognize the depth of your expertise.
Our creative had two parts. We kept running standard search ads with concise, benefit-focused copy. But for the new generative AI ad spots, which were still in beta on many platforms in early 2026, we tried out conversational prompts and ad formats that could appear inside the AI’s summarized answer. We tested short, embedded videos and even some interactive Q&A modules where a user could ask a follow-up question on the spot. Our traditional ads got a respectable CTR of 4.5%, but the experimental generative AI placements, even with fewer impressions, pulled a CTR of 7.2%. That shows how much more engaged users are when an ad is part of a helpful, synthesized answer.
We tightened our targeting by combining firmographic data (we only wanted companies with 500+ employees in manufacturing or retail) with behavioral signals that showed someone was researching supply chain topics. Using custom audiences in Google Ads and LinkedIn Ads, we went after job titles like “Supply Chain Director” and “Operations Manager.” We had to be that precise because our average CPL (Cost Per Lead) of $185 meant we couldn’t afford to waste money on unqualified prospects.
What Worked and What Didn’t
Our biggest win came from our investment in structured data markup (Schema.org). By carefully tagging our articles with schema like Article, FAQPage, and HowTo, we made it dead simple for AI models to pull out key facts and feature them in direct answers or rich snippets. This gave us a noticeable visibility boost inside AI-generated results, getting our brand in front of people even when they didn’t click. A Q1 2026 eMarketer report backs this up, finding that sites with good schema get their content cited 15% more often by AI. That matched our experience exactly.
On the other hand, our first batch of AI-generated ads with heavy-handed sales language completely bombed. We learned quickly that users in these AI environments are looking for real information and are very sensitive to a hard sell. When our ad copy sounded like a generic banner ad, engagement cratered. So we pivoted to creating informative micro-content that focused on solving a user’s pain point. For instance, an ad promising “20% cost savings” did terribly, while one asking “Struggling with inventory inaccuracies? See how AI can help” performed much better.
The campaign ran from January to June 2026. In that time, we brought in 4,050 qualified leads on a total ad spend of $695,000, which works out to a Cost Per Conversion of $171.60. Calculating a direct ROAS is tough with long B2B sales cycles, but we tracked the leads through our CRM. Based on our historical MQL-to-SQL-to-win rates, we projected the campaign would generate an ROAS of 2.8x over 18 months. This lines up with B2B SaaS benchmarks from a late 2025 HubSpot report on B2B marketing ROI.
Optimization Steps and Learnings
One of our best optimizations came from A/B testing different content formats to see what the AI would prefer. We found that content broken into bullet points or numbered lists that answered a specific question got pulled into AI answers far more often. This discovery led us to go back and reformat a ton of our old, dense content into more digestible, answer-first segments. We also built out semantic content clusters to cover topics from every possible angle. The whole point was to build a knowledge base so thorough that the AI would see us as an unimpeachable source.
We also learned that attribution in an AI-first world is a completely different beast. Your traditional last-click model is broken when a user gets information about you from an AI summary, does their own research, and then finally clicks on a branded ad a week later. We had to build a multi-touch attribution model that gave credit to those early AI-summary touchpoints. It was the only way to see the full customer journey and justify our budget for top-of-funnel content. It’s tough to set up, it requires your ad platforms and CRM to talk to each other perfectly.
We also realized that voice search optimization is more important than ever. As people use smart assistants, they ask questions using natural language. They talk differently than they type. Our team started digging through our own site’s search logs and customer service chat transcripts to find out what questions people were actually asking. If we saw “How does InnovateTech reduce shipping costs?” pop up a lot, we’d build a page specifically to answer that question, formatting it so an AI could easily read it out loud.
Winning in generative AI search is about understanding what people are trying to accomplish at a much deeper level. CMOs have to get good at predicting not just the query, but the underlying question and the format of the answer they want. It takes a mix of data analysis, great content work, and a budget for experimenting with ad formats that deliver actual utility. My advice is to start testing now, even on a small scale. The marketers who figure this out first are the ones who will own the next generation of digital marketing.
How does generative AI search differ from traditional search engines?
Generative AI gives you a direct, synthesized answer in conversational text by pulling information from many different websites. Traditional search just gives you a ranked list of links, and you have to do the work of finding the answer yourself.
What content strategies are most effective for generative AI visibility?
You need to create long-form, answer-focused content that addresses specific questions. Use structured data (Schema.org) markup so the AI can understand your content, and build out topical authority with content clusters. The goal is to make it easy for the AI to grab and cite your information.
How should marketers measure success in a generative AI search environment?
You have to look beyond clicks and impressions. Start tracking engagement with AI snippets, how often your brand is cited in AI answers, and use multi-touch attribution to see the entire journey. In the end, it still comes down to conversions and revenue.
Will generative AI eliminate the need for traditional SEO?
Generative AI will transform traditional SEO, not eliminate it. The basics of understanding user intent and creating quality content are still the foundation. The focus of SEO will just shift toward optimizing for AI interpretation and becoming a trusted source for AI models, instead of just ranking for keywords.
What role does first-party data play in generative AI marketing?
First-party data is your best tool for personalization. It lets you feed an AI signals about a user’s known preferences, which results in much more relevant content and ads. As AI becomes a more common go-between, this data is how you maintain a direct relationship with your customers instead of just handing it over to the search engine.
“HubSpot’s State of AEO 2026 found that 44% of marketers have made a business purchase based on brands they discovered through answer engines.”
