By 2026, Eleanor Vance had a problem. As marketing director for “Urban Bloom,” a hot online plant shop out of Atlanta’s Old Fourth Ward, her company’s growth had completely flatlined. Their social media was strong and their content was consistent, but the old SEO playbook wasn’t working on the new AI-powered search engines. People weren’t just typing keywords anymore. They were having conversations. Eleanor knew that getting a handle on AI search insights was the only way for Urban Bloom to survive. But where do you even start when you’re trying to turn weird AI summaries into an actual marketing campaign?
Key Takeaways
- Look at the intent clusters AI search like Google’s Search Generative Experience (SGE) creates to find what customers really need, going way beyond simple keyword lists.
- Build your content to answer a whole conversation, not just one question. Think about what the user will ask next and have the answer ready.
- Use AI search data as a research tool. When you see gaps in the market showing up in AI-generated summaries again and again, you’ve probably found a new product opportunity.
- A/B test your new AI-focused content against your old SEO pages. Don’t just look at traffic, measure things like how long people stay on the page and whether they actually buy something.
- Your marketing team needs to learn prompt engineering. It’s a new skill, but teaching them how to ask AI platforms smart questions is the only way to get smart, actionable answers for your campaigns.
Eleanor’s first instinct was to just do more keyword research, but the tools felt clumsy and outdated. A report telling her “succulent care tips” is a popular search is one thing, but it’s useless for understanding a query like, “What’s the easiest succulent to keep alive for someone who travels a lot, and where can I buy one online in Atlanta?” That query isn’t a keyword, it’s a short story with a problem that needs solving. The standard SEO metrics they were tracking from their Ponce City Market office, while good for a baseline, couldn’t see that deep. “We’re missing the forest for the trees,” she’d tell her team, sketching frantic flowcharts on the whiteboards.
Uncovering Latent Intent with AI Search Data
Getting the right data was the first real challenge. Her team’s analytics platforms were built for a world of keywords and click-throughs. Eleanor needed to see inside the black box of AI search results. She got her team into beta programs for AI search platforms, especially Google’s Search Generative Experience (SGE), which had become dominant since its big rollout in mid-2025. The goal was to understand the summaries and conversational answers the AI was constructing for people, not just to scrape the results themselves.
“The big change is learning what the AI thinks people are trying to accomplish, not just what they type,” Eleanor explained to Marcus, her lead content strategist. “When SGE gets a query like ‘low-light plants that purify air for a north-facing apartment,’ it pulls information from all over. We have to analyze that final, synthesized answer.” They started using a new breed of analytics tools, many of them still rough around the edges in 2026, that could parse these AI summaries and spot recurring themes and unmet needs. It was urgent work. A recent 2026 eMarketer report showed that almost 60% of US online searches now involved an AI-generated summary.
One of the first things they found was a huge group of searches all circling around “pet-safe, non-toxic plants for small spaces.” Urban Bloom had plenty of pet-safe plants, but they weren’t hammering the “small spaces” angle hard enough in their copy. They also saw that AI summaries often brought up specific plants like the Prayer Plant (Maranta leuconeura) or Boston Fern (Nephrolepsis exaltata) in response to general pet-safety questions. That was a direct order for Marcus. He and his team quickly launched a new blog series, “Compact Companions: Pet-Friendly Plants for Cozy Atlanta Homes,” full of care guides and direct links to buy the exact plants mentioned.
Crafting Content for Conversational AI
Next, they had to completely overhaul how they created content. Old-school SEO was about creating a single, long article to own a keyword. But AI search favors content that can answer a string of related questions and anticipate what the user wants to know next. “Think like a helpful librarian, not a keyword stuffing robot,” became Eleanor’s constant refrain in their marketing meetings.
They began rebuilding their blog posts and product pages from the ground up. Instead of one FAQ section at the very end of a post, they wove answers to follow-up questions right into the main text. An article on the Monstera Deliciosa, for example, would naturally flow into sections answering “why are my Monstera leaves turning yellow?” or “how often should I water a Monstera in Atlanta’s humidity?”. This structure made their content incredibly useful for AI models, which could then confidently pull complete, contextual answers for users who were asking multiple questions in a row.
A new IAB report on generative AI’s impact on advertising had just confirmed their hunch, showing that users who interact with AI search results spend more time on the answers. People want complete information, not a list of links. So Urban Bloom started loading up their product pages with comparison charts, specific environmental requirements, and even short videos showing how to care for the plant. They were building a complete, authoritative resource that an AI would have no problem recommending to its users.
Identifying New Product Opportunities
An unexpected win from digging through all this AI search data was the discovery of brand-new customer groups. Their analysis tools started flagging weirdly specific, recurring queries that pointed to a hole in the market. People kept asking for things like “subscription boxes for rare or unusual plants” and “plants that thrive in smart homes with automated lighting.”
“We never even considered a subscription box for ‘rare plant enthusiasts’ before,” Eleanor said during a strategy meeting at a coffee shop near Piedmont Park. “Our whole business is built on convenience and making plant ownership easy for beginners. But the AI is showing us a paying customer base that’s actively looking for something new and challenging.” That insight, which came entirely from the aggregated intent of thousands of AI searches, led directly to Urban Bloom piloting a “Curated Collector’s Box” for hard-to-find plants. They promoted it with content aimed squarely at answering “where to find unique indoor plants in Georgia,” and the launch blew past all their projections, showing that AI search could be an amazing tool for product development and market expansion.
They also noticed a constant drumbeat in the search data around “eco-friendly packaging for plant delivery.” Urban Bloom already used recyclable materials, but they weren’t talking about it. AI summaries consistently brought up sustainability when people asked about buying plants online. The team moved fast, updating the website and all their marketing to put their sustainable practices front and center. It was a direct response to what the AI data showed them users were worried about.
The Role of Prompt Engineering for Marketers
The deeper the team went, the more they realized that getting good information out of AI search models took actual skill. It wasn’t a passive process. It required prompt engineering. Asking a generic question like “What are people searching for?” got them generic, useless answers. But asking, “Analyze recent conversational AI search queries related to indoor plants, identify common pain points for first-time buyers in urban environments, and suggest content themes that address these issues, specifically mentioning solutions available in the Southeastern US” gave them gold.
Eleanor got her team formal training in prompt engineering. They learned how to build queries with specific constraints, tell the AI what kind of answer format they wanted, and tweak their prompts to get better results. It gave their own intuition a powerful new analytical tool. “You’re learning how to talk to a brilliant, but very literal, research assistant,” Marcus said, after getting particularly good at writing complex prompts. “The better your question, the better your answer.”
This new skill was especially powerful for competitive analysis. They could frame prompts to make the AI compare Urban Bloom’s offerings to their rivals’, showing them exactly where competitors looked strong in AI summaries and where they were weak. It gave them a clear map for how to position their own content and products to win.
Measuring Impact and Iterating
Of course, a new strategy is worthless if you can’t measure it. Urban Bloom had to create new KPIs that went beyond organic traffic. They started tracking things like “AI summary mentions” (how often their site was cited in an AI-generated answer), “multi-turn query completion rates” (a measure of whether users found everything they needed on their site after coming from an AI search), and the conversion rates on inquiries for new products they’d identified through AI. These evolving metrics gave them a much sharper picture of how they were performing.
They also ran constant A/B tests. They’d pit a new page, designed for conversational AI search, against an old, traditionally optimized page. The AI-optimized content consistently delivered higher engagement and more conversions on complex, informational searches. Their own internal data showed that over six months, articles built for multi-turn AI queries had a 15% longer average session duration and a 10% higher conversion rate for leads than the old keyword-stuffed pages.
Eleanor knew the work was never done. The AI search field would keep changing, and so would their strategy. But by fully committing to understanding customer intent through the lens of AI, Urban Bloom had not only figured out how to compete in the new environment but had also discovered new ways to grow their business. It proved they were adaptable and willing to look past the obvious metrics.
Using AI search insights completely changed Urban Bloom’s marketing. They stopped chasing keywords and started listening to what customers were actually trying to do, a shift that led them to new product lines and real, sustained growth.
What are AI search insights?
They’re the conclusions you can draw by analyzing how AI search engines like Google’s SGE interpret what users are asking. This includes looking at the summaries the AI creates, the follow-up questions it suggests, and the real intent it figures out from a person’s conversational query.
How do AI search insights differ from traditional keyword research?
Traditional keyword research is about finding popular search terms. AI search insights dig deeper, analyzing the full context of a query to understand a customer’s true goal, what their next question might be, and the specific problem they’re trying to solve.
Can AI search insights help with product development?
Absolutely. When you analyze patterns in AI summaries and user questions, you can spot problems people have that your current products don’t solve. This points directly to opportunities for new products or services, like how Urban Bloom created its “Curated Collector’s Box.”
What is prompt engineering in the context of marketing?
For a marketer, prompt engineering is the skill of writing very specific and detailed questions for AI tools to get back actionable intelligence. It’s how you turn a general AI into a powerful tool for competitor research, content strategy, and understanding what your customers want.
What metrics should marketers track for AI search performance?
You need to go beyond standard SEO stats. Start tracking how often your content gets mentioned in AI summaries (“AI summary mentions”), if users are getting all their answers on your site (“multi-turn query completion rates”), and the conversion rates from traffic that you know was driven by AI-informed content.
