By Editorial TeamBy 2026, it was obvious to any marketer paying attention that keyword targeting wasn’t the whole game anymore. For Sarah, marketing director at “Urban Sprout,” it was more than obvious, it was a problem. Her boutique plant delivery service, covering Seattle spots like Capitol Hill and Fremont, was hitting a wall. The team’s whole strategy on Microsoft Ads was built on precise bids for terms like “buy houseplants Seattle” and “succulent delivery Fremont.” It had worked for years. But now, their conversion rates were flatlining while search volume was fine. Sarah knew the reason: people weren’t just typing keywords anymore, they were talking to Cortana and other AI assistants. Her carefully built campaigns were missing these long, conversational queries, and she had to figure out how to catch them.
Key Takeaways
- Start using broad match modified keywords to grab those long, conversational AI searches, but be ready with a heavy-handed negative keyword list.
- Write ad copy that directly answers questions and solves problems, thinking about how real people talk.
- Go beyond keywords by using Microsoft Ads’ audience targeting, in-market and custom audiences are your best friends here.
- Get into your search query reports every two weeks, without fail, to find and adapt to the weird phrases AI is generating.
- Use structured data (schema.org) on your landing pages so you have a better shot at showing up in AI answer boxes.
The Plateau: When Traditional Keywords Aren’t Enough
For a long time, Urban Sprout’s Microsoft Ads campaigns were textbook perfect. They had ad groups for specific plants, delivery zones, and ads with clear offers like “Same-Day Plant Delivery Seattle.” They worked. But in late 2025, Sarah saw the CTR on their best ads start to slip, even though their impression share was solid. “It felt like we were shouting into a void,” she recalled. “People were seeing our ads, but they weren’t clicking with the same urgency.”
It wasn’t that people stopped wanting plants, and it wasn’t a new competitor. The answer was in an eMarketer report from early 2026 showing how fast voice shopping was growing, with people using assistants to start their research. The searches themselves were changing. Instead of typing “ficus Seattle,” a real person was asking Cortana, “Where can I find a pet-friendly ficus plant delivered to my apartment in South Lake Union today?” Urban Sprout’s keyword list was totally unprepared for that level of detail.
“Referral traffic from AI tools like ChatGPT and Gemini has tripled over the past year, and 44% of marketers say they’ve made a business purchase based on a brand they first discovered in an AI answer.”
Understanding the AI Search Shift: Cortana and Beyond
The whole game with AI search, driven by assistants like Cortana, is about understanding intent and context instead of just matching keywords. The queries get longer and more conversational, full of implied needs. For us advertisers, that means our keyword-heavy approach has to evolve into a full-blown Audience Experience Optimization (AEO) strategy.
“We realized our ads had to become conversations,” Sarah explained. “Cortana isn’t just matching words. It’s trying to answer a question. Our ads need to be the answer.” That was the big lightbulb moment. Instead of just bidding on “plant delivery,” they had to start thinking about the real questions people ask, like “What’s the best air-purifying plant for a small office?” or “Where can I get a last-minute gift for a plant lover near Pike Place Market?”
Re-engineering Microsoft Ads for Conversational Queries
The first tangible step for Urban Sprout was a complete audit of their existing Microsoft Ads campaigns. They began by expanding their keyword research beyond traditional tools. They looked at customer service logs, website chat transcripts, and even social media comments to identify common questions and phrases customers used when talking about plants and delivery. This qualitative data proved invaluable.
Using Broad Match Modifiers with Precision
The first tactical shift was getting smarter with broad match modifiers. Exact and phrase match were still good for the high-intent, short keywords, but to catch longer phrases, Sarah’s team started building out modified queries. They moved from simple stuff like “+plant +delivery +Seattle” to longer strings that sounded more like speech, such as “+where +can +I +buy +indoor +plants +online +Seattle” or “+best +succulent +gifts +for +delivery.”
Of course, this meant they had to get way more serious with their negative keywords. “You can’t just throw broad match out there and pray,” Sarah warned. “We were in the search query reports constantly, adding negatives for anything that wasn’t a fit, ‘plastic plants,’ ‘plant care tips free,’ you name it. It’s a never-ending job because you have to protect your ad spend.” That’s why they made bi-weekly SQR reviews mandatory. It was the only way to stay ahead of the garbage queries Cortana would sometimes send their way.
Crafting Ad Copy for AI Search: The Answer-First Approach
Their ad copy had to change completely. Instead of just listing what they sell, the ads started answering the unasked question. So for a query like “easy care plants for beginners,” the headline became “Beginner-Friendly Plants Delivered. Low Maintenance, High Style.” The description followed up: “Struggling with plant care? We curate resilient plants perfect for new owners, delivered across Seattle. Find your green thumb with Urban Sprout.”
The team also leaned heavily into Microsoft Ads’ responsive search ads. They maxed them out, feeding the system up to 15 headlines and 4 descriptions. The ad platform could then assemble the best combination on the fly to match a user’s specific conversational query. This approach to AEO worked because it let them answer a huge number of potential questions without having to build thousands of unique ads by hand.
Ad extensions got an overhaul, too. Sarah’s team updated all the structured snippets, callouts, and sitelinks to act like mini-FAQs. A sitelink for “Pet-Friendly Plant Collection” or “Same-Day Delivery Zones” gave people fast answers right in the search results, which meant the clicks they got were from people who were a much better fit.
Beyond Keywords: Audience Targeting and Intent Signals
Keywords weren’t enough. The team knew AI search uses tons of context and intent signals, so they started digging into Microsoft’s audience targeting tools to run alongside their keyword campaigns.
First, they started layering on in-market audiences. This let them target people who Microsoft’s data showed were already shopping for home and garden stuff or gifts, even if they hadn’t searched for a plant-related term yet. It was a great way to find interested buyers. They also used custom audiences from their own website traffic and customer lists for remarketing, hitting people who already knew them with ads that answered their likely next questions.
“We got a lot more granular with geo-targeting, too,” Sarah added. “We stopped just targeting ‘Seattle’ and built campaigns for Ballard, West Seattle, and other neighborhoods, with ad copy that mentioned local delivery info. Since Cortana uses location to figure out what a user wants, our ads had to be just as local.”
Measuring Success in the AEO Era
They had to change how they measured success. Conversion rate was still the main goal, of course, but the team started tracking new things. They watched for “new query types discovered” in the search query reports and kept a close eye on their average position for queries longer than five words. They even started monitoring on-page engagement to see if people were actually finding the answers they were looking for.
Three months after they started making these changes, the results were clear: a 15% jump in conversion rates from their Microsoft Ads campaigns, driven almost entirely by those long-tail queries. Even better, their average cost-per-click (CPC) on those longer searches actually went down because there was less competition. “By figuring out how people talk to AI, and how AI Agent Trends are shaping search, we’re now capturing a whole new stream of customers,” Sarah admitted.
This isn’t a one-and-done fix. The algorithms behind Cortana are always changing, so what works this quarter might be obsolete next. The only way to keep up is to stay in the data, keep testing new ad formats and targeting, and never forget that you’re trying to answer a real person’s question.
How does AI search, like Cortana, differ from traditional keyword search?
AI search tries to understand a user’s actual intent and the context of their question, which is why it can handle long, conversational sentences. It’s a big shift from traditional search that just matches keywords. The goal of an AI is to give a direct answer, not just a list of links.
What is Audience Experience Optimization (AEO) in the context of Microsoft Ads?
AEO means building your ad campaigns around the audience’s experience, especially now that they’re using AI assistants to search. You have to think past keywords. It’s about anticipating their questions, writing ad copy that gives them answers, and using all the audience signals you have (like in-market data) to be more relevant.
Should I still use exact match keywords for Microsoft Ads in 2026?
Absolutely. Exact match is still your best tool for high-intent, specific searches where you need total control and want to maximize conversion rates. You just can’t *only* use them anymore. You have to pair them with broader match types so you don’t miss out on all the conversational queries people are now using.
How can I find conversational search queries for my keyword research?
Stop relying only on keyword tools. Your best source is your own data: look at your internal site search logs, read your customer service chats and emails, and check social media comments. That’s where you’ll find the real language your customers use. And, of course, your Microsoft Ads search query reports are a goldmine for the weird and wonderful long-tail phrases you never would have thought of.
What role do landing pages play in optimizing for AI search?
Your landing pages are where the conversion happens, so they have to deliver on the promise of your ad. The page needs to immediately answer the question the person just asked their AI assistant and give them a clear path forward. If you add structured data (schema.org) to the page, you’re also giving AI systems like Cortana a clean summary of your content, which improves your odds of getting featured in an answer box.
