Google Conversational Ads: What SMBs Need to Do Right Now

01What Google Actually Changed (and Why Your Old Campaigns Are Quietly Failing)

A year ago, search ads were about keywords. Today, they’re about the conversation happening inside Google before anyone clicks. If your campaign still sounds like a keyword list, you’re already losing.

Google’s AI Mode is a structural change to how ads surface and convert in search. Instead of matching keywords to queries, Google’s AI serves ads based on conversational intent: what a user is trying to accomplish across a multi-turn search session, not just what words they typed into a box. Highlighted Answers and Conversational Discovery ads appear natively inside AI Overviews, making them part of the answer rather than adjacent to it.

Google Marketing Live 2026 introduced two formats that matter most: Highlighted Answers, which surface your ad content as a direct response inside an AI Overview, and Conversational Discovery ads, which appear as users navigate multi-turn AI Mode sessions.

The scale is real. Google Search revenue grew 17% in Q4 2025, and their CEO attributed that growth explicitly to AI features. Ads now appear in 25.5% of AI Overview results, up from 5.17% in early 2025, a 394% increase in less than a year. AI Mode hit 75 million daily active users by early 2026, four times higher than at launch in May 2025. That’s not an early-adopter curve anymore. If you’re still building campaigns for 2023-style search behavior, you’re building for a shrinking audience.

02The Part Nobody’s Mentioning: Your Keyword Strategy Just Got Weird

Keywords aren’t dead. They’re just a lossy proxy for what actually determines relevance now. And for a lot of SMBs, that proxy was the whole strategy.

Traditional search matched your ad to a query pattern. Conversational AI retrieves answers based on intent: what the user is trying to accomplish across the whole session, not just this specific search. A user asking “what’s the best way to manage a remote team” followed by “what tools help with async communication” isn’t typing the keywords your project management software campaign is bidding on. But they’re exactly your customer.

Data from Ahrefs and WordStream shows AI Overviews reduce CTR by 34-58% for top-ranked results. A small SaaS company bidding on “project management software” saw CTR drop 12% in Q4 2025 alone as AI Overviews started appearing on 40% of their impression share. That’s not an edge case.

Zero-click searches hit 93% in AI Mode queries versus 60% in traditional searches. If your entire ROI model depends on clicks, you’re optimizing for the 7% of interactions that result in one. Understanding how Google’s AI features are reshaping search results will save you the rude awakening of wondering why your impressions look fine but your conversions have quietly collapsed.

Why Google’s AI Can Tell You’re Pitching

Google’s AI Mode can tell the difference between an answer and a pitch. Most SMB search ads are pitches. Conversational relevance is the degree to which your ad matches the underlying intent of a multi-turn conversation, not just one query, and Google evaluates whether your ad is a useful continuation of the conversation the user is already having.

If your ad copy sounds like a search result title from 2019, it won’t perform in conversational contexts. That’s not a strategic failure. That’s a format mismatch, and the format changed without sending anyone a memo.

Where Budget Is Getting Wasted Right Now

If you’re running exact or phrase match campaigns on high-competition terms in healthcare, tech, or education, you’re competing in categories where AI Overviews trigger 87% of the time on desktop. Your ad is showing up below the fold in most of those auctions: behind the Overview, behind the Highlighted Answers from competitors testing new formats, behind everything a user is actually reading. That’s not a visibility problem you can bid your way out of.

03Your Ad Copy Needs to Sound Like Something a Human Said

Your current ad copy almost certainly fails the AI Mode test. Here’s how to check: read it out loud and ask whether it sounds like something a person would actually say in response to a question. Most traditional search ad copy fails immediately.

Traditional search ad copy is engineered around three things: keyword insertion, character limits, and a call to action. The result sounds like this:

• Before (traditional): “Top-Rated Project Management Software | Free Trial | Manage Teams Easily”
• Before (traditional): “Affordable Knee Pain Treatment | Same-Day Appointments Available”
• Before (traditional): “Business Insurance Quotes | Compare Rates | Save Up to 30%”

These work when the user typed three words into a search bar and wants to scan a list of options. They don’t work when Google’s AI is constructing a conversational answer and needs your ad to fit naturally inside it. Dropping a pipe-delimited headline into an AI Overview is roughly equivalent to handing someone a business card in the middle of a therapy session. Technically you made contact. Nobody’s calling.

Conversational ad copy for AI Mode sounds less like a headline and more like a direct response to what the user just asked:

• After (conversational): “If you’re coordinating a remote team, this is how we make async work actually work, without the 9pm Slack messages.”
• After (conversational): “Knee pain that’s been going on more than a few weeks usually needs imaging before treatment. We can see you this week and tell you what you’re actually dealing with.”
• After (conversational): “Most business owners overpay on insurance because they last shopped it when they started the business. Here’s what a current quote looks like for your industry.”

The structural shift: start with the user’s situation, not your product’s features. Acknowledge what they’re trying to figure out, then tell them what you do about it. It requires actually knowing what your customer is thinking when they search, not just which keywords they use.

04The Real Question: Should You Pause Your Old Campaigns or Run Them Parallel?

Don’t pause everything. Anyone telling you to blow up your traditional search campaigns and go all-in on conversational formats right now is selling something, or hasn’t actually run the tests.

Performance Max campaigns using AI show a 10-14% average increase in conversions at similar or better ROAS. That’s a real lift. But Performance Max and conversational ads aren’t a 1:1 replacement for every campaign type you’re running. There are still search queries where traditional formats win, and you don’t want to find out which ones the hard way by pausing everything in October.

The practical approach is parallel testing for 60-90 days:

1. Identify your highest-spend traditional campaigns. These are the ones where a CTR drop or CPC spike will hurt most.
2. Build conversational ad variants for those exact campaigns. Same offer, different copy structure. Write to the intent, not the keyword.
3. Run both simultaneously with 20-30% of budget toward the new format. Enough to generate real signal without catastrophic downside if the conversational ads underperform initially.
4. Set a 90-day review window. You need enough data to see conversion patterns, not just click rates. The first 30 days will look wrong regardless of which format is actually working.

The risk of moving too slowly is real. Every month that early adopters in your vertical run conversational ads and feed conversion signals back to Google’s AI, the algorithm gets better at serving their ads and comparatively worse at prioritizing yours. Late movers pay 2-3x CPC to catch up. The 90-day parallel test is the minimum viable commitment, not the cautious option.

05How to Read Your New Metrics (Because They’re Lying to You)

The dashboard looks familiar. The numbers look wrong. Nobody’s explained why. That’s going to be the quiet experience of a lot of SMBs for the next 12 months.

Which Metrics Are Now Misleading You

CTR is the big one. If your CTR is dropping and you’re in a vertical with high AI Overview coverage, that’s not your campaigns failing: that’s zero-click behavior in AI Mode being counted in your denominator. You can’t optimize your way out of that number by improving ad copy. It’s structural.

Impression share is also becoming less useful. An impression in an AI Overview context where the user never scrolls to your ad is not the same as an impression in a traditional SERP. They’re being counted identically in your dashboard. They’re not the same thing.

Stop Watching CTR. Start Watching These.

• Conversion rate per impression, not per click. If clicks are dropping but conversions are holding, your conversational ads are doing their job.
• View-through conversions. Users in AI Mode who see your Highlighted Answer and convert later without clicking will start showing up here. This metric was mostly noise in traditional search. It isn’t anymore.
• Assisted conversions in your attribution model. Conversational search creates longer consideration paths. Single-touch attribution will make conversational ads look like they don’t work. They’re often working two steps upstream of the conversion you’re crediting.
• Search term reports at the intent level. Stop reading these query-by-query. Start reading them for intent clusters: what problem was the person trying to solve?

Most SMBs are looking at last-click, keyword-level reporting and asking why the numbers don’t match their gut sense of what’s working. The reporting is measuring the wrong thing, not the campaigns. Fix the measurement before you fix the campaigns, or you’ll fix things that aren’t broken and break things that are.

06The Honest Timeline: When This Becomes Non-Negotiable

You probably have 6-12 months before not adapting becomes visibly, measurably painful. That’s not a long runway. It’s also not a crisis that requires burning everything down this week.

The forcing function won’t be Google deprecating traditional search ads. It will be your competitors’ data. Early adopters running conversational ads feed conversion signals back to Google’s AI, making the algorithm better at serving their ads and worse at prioritizing yours. You’re not locked out. You’re just buying your way in at a worse price each month you wait.

If you’re in healthcare, tech, education, or any high-consideration B2B category, move faster. These are the verticals where AI Overviews trigger most aggressively and where the traditional search ad playbook is becoming expensive fastest. A healthcare clinic running search ads on “knee pain treatment near me” is competing with AI Overviews 87% of the time. Some are already considering pausing search entirely because their ROI went negative before they understood what changed.

B2B advertisers should pay particular attention: Business Agent for Leads is currently in beta. Conversational ads will matter for B2B faster than trade coverage suggests. Build familiarity with the format before that beta opens, not after the waitlist closes.

07What to Actually Do This Week

One thing. Not five, not a full migration. One thing.

Take your highest-spend search campaign and write three conversational ad variants using the copy structure from the before/after examples above. Start with the user’s situation. Acknowledge what they’re trying to figure out. Tell them what you do about it. Keep it under the character limit. Put 20% of that campaign’s budget behind the new variants and let it run for 30 days before you touch anything.

You’re not committing to a new strategy. You’re generating data that will make every subsequent decision clearer. If the conversational variants outperform, you have a mandate to move budget. If they underperform, you’ve learned something specific about your audience that your keyword reports never told you. Either outcome is worth 20% of budget for a month.

The businesses that figure this out in the next six months will have cleaner data, lower CPCs, and a head start on an algorithm that rewards conversion history. The ones that wait will spend 2026 wondering why their search budget stopped working and bidding against people who already know the answer.

Start with the one campaign. The rest follows from there.