- Heidi Schwende7 days ago6 min read
What It Actually Means for Your Marketing Budget
Advertising has arrived in AI search. Google expanded ads in AI Overviews to 12 countries in December 2025. OpenAI is building ad prototypes for ChatGPT. Perplexity launched ads a year ago, then pumped the brakes when they couldn't prove ROI.
If you're running paid search campaigns or relying on organic traffic for revenue, this isn't theoretical anymore. It's reshaping how your prospects find you—and what it costs to reach them.
Google quietly expanded advertising within AI Overviews to Australia, Canada, India, and nine additional English-speaking markets on December 19, 2025. No press release. No Marketing Live announcement. Just an updated support doc. The company also started testing ads in AI Mode, their ChatGPT-style conversational search experience.
The speed of this rollout matters: ads now appear alongside roughly 40% of AI Overview search results, up from just 3% in January. That's not gradual adoption. That's a fundamental shift in how Google monetizes the new search experience.
Meanwhile, OpenAI spent 2025 in a very public dance with advertising. CFO Sarah Friar confirmed they're "exploring ad models." They tested in-app promotional messages. Users hated them. The feature got yanked.
Then in early December, CEO Sam Altman declared a "code red"—telling employees to deprioritize advertising work and focus on improving ChatGPT before they lose users to Google's Gemini and Anthropic's Claude. Work on ads, shopping agents, and a proactive assistant called Pulse got shelved.
That lasted about three weeks. By Christmas, reports surfaced that OpenAI employees are actively mocking up ad formats: sponsored sidebars, priority placement for paid content in responses, and—most notably—using chat histories to target ads. Only about 2.1% of ChatGPT queries involve explicit buying intent, but all that conversational context is too valuable to leave on the table.
The financial pressure explains the whiplash. Even with annualized revenue hitting $10 billion, OpenAI's operating losses are projected to reach $74 billion annually by 2028. Only 5% of ChatGPT's 400 million users pay for subscriptions. The ad model isn't optional—it's inevitable.
Perplexity is the cautionary tale. They launched ads in November 2024 promising premium CPM rates and access to affluent, educated users. By October 2025, they stopped taking new advertisers. The problem wasn't demand—it was measurement. Advertisers couldn't track CTRs or ROAS with any reliability. That's a non-starter for anyone accountable to revenue outcomes.
The Numbers You Should Actually Care About
Semrush analyzed 10 million keywords across 2025 and found AI Overviews surged to nearly 25% of queries by mid-year before pulling back to around 16% by November. Google is still figuring this out in real time.
The click data tells the real story. Seer Interactive tracked over 25 million impressions from June 2024 to September 2025 and found organic CTRs for informational queries with AI Overviews dropped 61%. Paid CTRs on those same queries fell 68%. When AI Overviews appear, Adthena found paid search CTRs decline an additional 8-12 percentage points.
Google's response? They claim reduced clicks get offset by 10% more search volume overall, creating "more opportunities for monetization." That's fine for Google's revenue. It's cold comfort for your cost-per-acquisition.
The user behavior shift is the real story. McKinsey found that 44% of AI search users now say it's their primary source for buying decisions—beating traditional search at 31%. Two-thirds of consumers believe AI will replace traditional search within five years. And 82% say AI-powered search is more helpful than Google's traditional results.
What Google Isn't Telling Advertisers
Right now, you can't bid specifically for AI Overview placements. You can't target conversational intent separately. You can't see granular performance metrics for AI impressions. Your existing Search, Shopping, and Performance Max campaigns are eligible to show in AI Overviews, but you have no control over when or how.
Google says they're "learning" and will "announce when there's more to share." Translation: they're making you pay for impressions they can't yet prove are worth paying for.
The matching methodology is also different. Instead of keyword selection driving eligibility, AI Overviews analyze semantic relationships between queries, generated content, and available ads. Your carefully structured keyword strategy may not map to what actually gets served.
Google is excluding sensitive verticals from AI Overview ads—healthcare, finance, politics, gambling, alcohol. If your business touches any of these, you're not playing in this sandbox yet.
The Real Strategic Question
Here's what nobody's asking loud enough: does advertising in AI search actually work for advertisers?
Perplexity paused their ad business specifically because they couldn't prove ROI. OpenAI pulled their shopping feature after user backlash, declared a "code red" to fix product quality, then went right back to designing ad formats three weeks later. Google won't release performance data. The platforms are charging premium rates for an ad format with no established benchmarks.
The OpenAI whiplash is telling. They know ads risk alienating users who came to ChatGPT specifically because it wasn't another ad-stuffed Google experience. But the math doesn't work without monetizing free users. Every AI platform is caught in the same trap: the infrastructure costs are staggering, subscription conversion rates are abysmal, and investors want returns.
If you're a mid-market company with real revenue targets, the smart play isn't rushing to "test AI ads." It's demanding the same accountability from AI placements that you'd expect from any other channel.
What This Means for Organic Visibility
The ad expansion accelerates a trend that's already gutting organic traffic for informational queries. Semrush projects AI search could overtake traditional organic traffic for many verticals by early 2028. If Google shifts its default experience to AI Mode, that timeline compresses. (I wrote about this in more detail in CMOs: Your 2026 SEO Budget Needs A Reality Check.)
Only 16% of brands systematically track AI search performance. That's a problem when your own sites typically make up just 5-10% of what AI systems reference in their answers. The rest comes from affiliates, reviews, forums—content you don't control.
The companies winning in AI visibility publish original research, maintain consistent brand messaging across channels, and present information in formats AI systems can easily extract and quote. Generic SEO content that exists purely to capture long-tail traffic? That's exactly what AI Overviews eliminate.
This isn't about whether to embrace AI search—that decision has been made for you. The question is how to maintain visibility and conversion efficiency while the platforms figure out what they're selling.
One development worth watching: OpenAI is building ad targeting based on users' chat histories. That's intent and context data that makes Google's keyword targeting look primitive. If they execute without triggering a privacy backlash, the targeting precision could be valuable. But that's a big "if"—and the measurement infrastructure still doesn't exist.
Stop optimizing for impressions. Start tracking AI citations as a distinct KPI. The brands getting cited today are building moats their competitors will struggle to cross.
Push back on any agency or platform rep telling you to increase AI search spend without proving it converts. Until the measurement infrastructure exists, you're subsidizing Google's learning curve.
Double down on what AI systems reward: original research, expert perspectives, consistent brand authority. The same signals that drive traditional SEO success drive AI visibility—but you're optimizing to be quoted, not clicked.
And pay attention to what happens next. Google adding AI Mode ads signals where this is headed. OpenAI's "code red" pause lasted all of three weeks before they went back to designing ad formats—that tells you everything about the financial pressure these platforms face.
The question isn't if advertising colonizes AI search, but whether the targeting and measurement tools mature fast enough to make it worth your budget.
My Recommendations:
- Don't rush to "test AI ads" — demand the same ROI accountability you'd expect from any other channel
Don't rush to "test AI ads" — demand the same ROI accountability you'd expect from any other channel
- Track AI citations as a distinct KPI — not just impressions or rankings
Track AI citations as a distinct KPI — not just impressions or rankings
- Push back on agencies and platform reps asking you to increase AI search spend without conversion proof
Push back on agencies and platform reps asking you to increase AI search spend without conversion proof
- Publish original research and expert perspectives — that's what AI systems cite
Publish original research and expert perspectives — that's what AI systems cite
- Maintain consistent brand messaging across channels — AI pulls from everywhere, not just your site
Maintain consistent brand messaging across channels — AI pulls from everywhere, not just your site
- Stop optimizing for clicks, start optimizing to be quoted
Stop optimizing for clicks, start optimizing to be quoted
- Monitor the measurement gap — until platforms prove performance, you're subsidizing their learning curve
Monitor the measurement gap — until platforms prove performance, you're subsidizing their learning curve
- Watch OpenAI's chat history targeting — could be valuable if they execute without privacy backlash
Watch OpenAI's chat history targeting — could be valuable if they execute without privacy backlash
- Keep investing in traditional SEO fundamentals — same signals drive AI visibility
Keep investing in traditional SEO fundamentals — same signals drive AI visibility
- Track which of your keywords trigger AI Overviews — know where your traffic is actually going
Track which of your keywords trigger AI Overviews — know where your traffic is actually going
The platforms want your ad dollars now. Make them earn it.
Related Reading:
- CMOs: Your 2026 SEO Budget Needs A Reality Check — Why organic traffic patterns are becoming less predictable and what to do about itGoogle's AI Shopping Features in 2026: What Ecommerce Brands Need to Know — How product discovery is changing with AIPerformance Marketing Priorities for 2026 — What serious CMOs are actually doing
CMOs: Your 2026 SEO Budget Needs A Reality Check — Why organic traffic patterns are becoming less predictable and what to do about it
Google's AI Shopping Features in 2026: What Ecommerce Brands Need to Know — How product discovery is changing with AI
Performance Marketing Priorities for 2026 — What serious CMOs are actually doing
Sources:
- Google Ads HelpPPC LandThe InformationMacRumorsFortuneEMARKETERSearch Engine LandSeer InteractiveMcKinsey & CompanySemrush
PPC Land
MacRumors
Fortune
EMARKETER
McKinsey & Company
Semrush
- Paid Search AdvertisingPaid Search
