The Study That Should Reset Your Ad Budget
SE Ranking ran 50,032 commercial keywords through Google AI Mode on June 30, 2026. Twenty niches, US only, fresh data. The findings should make every performance marketer sit up.
29.45% of those queries returned a text ad. That's nearly one in three. In 71% of ad cases, two competing ads sat side by side in the answer block. Google went from zero ads in AI Mode to a third of commercial queries in under a year.
But the number that should actually change your budget is this: 88% of advertisers who paid for that placement were not cited as a source in the AI answer. At the exact URL level, it's 1.95%. And 85% of those same advertisers don't rank organically for the keyword they're bidding on.
The ad buys you the slot. It does not buy you the citation. Those are two separate games wearing the same interface.
What Predicts Whether Your Ad Shows
This is where it gets useful for planning. SE Ranking found that CPC is the single best predictor of ad presence in AI Mode. Keywords under $2 CPC showed ads 24.33% of the time. Keywords at $10 and above? 53.56%. A straight line.
Search volume? No correlation. Keyword difficulty? None. The metrics most PPC managers use to plan campaigns told you nothing about whether your ad would appear in AI Mode.
The niche swings are enormous. Pets returned ads on 72.38% of commercial keywords. Healthcare sat at 2.64%. That's a 70-point gap between the top and bottom. If you're in a high-CPC vertical like legal, finance, or home services, assume AI Mode is already showing competitor ads against your keywords. If you're in healthcare, you've got breathing room, but it won't last.
Google's own bidding changes compound this. The August 17 Smart Bidding update we covered recently means budget-limited campaigns will see bidding behavior shift. Combine that with AI Mode's ad expansion and you're looking at a paid search environment that behaves nothing like it did six months ago.
The Control That Makes This Trustworthy
SE Ranking didn't just compare advertisers to non-advertisers and call it a finding. They controlled for Domain Trust, backlinks, referring domains, and organic standing. They matched advertisers against non-advertising domains with the same authority profile and the same organic footprint.
The result was clean. Advertisers were cited no more often than equivalently authoritative non-advertisers. The ad spend had zero measurable effect on whether AI Mode cited the domain as a source in the answer.
This is the clearest proof yet that generative engine optimization is its own discipline, separate from paid media. If someone tells you AI Mode spend will "build your AI visibility," ask them for the data. The SE Ranking study is the data, and it says the opposite.
This matters because Google already generates synthetic images of your brand in AI Overviews without your input. The visual layer and the textual citation layer are both earned, not bought. The ad is the one thing you can pay for, and it's the one thing that doesn't carry over into the citation.
A Few Advertisers Own Entire Niches
Here's a detail that should worry anyone competing in high-ad verticals. Across all 50,032 keywords, 2,930 unique advertisers generated 25,243 ad appearances. Spread evenly, that's about 8.6 appearances per advertiser. They're not spread evenly at all.
In Pets, 68 advertisers split 3,514 appearances. That's an average of 51.68 appearances each. A small club is winning the same slots over and over. In Entertainment and Hobbies, 49 advertisers split 2,622 appearances (53.51 each). These niches have effectively been locked down by a handful of brands.
Meanwhile in Healthcare, 36 advertisers split just 91 appearances (2.53 each). In News and Politics, 81 advertisers split 211 appearances (2.6 each). Wide field, thin coverage.
What this means: if you're entering a niche where a few advertisers already own the AI Mode ad slots, you're not bidding into an open auction. You're trying to break into a club. CPC alone won't tell you this. You need to look at how many advertisers are competing and how often they appear.
The Measurement Gap Is Closing
Here's the other half of this story. Google Analytics 4 added an "AI Assistant" channel on May 13, 2026. It captures traffic from ChatGPT, Gemini, Claude, and other AI chatbots as a distinct channel, separate from organic search.
Until now, AI assistant traffic was invisible. It was lumped into organic search or referral data. You couldn't tell whether a visitor came from Google's blue links or from ChatGPT recommending your product. Now you can.
The channel only counts forward from May 13, so you need a few months of data before trends become reliable. But if you're investing in GEO, and the SE Ranking study strongly suggests you should be treating it as separate from paid, this is how you measure whether it's working. Most marketers still can't report AI assistant traffic as a line item. Being able to puts you ahead of the field.
LinkedIn quietly added something useful too. Post analytics now split reach into in-network and out-of-network impressions. In-network is your followers and connections. Out-of-network is strangers the algorithm decided to show your content to. If 95% of your impressions are in-network, you're performing for your own audience. If 40% are out-of-network, the algorithm is actively distributing you to new people. That's the number that predicts growth, not follower count.
The through line across all of this: paid visibility and earned visibility are finally measurable as separate things. Google split them in AI Mode (ads don't earn citations). GA4 split them in analytics (AI traffic has its own channel). LinkedIn split them in social (in-network vs out-of-network). The tools caught up to the reality. The question is whether your strategy does too.
What to Do This Week
Three moves, all cheap.
First, pull your top 50 commercial keywords by CPC. Anything above $5 is likely already showing competitor ads in AI Mode. Anything above $10 is almost certainly showing two. Sort by CPC and you've built your AI Mode ad risk profile in five minutes.
Second, for each of those keywords, check whether you rank organically. If you're in the 85% who don't, your cheapest AI visibility win is not more ad spend. It's closing the organic gap. An SEO sprint on 10 keywords where you already bid will do more for your citation rate than doubling your AI Mode budget.
Third, set up the GA4 AI Assistant channel. Go to Admin, verify AI Assistant is active, and start pulling weekly reports. Three months from now you'll have a baseline that most of your competitors don't.
And while you're at it, ChatGPT's own ad business is missing forecasts by 90%. When the two biggest AI platforms are both struggling to prove ad value, the idea that ad spend builds "AI visibility" gets even harder to justify.
The Two-Track Future
Google AI Mode runs two parallel auctions and lets everyone pretend they're one. The paid auction decides who gets the ad slot. The citation auction decides who gets referenced in the answer body. Your money influences one. Your content, authority, and structured data influence the other.
The agencies that understand this separation will build two strategies. The ones that don't will keep selling "AI visibility packages" that are really just ad buys with a better name. The SE Ranking data gives you the language to ask which one you're buying.
The data is from June 30. Whatever you're seeing today is already higher. The 29.45% ad rate is the floor, not the ceiling. And the 88% non-citation rate is the number your CFO needs to see before the next budget review.