AI Mode vs. AI Overviews: Two Behaviors, One Google

By NewSunSEO | May 30, 2026

⏱ 8 min read · Last updated 2026-05-30

New behavioral research shows that Google’s two AI surfaces are not the same search at all. In Google AI Mode, users treat the AI’s shortlist as a decision: 88% accept the list it hands them and 74% pick the very first option. In AI Overviews, those same users behave like shoppers, scrolling back down to compare alternatives roughly 50% of the time. One surface rewards being chosen. The other rewards being compared. For a service business in Nassau or Suffolk County, that split changes how you compete for the click.

Why It Matters

The old local game was simple: rank in the top three blue links and earn the click. That game is shrinking. AI Overviews and AI Mode now sit above - or in place of - those links for a growing share of searches, and they decide which handful of businesses even get named. Google’s own documentation on AI features in Search confirms these surfaces pull from the open web rather than a closed index, so the content you publish still decides whether you appear.

Here is the hard part for a local owner. If the AI shortlists three plumbers and a homeowner picks the first one 74% of the time, positions four through ten effectively don’t exist for that query. This isn’t a future problem - it’s the search results your customers are looking at this week. And because the two surfaces reward different things, optimizing for one while ignoring the other leaves real revenue on the table.

What’s New / How It Works

The headline finding is that intent splits by surface. AI Mode is conversational and decisive. A user asks a question, gets a curated shortlist, and treats it like a recommendation from a knowledgeable friend - they rarely second-guess it. AI Overviews, by contrast, appears at the top of a normal results page, so the full list of links is still sitting right there. Users use it, scrolling back down to weigh alternatives about half the time.

The practical translation is two different jobs. In AI Mode you are fighting to be the named, trusted pick, which is won by authority signals - consistent citations, strong reviews, and content that answers the question directly without hedging. In AI Overviews you are fighting to survive a comparison, which is won by specificity - the concrete detail, price range, or first-hand local fact a generic competitor can’t fake.

The research also surfaced a citation shift worth acting on this quarter: LinkedIn is now the #2 source AI engines cite, behind only YouTube. The blueprint that gets cited is consistent - original first-person posts, clear takeaways near the top, a named author, and a specific claim or number rather than vague advice. The pattern is clear: the engines are rewarding provenance over volume.

The Numbers

• 88% of users accept the shortlist Google AI Mode hands them.
• 74% of AI Mode users select the first option on that shortlist.
• In AI Overviews, users scroll backward to compare options about 50% of the time.
• LinkedIn ranks as the #2 AI citation source, behind YouTube at #1.
• Google Ads begins deleting hourly, daily, and weekly data older than 37 months starting June 2025 - export it first.
• Standard display campaigns migrate to Demand Gen by January 2027.
• OpenAI is rolling out pay-per-conversion ads inside ChatGPT.

In AI Mode people tend to trust the shortlist and take the top pick, while in AI Overviews they tend to shop around. If you optimize for one and ignore the other, you leave about half the funnel on the table.

What Comes Next

Two ad-platform changes deserve a calendar entry. First, Google Ads data deletion: beginning June 2025, hourly, daily, and weekly reporting data older than 37 months is being purged. If you rely on multi-year performance trends - seasonality for a roofing or landscaping business, say - export that granular data now, because once it’s gone it isn’t coming back.

Second, standard display ads are being retired into Demand Gen by January 2027, a shift Google details in its Demand Gen campaign documentation. Demand Gen blends display, Discover, and YouTube inventory with AI-driven targeting, so the creative and audience setup you’re used to will need rebuilding. Start testing it now rather than waiting for the forced migration.

On the search side, OpenAI’s pay-per-conversion ads in ChatGPT signal that AI assistants are becoming commercial real estate, not just answer engines. That makes earning organic AI citations - the kind LinkedIn and YouTube content is winning right now - more valuable, because paid placement inside chat is coming whether you bid or not.

What This Means for You

If you run a business in Nassau or Suffolk, here’s the operator-level read. Build content that wins both AI surfaces: answer the question cleanly enough to be the AI Mode pick, and pack in the first-hand local specifics that survive an AI Overviews comparison. That’s the same non-commodity principle we covered in why non-commodity content wins in AI search - provenance and specificity beat volume. And if your best pages are stale, refreshing them works better than publishing more.

Tactically, that means tightening your answer-first copy with help from SEO copywriting built for AI citation, shoring up the Google Business Profile and citation consistency that feed the AI Mode shortlist through local SEO, and making sure the page the AI sends people to actually converts - a web design problem as much as a content one.

Keep your LinkedIn and social profiles active and consistent with Feedsta, an AI social media manager that keeps the channels AI engines now cite from going quiet. And run a free scan on BizScoreAI to see your AI Visibility Score - how often ChatGPT, Gemini, and Perplexity actually name your business - before you decide where to spend. We don’t promise rankings; we help you become the answer the engines have a reason to choose.

The Bigger Picture

The takeaway from this research isn’t that AI broke search; it’s that “search” is now at least two different behaviors sharing the same logo, and the businesses that win treat them that way. Following that approach means you are not chasing the algorithm but giving it a reason to choose you.

Frequently Asked Questions