AI Overviews vs. AI Mode: User Behavior & SMB Strategy

Introduction

Most of the small-business owners we talk to use “AI Overviews” and “AI Mode” interchangeably. That made sense a year ago. It does not in 2026. Google's two AI surfaces are now distinct enough — in interface, in intent, in user behavior — that planning content and ads for both at once produces work that is mediocre on both.

A new piece of research from Eric Van Buskirk at Clickstream Solutions, covered this week by Search Engine Land, studied roughly 846,000 U.S. Google search sessions from February and March 2026, using anonymized clickstream data from Surfer SEO. The finding, broadly: AI Overviews users compare and scroll. AI Mode users accept the answer and stop. The difference is large, repeatable, and has direct implications for how a small business should split its limited time and budget.

This post is the strategic translation of that data. We will walk through what the study found, what we trust about it (and what we are watching for), and the concrete content and ad-strategy decisions that fall out of the contrast. There is a Fort Wayne section toward the end with vertical-specific scenarios for HVAC, dental, and legal practices.

Key Takeaways

AI Mode is a closed loop. In the Clickstream Solutions study, 88% of AI Mode tasks involved accepting the AI's shortlist as-is, 74% selected the first-ranked item, and 64% clicked nothing at all.
AI Overviews is a comparison environment. With AI Overviews present, 47.5% of scrolling was backward (vs. 27% without), and cursor activity covered roughly 83% of the viewport as users compared multiple options.
Different surfaces, different jobs. The pages that win in AI Overviews are comparison-friendly, list-formatted, and skimmable. The pages that win in AI Mode are conversational, complete, and depth-first.
The data is preliminary. It comes from one study, on U.S. sessions, in early 2026. The directional finding is consistent with what we are seeing in client testing — but the specific percentages will move.
For most SMBs, the right response is not to chase AI Mode (where you cannot directly buy ads or guarantee citations). It is to harden AI Overviews coverage, then experiment with AI Mode opportunistically.

The Two Surfaces, In One Paragraph Each

AI Overviews sits above a traditional results page. When you ask Google a question, the AI Overview is generated at the top, followed by the regular blue-link results, the ads, the map pack, and everything else. The user is still inside the traditional Google environment, with an extra summary at the top. They can read the summary, click the cited sources, scroll down to the regular results, compare a few options, and decide. Per Google's AI features documentation, the design intent is to make complex queries easier to start, not to replace the rest of the SERP.

AI Mode is a conversational surface. Per Google's AI Mode help documentation, the user opts into AI Mode (or arrives via a dedicated tab), asks a question, and gets a complete, conversational answer in-thread. They can follow up. They can refine. The traditional results list is not the primary affordance — the conversation is. This is a different product, and the user knows it.

That structural difference produces the behavioral difference. Different UX, different expectations, different behavior.

What the Clickstream Solutions Study Actually Found

The study design, per the Search Engine Land report, analyzed about 846,000 U.S. Google search sessions captured in February and March 2026, using anonymized clickstream data from Surfer SEO. The Clickstream Solutions team, led by Eric Van Buskirk, has done prior work in this space — referenced in the report are earlier studies in May 2025 (a 70-user think-aloud study), October 2025 (a 250-session AI Mode study), and April 2026 (185 high-stakes purchase sessions).

The headline findings:

Behavior	AI Overviews Present	Comparison Condition	Source
Reverse (backward) scrolling share	47.5% of all scrolling	27% without AI Overviews	Clickstream Solutions
Cursor scatter across viewport	~83% of viewport area	Less when no AI Overview	Clickstream Solutions
Navigational cursor scatter score	27.5	19.7 baseline	Clickstream Solutions
Time-on-page spread by intent type	6-point spread	20-point spread without AIO	Clickstream Solutions
Brand-search active session at 21s	45.8% still active	12% without AIO	Clickstream Solutions

For AI Mode specifically:

88% of AI Mode tasks involved accepting the AI's shortlist as-is.
74% of AI Mode tasks selected the first-ranked item in the AI's response.
64% of AI Mode tasks ended with no outbound click at all.

We will note up front: this is one study, on U.S. sessions, over a short window. The directional finding — AI Mode is a closed loop and AI Overviews is a comparison environment — is consistent with what we see across client testing. The specific percentages will move. Treat the numbers as orienting, not as benchmarks to target.

How to Read the Data Without Overclaiming

A few honest caveats before the strategy section.

Selection effect. The users who have opted into AI Mode are not a random sample of all searchers — they have specifically chosen the conversational surface. Their willingness to accept the AI's shortlist may partly reflect their willingness to use AI Mode in the first place. As AI Mode rolls out more broadly, the click-rate behavior could shift.

Intent mix. The study covered “tasks,” but a mix of informational, transactional, and navigational queries will produce very different click patterns. The 64% no-click figure in AI Mode is unsurprising for informational queries (the answer is on screen, the user got what they wanted, no further click needed). It would be more surprising for transactional queries. The aggregate number obscures that mix.

The metric is behavior, not value. A user who reads an AI Overview, gets the answer, and does not click is not necessarily a lost user — they may have learned the thing they came to learn and walked away satisfied. That has real value for the publisher of the cited source (brand impression, trust deposit) even when no click registers. The economics of zero-click are not zero, just different.

With those caveats noted, the directional reading is solid: the two surfaces produce meaningfully different behavior, and that behavior is what should drive your content and ad strategy.

A large monitor showing an abstract heat-map style visualization of mouse and scroll activity on a search page, with bright clusters at the top and middle, viewed in a small office.

What This Changes About Your Content Strategy

Start with the obvious split, then refine.

Content patterns that win in AI Overviews

AI Overviews users compare. They scroll back. They open multiple sources. Their cursor moves across most of the viewport, suggesting active scanning rather than passive reading. The content patterns that win this environment:

Listicle-style structures. Numbered or labeled options with a clear ranking or differentiation cue. The Surfer/Clickstream finding that listicles dominate AI engine citations is consistent with this — we wrote about it in The Content Formats AI Engines Cite Most.
Comparison tables. When the user is in evaluate-and-compare mode, a table that does the comparison work for them is the most useful thing on the page. The page does not have to be a “best X vs Y” listicle — even a single product or service page benefits from a comparison table inside it.
Visible structure and intermediate headings. Reverse-scrolling means users are skipping around, then coming back. Headings act as navigational anchors. A long page with no internal structure forces them to re-read; a well-anchored page lets them jump back to the right section.
Concise, quotable lead paragraphs. AI Overviews increasingly pull short quotes directly from the cited source. The first paragraph of every section is the lift candidate.

What does not work as well in this environment: deep narrative essays, single-track storytelling, anything that requires the user to read top to bottom to get value.

Content patterns that win in AI Mode

AI Mode users accept the answer in-thread. They follow up with refinement questions. They do not typically click out unless the answer is incomplete. The content patterns that win this environment look different:

Deeply complete answers to specific questions. An AI Mode user wants a single, comprehensive response. If your content is the source the AI lifts that response from, you win — even if the user does not click. Because AI Mode reasoning often pulls from multiple sources, being one of the cited sources is a meaningful brand signal.
Conversational structure with implicit follow-up handling. Content that anticipates the second and third question the user is likely to ask, and answers them inline, gets re-cited as the user asks follow-ups. Content that answers only the literal question gets surfaced once and then dropped.
Definitive framing. AI systems are looking for sources that can be quoted authoritatively. Hedged, qualified, every-possible-caveat writing gets paraphrased away. Direct, falsifiable claims (with the data to back them) get quoted.
Visible expertise markers. Author bio, credentials, named expertise. AI Mode disproportionately surfaces sources it can attribute to a specific named expert, partly for citation reliability and partly for user trust.

We covered the underlying authority mechanics in The 4 Signals That Define AI Search Visibility in 2026.

Should you write two versions of every page?

Mostly no. Most pages should be optimized for AI Overviews first because that is where most SMB traffic still lives — AI Mode adoption is real but not yet at the volume that justifies separate content for it. The right model is to write the AI Overviews version (skimmable, comparison-friendly, quotable) and audit it for AI Mode readiness (does the lead paragraph answer the question completely, are follow-ups anticipated, is the author credentialed).

There are categories where the AI Mode version deserves its own page. Highly conversational, advice-heavy topics — “how do I decide between two providers,” “what should I expect during X process” — work better as standalone depth-first content. For most service-page and product-page content, the unified-with-audit model is enough.

An overhead shot of two printed page layouts side by side on a desk — one styled as a numbered listicle with short bullets, the other as a long-form conversational article — connected by a hand-drawn arrow.

What This Changes About Your Ad Strategy

Honest disclosure up front: ad formats inside AI Mode are still rolling out and the buying surface is not fully open. Most of what we say below is forward-looking, based on what Google has announced and what we are seeing in beta surfaces.

What changes in AI Overviews ad performance

AI Overviews push paid ads further down the page. We covered the mechanics in How AI Overviews Are Reshaping Paid Search. The behavioral data adds nuance: because AI Overviews users compare and scroll, ads that appear below the Overview can still get attention — they are not invisible, just postponed. The 47.5% backward-scrolling figure means users return to upper SERP positions after exploring lower ones, which means the ad slot below the Overview is now a comparison-mode slot, not a top-of-funnel slot.

Practical implication: ad copy that reads as one option among many (“compare local HVAC repair quotes,” “see plumbing options near you”) will outperform ad copy that demands a click without context. The user just read a comparative AI summary; they are still in comparison mode.

What is changing in AI Mode ad surfaces

Google has been previewing conversational ad formats in AI Mode for several months — small, in-thread sponsored callouts that appear inside the AI's response. The economics are still being worked out. Two implications worth tracking:

Click-through metrics will look terrible by traditional standards. With 64% of AI Mode sessions ending no-click, sponsored slots in AI Mode are going to be measured against a baseline where most users do not click anything. Optimizing for CTR in that environment is a mistake. The right metrics are brand-impression value, share of conversational mentions, and assisted conversion lift.
The Performance Max-style automation that has dominated Google Ads for two years is a poor fit for conversational surfaces. Performance Max optimizes across surfaces using opaque targeting. AI Mode rewards contextually-tight messaging. We expect Google to publish a specifically conversational ad format — and the SMBs who are ready to test it on day one will have an advantage measured in months, not weeks.

If you are running ads today and want a baseline that holds through the transition, our recommendation is: keep your current search campaigns active, add a small experimental budget for any AI Mode beta format your account gets access to, and measure both against assisted-conversion lift rather than last-click attribution. The economics of conversational search are too different from keyword search to be evaluated with the same yardstick.

A marketer at a standing desk reviewing an ads dashboard on a large monitor showing two campaign panels labeled with short generic terms, in a softly lit office.

Fort Wayne and Northeast Indiana: The Vertical-Specific Reads

A wide street view of an Auburn or Fort Wayne Midwestern downtown at golden hour, showing a row of small businesses including what could be a service garage, a dental practice storefront, and a law office facade.

For our Northeast Indiana clients, here is how the AI Overviews / AI Mode split looks across three common verticals. The point is not to be exhaustive — it is to show what “design content for one surface, audit for the other” looks like in practice.

Fort Wayne HVAC and home services. The dominant query pattern is “best HVAC company in Fort Wayne” and variations. AI Overviews users on this query scroll through the cited sources, then look at Google Business Profile listings below. AI Mode users get a shortlist of 3-5 names and take the top one ~74% of the time (per the Clickstream data). The content strategy: a clean listicle-style page that explains how to choose an HVAC provider in Fort Wayne (for AIO), with a strong “Why we are different in DeKalb/Allen County” depth section (for AI Mode). The ad strategy: search ads stay; conversational copy that reads like one option among many; map-pack hygiene matters more than ever because AIO users keep scrolling down to it.

Fort Wayne dental practices. Dental queries split sharply between “find a dentist” (transactional) and “is X procedure safe / how much does X cost” (informational). The informational queries are AI Mode's sweet spot — users want a complete answer in-thread, accept the dental practice cited as the source, and may never click. The transactional queries still run through AIO and the map pack. Content strategy: deep, specific, FAQ-style content on procedures and pricing (for AI Mode citation), plus a clear “schedule a consultation” path on the practice site (for AIO transactional). HIPAA considerations apply throughout — same as we covered in Fort Wayne Healthcare Reviews: HIPAA Meets Local SEO.

Fort Wayne legal practices. Legal queries are almost all informational on the front end (“what should I know about X claim”) and transactional on the back end (“attorney near me”). AI Mode is increasingly the discovery surface for informational legal questions, which raises the value of standalone-thorough content on common claim types. AI Overviews still drives the transactional “attorney near me” pattern. Content strategy: a comprehensive, conversational practice-area page per claim type (for AI Mode), plus a tight, schema-marked attorney-bio and contact path (for AIO and the local pack). The ad strategy is conventional — legal is still a search-ads-and-Local-Service-Ads category — but the content layer is where AI Mode citation lives.

The throughline across all three verticals: the same business benefits from optimizing for both surfaces, but the specific page types are different. Build the listicle-style discovery content for AIO. Build the depth-first practice content for AI Mode. Audit each for the opposite surface. Watch the preferred sources feature carefully because it will change which sources get re-surfaced in your local AI search results.

How to Baseline Your Own AIO vs AI Mode Visibility This Week

A four-step audit you can run yourself:

List your top 20 queries — the ones you most want to win in your category and geography.
For each query, run AI Overviews in a fresh Google session and note: do you appear, where, with what citation text. Screenshot everything.
For each query, run AI Mode in a fresh session and note: do you appear in the cited sources, how does the AI describe you, what follow-up questions does the user surface get nudged toward.
For each query, compare your appearance pattern to the top three competitors. Where you appear in AIO but not AI Mode (or vice versa), that is a content gap. Where you do not appear in either, that is a deeper authority problem.

We have a longer version of this audit, with template tracking sheets, baked into our AEO Service work. For the orientation to the broader May 2026 Google search changes, the 2026 Search Overhaul piece is the right starting read.

A printed audit tracking sheet on a clipboard with 20 numbered rows and four columns labeled query, AIO, AI Mode, and notes, partially filled with handwritten check marks and short notes.

How Button Block Helps

We are an AI-powered digital agency in Auburn, Indiana, working with small and mid-size businesses across Fort Wayne, Northeast Indiana, and the broader Midwest. The AIO-vs-AI Mode split is one of the live questions we are working through with clients this quarter — what to keep doing, what to stop doing, what to start doing. If the audit above is producing more questions than answers for your business, get in touch. We will tell you which surface deserves your time and budget first, and we will tell you honestly when the right answer is “neither, fix something else.”

Ready to Map Your AIO and AI Mode Visibility?

Button Block builds AEO content strategies and runs the baseline audits that show where your brand currently appears across both Google AI surfaces.

Explore Our AEO Service Get a Free Visibility Audit

Frequently Asked Questions

Both stick around, at least through 2026. They serve different jobs. AI Overviews lives in the standard SERP for users who want a fast summary plus the ability to compare links. AI Mode is a separate conversational surface for users who want a complete answer in-thread. Google has been clear (in product communications and at Google Marketing Live 2026) that the two are complementary, not replacements. Most users will see both — AIO on standard searches, AI Mode when they specifically opt in.

Two reasons. First, being cited inside an AI Mode answer is itself a brand asset — the user sees your name and a quote, even if they do not click. That deposits trust the same way being quoted in a major publication does. Second, the 36% of users who do click in AI Mode tend to be much higher intent. They have read the AI's full answer and decided your specific source is worth a deeper look. The conversion rates on those clicks, in our client testing, are markedly better than equivalent traffic from standard search.

For most SMBs, not yet — the surface is still rolling out, ad formats are in beta, and the buying interface is incomplete. The right posture is to keep your existing search campaigns running, get on the waitlist for any AI Mode ad format Google offers your account, and prepare ad creative that reads as conversational (one option among several, contextually appropriate, not interruptive). When the format opens up at scale, the SMBs who have already drafted the creative will be ready in days, not months.

The honest answer: with effort. Google does not yet provide a clean dashboard for AI Mode appearance. The practical method is to run manual prompt tests — a list of 20-50 queries you care about, run regularly through AI Mode, with notes on whether your domain appears in the citation strip and how the AI describes your content. Several third-party tools (BrightEdge, Profound, Otterly, and similar) have rolled out AI Mode tracking in early 2026; results vary by tool and by category. Manual baseline plus one third-party tool is the realistic stack.

Preferred sources is a user-side affordance that highlights chosen sources in both AIO and AI Mode. It will likely amplify the behavior pattern documented in the Clickstream study rather than change it: AIO users who have flagged preferred sources will gravitate toward those highlighted citations during comparison; AI Mode users will be more likely to see and accept answers grounded in their preferred sources.

For most Fort Wayne SMBs, the answer is roughly 70/30 in favor of AI Overviews work for now, then re-evaluate every quarter. AIO is where the majority of local discovery still happens — "best HVAC in Fort Wayne," "dentist near Auburn," "DeKalb County attorney" — and listicle-style, comparison-friendly content shows up there first. AI Mode is the right place to invest the remaining 30% on depth-first content (procedure FAQs, decision guides, practice-area explainers) that gets cited in the conversational thread. The split should bend toward AI Mode in regulated and informational categories.

Probably not exactly. The current AI Mode user base is self-selected — people who have opted in are likely more comfortable accepting an AI's answer at face value. As AI Mode rolls out to more users by default, the click rate will probably tick up (more skeptical users will click to verify), but the directional difference between AIO and AI Mode behavior is likely to persist because it follows from the UX of each surface, not just from the user base.

Is AI Mode replacing AI Overviews, or do both stick around?: Both stick around, at least through 2026. They serve different jobs. AI Overviews lives in the standard SERP for users who want a fast summary plus the ability to compare links. AI Mode is a separate conversational surface for users who want a complete answer in-thread. Google has been clear (in product communications and at Google Marketing Live 2026) that the two are complementary, not replacements. Most users will see both — AIO on standard searches, AI Mode when they specifically opt in.
If 64% of AI Mode users do not click, why bother optimizing for AI Mode at all?: Two reasons. First, being cited inside an AI Mode answer is itself a brand asset — the user sees your name and a quote, even if they do not click. That deposits trust the same way being quoted in a major publication does. Second, the 36% of users who do click in AI Mode tend to be much higher intent. They have read the AI's full answer and decided your specific source is worth a deeper look. The conversion rates on those clicks, in our client testing, are markedly better than equivalent traffic from standard search.
Should I be running ads in AI Mode now?: For most SMBs, not yet — the surface is still rolling out, ad formats are in beta, and the buying interface is incomplete. The right posture is to keep your existing search campaigns running, get on the waitlist for any AI Mode ad format Google offers your account, and prepare ad creative that reads as conversational (one option among several, contextually appropriate, not interruptive). When the format opens up at scale, the SMBs who have already drafted the creative will be ready in days, not months.
How do I know if my pages are showing up in AI Mode?: The honest answer: with effort. Google does not yet provide a clean dashboard for AI Mode appearance. The practical method is to run manual prompt tests — a list of 20-50 queries you care about, run regularly through AI Mode, with notes on whether your domain appears in the citation strip and how the AI describes your content. Several third-party tools (BrightEdge, Profound, Otterly, and similar) have rolled out AI Mode tracking in early 2026; results vary by tool and by category. Manual baseline plus one third-party tool is the realistic stack.
What is the relationship between the AIO/AI Mode behavior split and the new "preferred sources" feature?: Preferred sources is a user-side affordance that highlights chosen sources in both AIO and AI Mode. It will likely amplify the behavior pattern documented in the Clickstream study rather than change it: AIO users who have flagged preferred sources will gravitate toward those highlighted citations during comparison; AI Mode users will be more likely to see and accept answers grounded in their preferred sources.
How should a Fort Wayne or Northeast Indiana small business split its content time between AI Overviews and AI Mode?: For most Fort Wayne SMBs, the answer is roughly 70/30 in favor of AI Overviews work for now, then re-evaluate every quarter. AIO is where the majority of local discovery still happens — "best HVAC in Fort Wayne," "dentist near Auburn," "DeKalb County attorney" — and listicle-style, comparison-friendly content shows up there first. AI Mode is the right place to invest the remaining 30% on depth-first content (procedure FAQs, decision guides, practice-area explainers) that gets cited in the conversational thread. The split should bend toward AI Mode in regulated and informational categories.
Will this user behavior pattern hold as AI Mode adoption grows?: Probably not exactly. The current AI Mode user base is self-selected — people who have opted in are likely more comfortable accepting an AI's answer at face value. As AI Mode rolls out to more users by default, the click rate will probably tick up (more skeptical users will click to verify), but the directional difference between AIO and AI Mode behavior is likely to persist because it follows from the UX of each surface, not just from the user base.

Sources & Further Reading

Search Engine Land: searchengineland.com/users-behave-differently-in-ai-overviews-vs-ai-mode — Coverage of the Clickstream Solutions clickstream study (May 2026).
Clickstream Solutions: clickstreamsolutions.com/research — Eric Van Buskirk's ongoing AI search behavior research.
Surfer SEO: surferseo.com/blog — Provider of the anonymized clickstream data used in the study.
Google Search Central: developers.google.com/search/docs/appearance/ai-features — Official AI features documentation for Search.
Google: support.google.com/websearch/answer/14901683 — About AI Mode in Search.
Google Search Central: developers.google.com/search/docs/fundamentals/creating-helpful-content — Creating helpful, reliable, people-first content guidance.
Google Ads Help: support.google.com/google-ads/answer/10724817 — About Performance Max campaigns.
Schema.org: schema.org/Article — Structured data Article documentation.