Quick Takeaways (read this first)

• Assume the AI answer can be fully expanded by default. Your blue link may start below the fold even when you rank top 3.
• Build two layers on every target page: (1) an 80–160 word “citation block” Google can safely lift, and (2) a “post-answer payoff” asset that makes scrolling rational after the answer is complete.
• Rewrite intros for extractability: short sentences, explicit entities, constraints, and numbers—so the model can quote you without ambiguity.
• Measure impact with a dual lens: Search Console’s generative AI controls/reporting + behavioral analytics (scroll depth, engaged time, assisted conversions).
• Run a 7-day sprint: recon 20 queries → update 5 pages → instrument scroll events → validate citations + clicks.

What changed: from “collapsed answer” SEO to “fully expanded answer” SEO

Historically, a collapsed AI Overview still left room for classic organic results to compete for attention. The user had to opt into more AI content by clicking “Show more.” Now, Google says AI Overviews can dynamically expand to their full state for some queries, removing that friction and pushing organic results further down the page.

Two practical implications follow:

1. Click probability drops because the user sees a complete answer without doing anything extra.
2. Your “rank #1 = win the click” model becomes less reliable, especially for definitional/how-to queries where the Overview satisfies intent.

This isn’t speculation; it’s based on Google’s confirmation reported by Search Engine Land: Google is dynamically expanding AI Overviews for some queries. Secondary commentary has also pointed out the practical outcome: effectively, the “Show more” moment is disappearing for those queries, changing how visibility is earned: Google removed the ‘Show More’ button on AI Overviews (commentary referencing confirmation).

Why “scroll-depth SEO” is the right mental model

When the AI answer is fully expanded, you’re competing with a complete response, not a teaser. So your job isn’t only “get clicked.” It’s:

• Get cited (become the grounding source inside the Overview)
• Earn the scroll (give the user a reason to move past the full answer)
• Convert post-answer intent (capture the “now what?” moment after the user understands the basics)

We’ve found teams win fastest when they design pages like a two-act play: Act 1: citation eligibility, Act 2: irresistible payoff.

The new page blueprint: citation block + post-answer payoff

Layer 1: The 80–160 word “citation block”

Your citation block is a self-contained passage that can stand alone in an AI Overview. It should read like something a careful editor would quote.

What a great citation block includes

• Definition + scope: what the thing is, and what it isn’t
• Constraints: the “depends on” caveats (region, platform, pricing tier, timeframe)
• Numbers: thresholds, ranges, time estimates, or step counts
• Named entities: tools, standards, roles, or system components
• Attribution cues: “based on Google Search Console data,” “per GA4,” “tested across X pages”

Good vs. bad citation block (realistic example)

Bad (too vague to cite):
“Scroll-depth SEO helps you get more visibility in AI Overviews by improving your content structure and user experience.”

Good (citable, bounded, specific):
“Scroll-depth SEO is the practice of engineering pages so they (1) contain citation-ready passages that AI Overviews can quote and (2) deliver a second-step asset immediately below the fold (e.g., a checklist or calculator). For informational queries where AI Overviews expand by default, this approach shifts success from ‘rank → click’ to ‘cite → scroll → convert.’ In practice, we aim for a 90–140 word passage with explicit entities, constraints, and at least 2–3 measurable details.”

Layer 2: The “post-answer payoff” (your scroll magnet)

If the AI answer is complete, the user needs a reason to keep going. The best scroll magnets are assets that the Overview can’t fully satisfy because they’re contextual, interactive, or decision-specific.

High-performing payoff assets we recommend:

• Decision trees (choose X vs Y with clear branching logic)
• Templates (copy/paste email, PRD, policy, prompt library)
• Calculators (pricing, ROI, time-to-value, capacity planning)
• Comparison tables (features, tradeoffs, who it’s for)
• Checklists (step-by-step implementation with acceptance criteria)
• Local/inventory modules (availability, shipping estimates, compatibility)

Placement rule: Put the payoff immediately after the citation block—often right below where the fold likely lands on mobile. Your goal is to make the next scroll feel like an obvious next step.

The 7-day “Scroll-Depth SEO” sprint (field-tested workflow)

This sprint is designed for in-house teams who need a measurable response quickly. You’ll prioritize the pages most harmed by auto-expanded Overviews, update them with the two-layer blueprint, and instrument scroll + conversion signals to see whether demand is being cannibalized or reshaped.

Day 1: SERP recon on your top 20 AEO queries

Pick 20 queries that historically drove high-intent informational traffic (SaaS: “best X for Y,” “how to implement Z,” e-commerce: “size guide,” “materials comparison,” publishers: “what is,” “why does”).

What to capture (do this per query)

1. Screenshot the SERP on mobile and desktop.
2. Record “organic displacement”: how far down your result appears when the Overview is expanded.
3. Note citation sources: which domains get cited, and what kind of passages are quoted (definitions, lists, steps, stats).
4. Classify intent: “definition,” “comparison,” “process,” “troubleshooting,” “pricing.”

Prioritization rule: Start with queries where (a) AI Overview is present and expanded, (b) your listing is pushed below the fold, and (c) the query still matters to pipeline or revenue.

Day 2: Build a “citation map” for 5 priority pages

For each page, identify 2–3 passages that could become citations. If you already have them, great—move them to the top and tighten them for extractability. If you don’t, you’ll write them.

Tip: If you’re already working through GSC’s generative AI reporting, pair this sprint with our measurement workflow in GSC Generative AI Report Went Global: 7‑Day AEO Sprint.

Day 3: Rewrite above-the-fold for extractability (without killing UX)

Above the fold is no longer “hook-only.” It’s now “hook + cite.” You want the first screen to be both human-friendly and model-friendly.

Extractable intro checklist

• Keep sentences under 20–22 words where possible.
• Use explicit nouns (avoid “this,” “it,” “they” without antecedents).
• Include one mini-heading that matches the query (“What is X?”, “X checklist”, “X vs Y”).
• Use a 3–5 bullet list for steps or criteria.
• Add 1–2 constraints (platform, region, timeframe) to reduce hallucination risk.

Common mistake

Writing a “brand essay” intro. If your first 200 words are positioning, history, or generic context, you’ve made it harder for Google to lift a safe, bounded answer. Put your perspective later—after you’ve earned the citation.

Day 4: Add the “second-question hook” + payoff asset

Right after the citation block, add a line that anticipates the user’s next question. This is your scroll trigger.

Second-question hook templates (copy/paste)

• “If you’re choosing between X and Y, here’s the exact comparison table we use (with decision rules).”
• “If you need to implement this in under 30 minutes, use the checklist below—each step has a pass/fail test.”
• “If your results still look wrong, jump to the troubleshooting section—these are the three failure modes we see most.”

Then deliver the asset: a table, checklist, calculator, or template. Don’t bury it 1,500 words down. Put it where the user lands immediately after the AI Overview experience ends.

Day 5: Instrument scroll depth + post-answer conversions

Search Console is still largely impression-led for generative surfaces, so behavioral analytics becomes your differentiator. You need to know: are users who land on the page actually reaching the payoff?

What to track in GA4 (minimum viable)

• Scroll depth events: 25%, 50%, 75%, 90% (or “reached payoff module” as a specific event).
• Engaged sessions and average engagement time.
• Micro-conversions: template downloads, tool usage, newsletter signups, “copy” button clicks.
• Assisted conversions from informational pages (especially for SaaS).

Also make sure you understand the new Search Console controls and reporting context. Google has documented a search generative AI control in Search Console Help: Search generative AI control. And Google’s own announcement on “new opportunities, control and insights for website owners” (updated Aug 31, 2026) gives the clearest framing for how site owners should think about visibility and control: New opportunities, control and insights for website owners.

If you need confirmation that the tooling is live and how it’s being observed in the ecosystem, this overview is useful: Google Search Console Generative AI Performance Report & AI Control Live.

Day 6: Validate citations manually (and log patterns)

For each of the 5 updated pages, spot-check the target queries daily for a week. You’re looking for:

• Was your domain cited?
• Which passage was used? (did it match your citation block?)
• What format won? (paragraph, bullets, steps, table summary)
• Who else got cited? (and what do they have that you don’t?)

Tip: if you’re seeing Google cite short highlighted passages more than whole pages, align with the passage-engineering approach in Google AI Mode Cites 117-Word Paragraphs: Optimize Them—the same “liftability” principle applies here.

Day 7: Compare pre/post impact and decide your scale plan

At the end of the week, you should be able to answer three operational questions:

1. Are we earning more citations? (manual checks + GSC generative AI context)
2. Are we earning deeper engagement? (scroll depth to payoff + engaged time)
3. Are we recovering business value? (micro + assisted conversions even if clicks drop)

If you see citations rising but clicks falling, that’s not automatically failure—it may mean your page is becoming the reference layer. Your next move is to strengthen the payoff so the users who do click become more valuable.

Three real-world examples: what “scroll-depth SEO” looks like in practice

Example 1 (SaaS): “What is SOC 2 Type II?” → template payoff

Problem: The AI Overview fully answers “what it is” and “why it matters,” pushing organic results down. Users who click are often looking for the next step: “what do I do now?”

Implementation:

• Citation block (120 words): Define SOC 2 Type II, clarify timeframe (typically a 3–12 month observation period depending on auditor and readiness), list the Trust Services Criteria, and state who needs it (B2B SaaS selling to mid-market/enterprise).
• Second-question hook: “If you’re planning a SOC 2 Type II in the next 90 days, use this readiness checklist to avoid rework.”
• Payoff asset: A downloadable readiness checklist + a simple “audit timeline calculator” (inputs: current controls maturity, headcount, auditor availability; outputs: estimated timeline ranges).

Why it works: The AI can summarize SOC 2, but it can’t credibly produce your operational checklist and calculator inputs without risking errors. You’ve created a post-answer reason to scroll and convert.

Example 2 (E-commerce): “Nylon vs polyester backpack” → comparison table + decision rules

Problem: Expanded Overviews often give a generalized comparison (durability, water resistance) that satisfies casual shoppers. Your product pages lose the “research click.”

Implementation:

• Citation block (100–140 words): Define nylon vs polyester in the context of backpacks, include measurable cues (denier ranges, typical tradeoffs), and list constraints (coatings, weave, and construction matter more than fiber alone).
• Second-question hook: “If you’re deciding for commuting vs hiking, use the decision table below—pick your use case and we’ll recommend the material + coating combo.”
• Payoff asset: A use-case table with decision rules (e.g., “daily rain + abrasion → nylon + PU coating,” “budget travel → polyester + thicker weave”), plus a module that filters in-stock products matching the rule.

Why it works: You turn a generic comparison into a guided purchase decision. Even after reading the Overview, the user still needs the “what should I buy?” layer.

Example 3 (Publisher/Media): “How to fix ‘DNS_PROBE_FINISHED_NXDOMAIN’” → troubleshooting flowchart

Problem: AI Overviews often list 6–10 fixes. That’s enough to stop many clicks. The remaining users are stuck and need a diagnostic path, not a list.

Implementation:

• Citation block (90–130 words): Define the error, identify primary causes (DNS cache, ISP DNS, host file, domain resolution), and include constraints (Windows/macOS/Chrome steps differ).
• Second-question hook: “If you’ve tried two fixes and it still fails, use this 90-second flowchart to identify the root cause.”
• Payoff asset: A flowchart: “Does it happen on mobile data?” → “Does it happen on another device?” → “Does nslookup resolve?” Each branch links to the exact fix with screenshots.

Why it works: AI lists are easy; diagnostics are hard. A flowchart reduces frustration and earns scroll depth and backlinks over time.

Engineering “citation-ready passages”: the technical writing rules that actually matter

Rule 1: Write passages that are safe to quote out of context

AI Overviews often quote or paraphrase small sections. If your passage depends on previous paragraphs to make sense, it’s fragile.

Actionable tip: Pretend your paragraph will be screenshotted and shared alone. Add the missing noun, the constraint, and the number inside the paragraph.

Rule 2: Prefer “bounded claims” over absolute claims

Absolute claims are risky for a model to cite (“X always works”). Bounded claims are safer (“In GA4, X typically indicates Y when Z is true”).

Good bounded pattern: “For most SaaS sites with 50k–500k monthly sessions, we see scroll-depth events become stable after ~7 days of data.”

Rule 3: Use micro-structure that models can lift

• Mini headings that mirror query language
• Bullets for criteria and steps
• Short definitions followed by one “why it matters” sentence

Rule 4: Put the “best paragraph” earlier than feels natural

Classic editorial pacing builds up to the answer. Scroll-depth SEO flips that: you lead with the extractable answer, then earn the scroll with payoff.

If you want a deeper grounding-oriented framework (especially as Google shifts documentation and emphasis), pair this with Google Quietly Killed Schema Docs: New AEO Grounding.

Common mistakes we’re seeing (and how to fix them fast)

Mistake 1: Chasing “AI Overview optimization” while ignoring click magnets

Many teams stop after rewriting a definition. That may earn citations, but it won’t recover business value if clicks decline.

Fix: Every updated page must ship with a payoff module (table, template, tool) located immediately after the citation block.

Mistake 2: Hiding the differentiator at the bottom

If your unique data, comparison, or template is 2,000 words down, users won’t reach it—especially after reading a full Overview.

Fix: Move the asset up. Treat it like your new “above the fold.”

Mistake 3: Measuring only clicks and rankings

Auto-expanded Overviews can reduce clicks even when you’re “winning” brand exposure and citations.

Fix: Add scroll depth + micro-conversions. Use GSC generative AI context for visibility, but use behavioral analytics to judge value.

Mistake 4: Writing content that’s easy to paraphrase but hard to cite

Fluffy intros, vague claims, and missing constraints make your content less quotable.

Fix: Tighten to 80–160 word citation blocks with explicit entities + numbers + boundaries.

FAQ: fast answers your team will ask

Should we try to opt out of AI Overviews?

Usually, no—unless you have a strong business reason (e.g., content licensing constraints, high cannibalization with no downstream conversion). Google has provided controls and documentation for site owners, so treat it as a strategic decision, not a reflex. Start by measuring: citations + scroll depth + assisted conversions before you consider opting out.

How long does it take to see changes?

For behavioral metrics (scroll depth, micro-conversions), you can see a signal within days. For citations inside AI Overviews, expect volatility; you’ll often need 1–3 weeks to see stable patterns, but the sprint is designed to produce directional evidence in 7 days.

Do we need new schema for this?

Schema can help clarify entities, but this shift is primarily a content engineering + UX placement problem. Your fastest wins come from extractable passages and payoff modules, not markup alone.

What if the AI Overview already answers everything?

Then you must change what “everything” means. Build the next-step layer: decision tools, templates, real data, localized availability, or implementation checklists with acceptance criteria. The Overview can summarize; it can’t replace your operational utility.

What to do next (action steps you can assign today)

1. Pick 20 queries where Overviews appear and your results are displaced below the fold.
2. Select 5 pages with business impact and high displacement.
3. Write one citation block per page (80–160 words; definition + constraints + numbers).
4. Add a second-question hook directly under the citation block.
5. Ship one payoff asset per page (template/table/calculator/checklist).
6. Instrument scroll depth and a payoff interaction event in GA4.
7. Run 7 days of spot-checks for citations and log which passages win.
8. Decide scale: roll the pattern across your top 50–100 informational pages.

Try our AEO workflow (dashboard + Chrome extension)

If you want to operationalize this as a repeatable system—query recon, citation block QA, and measurement loops—try the AEO tool dashboard. You can sign up here: https://aeotool.ai/register.

And if you want faster page-by-page checks while you’re doing SERP recon and content updates, install our Chrome extension: AEO Analyzer Chrome extension.