The central argument here is uncomfortable for anyone who built their growth on informational SEO: in 2026, ranking well and being visible are no longer the same thing. The gap between impressions and clicks isn’t an analytics anomaly. It’s the shape of a structural change that’s been building since mid-2024 and is now moving faster, not slower.

Here’s what the current data shows, what the practitioners actually working inside this shift are saying, and — more usefully — what the specific actions are that correlate with winning rather than managing decline.

Section 01 What Is Happening Right Now — and Why It’s Different From Anything Before

Let’s be blunt about the timeline: Google AI Overviews went from appearing in 13% of queries to 48% of all searches in the space of a few months. For informational queries — guides, how-tos, definitions, comparisons — coverage exceeds 70%. Meanwhile, Google AI Mode has reached 75 million daily active users, and 93% of those sessions end without a single click.

That’s the landscape. And it’s not just a traffic story. It’s a monetisation story. Ads now appear in 25.5% of AI Overview results, up from 5.17% in early 2025. Google isn’t showing summaries out of altruism. It’s building a walled garden where commercial transactions happen before users ever visit a publisher’s site. Organic traffic isn’t being destroyed. It’s being rerouted.

Three things happened simultaneously to create the current pressure. AI Overviews scaled from limited to pervasive. The March 2026 core update penalised aggregator-style content that synthesises third-party information. And buyer research began migrating to AI chat interfaces before users reach a search engine at all. ChatGPT is now the fourth most visited website globally, pulling over 5 billion monthly visits. Nearly 35% of Gen Z already use AI chatbots as their primary search tool.

None of these are temporary. They’re three interlocking commercial trends that are accelerating together. The teams treating this as a phase are now 18 months behind the teams who responded to it as the new normal.

48% of all Google searches now show an AI Overview Digital Applied, March 2026
−58% CTR drop at position 1 when AI Overview present Ahrefs, December 2025
93% of Google AI Mode sessions end with zero clicks Semrush, September 2025
+35% more organic clicks if cited inside an AI Overview Seer Interactive / Digital Applied

Section 02 The Data — What the Numbers Actually Show (and What They’re Missing)

The evidence base has gotten stronger since mid-2025. What started as directional signals from small samples is now corroborated across multiple large-scale, named-methodology studies. But the aggregate numbers hide important variation — and that variation is where the strategy actually lives.

The CTR collapse: the numbers that changed the most

In April 2025, Ahrefs measured a 34.5% CTR decline at position one when AI Overviews appeared. By December 2025, the same methodology produced a 58% decline. That near-doubling in eight months is the scariest data point in current SEO literature — not because of the absolute number, but because of the trajectory.

Pew Research’s 68,000-query controlled study of 900 real U.S. adults found an 8% click rate with AI Overviews present versus 15% without. That’s a 46.7% relative reduction — still the most methodologically sound headline figure because it uses actual browsing behaviour rather than impression sampling. The Amsive study across 700,000 keywords found a 15.49% average decline, which appears conservative but covers a much wider query mix including branded and navigational searches that are less AI-affected.

Here’s the more actionable version of this data, though: only 1% of searches with an AI Overview lead to a user clicking a link within the summary itself. And 26% of users end their session entirely after reading an AI answer, versus 16% on pages without one. Users are getting what they need and simply stopping.

AI Overview appearance rate by query type — 2026
Informational / how-to
70%+
B2B Technology queries
82%
Question-based queries
60%
Commercial / comparison
8.7%
E-commerce / transactional
4%
Navigational
~3%
Sources: WebFX 2026 benchmarks · Semrush · JetDigitalPro

That chart tells you where the safe ground is. E-commerce and transactional queries are relatively protected — 4% AI Overview exposure means traditional rankings still deliver most of the value. B2B tech at 82% exposure is essentially the opposite. If your business runs on “what is X” and “how to do Y” content in a technical category, you are in the highest-exposure segment in the ecosystem.

The citation gap: the number most teams haven’t noticed yet

In July 2025, approximately 75% of pages cited in AI Overviews also ranked in the organic top 10. By early 2026, that overlap had dropped to 38%. That’s a 37-percentage-point collapse in seven months in the correlation between ranking and citation.

What replaced it? Content depth is emerging as the dominant citation driver. Pages above 20,000 characters average about 10 AI citations each. Pages under 500 characters average 2.4. Structured content also significantly outperforms: comparison pages with 3 tables earn 25.7% more citations, and validation pages with 8 list sections earn up to 26.9% more.

Key metrics — Without AI Overview vs. With AI Overview, 2026
Metric Without AI Overview With AI Overview Source
Organic CTR, position 1 ~15% ~8% Pew Research
Position-1 CTR drop Baseline −58% (Dec 2025 data) Ahrefs
CTR boost: cited in AIO Baseline +35% organic / +91% paid Seer Interactive
Top-10 ↔ AIO citation overlap ~75% (July 2025) 38% (early 2026) Ahrefs
AI Overview query coverage <5% (2024) 48% all queries (Mar 2026) Digital Applied
Ads in AI Overview results 5.17% (early 2025) 25.5% (Q1 2026) Digital Applied
Session end rate after AI answer 16% 26% Pew Research
AI Mode zero-click rate n/a 93% of sessions Semrush
Sources: Pew Research Center (July 2025) · Ahrefs (2025–2026) · Seer Interactive · Digital Applied (March 2026) · Semrush (September 2025)

What the numbers are genuinely missing

Traffic from AI platforms — ChatGPT, Perplexity, Gemini — is growing fast but remains approximately 1/34th the size of traditional SEO traffic. GEO (Generative Engine Optimization) traffic grew 527% year-over-year, which sounds transformative until you see that baseline. The channels are additive, not substitutional — yet. The risk is ignoring them while they’re still small, only to find them significant 18 months from now when optimising retroactively is harder.

There’s also a measurement gap nobody has cleanly solved. Most GA4 setups don’t capture AI referral traffic reliably. You need to manually filter for chatgpt.com, perplexity.ai, and gemini.google.com as referral sources. Searches influenced by AI — someone who reads a Perplexity summary then types a branded query directly into Google — don’t appear in any analytics as AI-influenced at all.

Measurement Action

Set up a dedicated UTM segment or referral filter in GA4 now to isolate AI-sourced traffic. Include: chatgpt.com, perplexity.ai, gemini.google.com, claude.ai, and bing.com (for Copilot). This instrumentation is an urgent priority — waiting for the data to appear on its own means you’re already flying blind.

Section 03 What the People Actually in the Data Are Saying

There is a clear consensus among practitioners who work with live data: the KPI for SEO and website growth has shifted from rankings to retrievability. But there’s meaningful disagreement about pace, severity, and what to prioritise first.

Lily Ray (Amsive): the brand signal shift

Lily Ray’s consistent position since late 2025: clicks will decline, brand visibility becomes the new KPI, and third-party corroboration is now an AI trust signal, not just a ranking signal.

That last point is genuinely underappreciated. AI Overviews don’t cite a source because it ranks — they cite it because independent sources corroborate what it says. Backlinks always functioned as trust proxies. They now simultaneously function as AI corroboration signals. The mechanism is the same; the consequence has expanded. If your domain earns no external citations, an AI system has no corroboration for your claims. You’re effectively invisible to it regardless of your ranking position.

Her January 2026 analysis also found sharp visibility declines across brands publishing high volumes of self-promotional listicles — specifically articles ranking the company’s own product as the best solution. Some had published over 200 such pieces. The teams who built the most aggressive 2022-era content operations are now managing the largest remediation projects.

Yamini Rangan (HubSpot): the bluntest data point in this cycle

HubSpot CEO Yamini Rangan was direct about it: HubSpot lost between 70–80% of organic traffic between November 2024 and Q2 2025. She attributed it explicitly to AI Overviews answering queries before users clicked through.

“AI Overviews are giving answers, and fewer people are clicking through to websites.” — Yamini Rangan, CEO, HubSpot, Q2 2025

The HubSpot case is instructive because they had one of the most sophisticated SEO operations in B2B. The lesson isn’t that SEO failed — it’s that a content strategy built almost entirely on broad informational keywords has a structural ceiling when AI summarises those exact topics at the top of every relevant SERP. That’s the content model most SaaS blogs still operate on today. If you’re running it, you’re looking at the same structural exposure HubSpot faced, just smaller.

Cyrus Shephard: what the winning sites actually look like

Cyrus Shephard’s site-level analysis of 400+ domains found organic traffic winners were measurably more likely to: offer a product or service users can transact with directly, let users complete tasks on-site, own proprietary data nobody else can replicate, maintain tight topical focus, and show stronger branded search signals. Interesting that branded search made the list — it suggests that off-site awareness work is now defensively necessary, not just aspirationally nice-to-have.

Kaspar Szymanski: the useful dissent

Former Google engineer Kaspar Szymanski argues that the fundamentals — brand recognition, user trust, site performance, technical crawlability — will remain consistent regardless of how AI surfaces change. Worth holding seriously. Every major SEO disruption cycle has eventually settled into a landscape where fundamentals reasserted dominance. The risk is assuming that stabilisation will come quickly enough to justify waiting. Based on current trajectory, that bet looks wrong.

Section 04 GEO: What Generative Engine Optimization Actually Is and Why It’s Not a Replacement

GEO — Generative Engine Optimization — is the practice of structuring content to appear in AI-generated answers rather than traditional ranked results. It’s the fastest-growing new discipline in search marketing and it’s also the most misunderstood.

Here’s the thing most coverage gets wrong: GEO traffic grew 527% year-over-year. That number will be in every deck you see this quarter. What those decks often omit: GEO traffic is still approximately 1/34th the size of traditional SEO traffic. You cannot pivot your entire programme to GEO and expect equivalent reach. Not yet.

The Two-Surface Visibility Model — 2026
SURFACE 01
Traditional Rankings
Blue links. Position 1–10. Still ~34× larger than AI traffic. Core technical SEO, E-E-A-T, authority still drives this.
SURFACE 02 — FAST GROWING
AI Citation Layer
AI Overviews, AI Mode, ChatGPT, Perplexity. Citation drives +35% organic CTR. Requires different signals than ranking.
THE OUTCOME
Combined Visibility
Sites cited inside AIOs while also ranking earn disproportionate traffic. Neither channel alone is sufficient.

What GEO actually requires — and this is where it intersects with good SEO rather than replacing it:

Content depth over breadth. Pages above 20,000 characters average 10 AI citations. Under 500 characters: 2.4. This isn’t a license to pad word counts — it’s a correlation with genuine topical comprehensiveness that AI systems reward. Freshness above 90 days is now a meaningful signal; content not updated within that window saw 20–40% traffic losses in post-March 2026 analysis. Structured formatting — comparison tables, numbered lists, FAQ sections — earns 18–27% more citations than equivalent unstructured content.

And here’s one that most teams aren’t tracking yet: 44.2% of all LLM citations come from the first 30% of a page’s text. Your introduction and opening argument are doing more citation work than your conclusion. If your first 500 words are setup and context-building rather than direct answers, you’re ceding the most citation-active real estate on the page.

The llms.txt Signal

Similar to robots.txt, llms.txt is a new file that tells large language models how to interact with your site — what to crawl, what to prioritise, what to exclude. Technical SEO benchmarks in 2026 now include it alongside HTTPS, clean robots.txt, and structured data as a baseline requirement for AI crawlability. If your site doesn’t have one, AI systems have no guidance on how to treat your content. Add it to the technical audit checklist before the next update cycle.

Section 05 The March 2026 Core Update — What 600,000 Pages Revealed

Google completed the March 2026 core update on April 8 — the first broad core update of the year, arriving close behind a spam update and a February Discover update. Three distinct quality assessments inside six weeks means your current visibility reflects a compounded signal, not a single algorithmic swing. If your traffic moved significantly in Q1, look at all three.

The JetDigitalPro analysis that synthesised Ahrefs, Semrush, and Originality.ai data across 600,000+ pages between December 2025 and March 2026 produced the clearest picture of winners and losers:

  • ↓ Affiliate marketing sites: 71% saw negative ranking impact
  • ↓ Generic AI-generated content without human editorial oversight: 60–80% traffic drops
  • ↓ Content not updated within 90 days: 20–40% traffic losses
  • ↑ Sites with original data and case studies: +22% visibility gains
  • ↑ Pages with detailed author bios and verifiable credentials: 72% of top-ranking sites now include them

One data point that cuts through the noise: there was a near-zero correlation (0.011) between AI use in content creation and ranking penalties. Google is not penalising AI-assisted content. It is penalising content that lacks unique insights regardless of how it was produced. The distinction is the output, not the method. Teams spending energy on AI detection avoidance are solving the wrong problem.

2023 SEO playbook — losing ground
  • Rank for informational keywords at scale
  • Content volume over topical depth
  • Rankings + impressions as success KPIs
  • Broad keyword-targeted long-form articles
  • Comparison roundups with affiliate structure
  • Self-promotional “best of” listicles
  • Annual content strategy review cycles
2026 citation-first strategy — working now
  • Optimise for AI citation and traditional ranking simultaneously
  • Proprietary data + first-hand author experience
  • Measure brand visibility, citation rate, and conversion quality
  • Tight topical authority + transactional capability
  • Original research with named methodology
  • Content requiring a click to complete a genuine task
  • Quarterly refresh cadence with llms.txt and schema maintenance

Section 06 The 6-Question Citation Readiness Framework

Before publishing any page — or deciding whether to revise, consolidate, or cut an existing one — run it through these six questions. They’re derived from the patterns that consistently correlate with AI Overview citation across the current literature, cross-referenced against the March 2026 update losers.

Citation Readiness Evaluation — 2026 Edition
  1. Does this page enable a user to complete a real task or make a real decision — not just learn about a topic? AI explains topics. Your page needs to do more than that to earn a click. If the only value is informational and the information can be summarised in four sentences, your click moat is structurally zero.
  2. Does the author have verifiable first-hand experience — demonstrable, not just claimed? The first E in E-E-A-T is Experience. That means the author has done the thing, and evidence exists beyond their own assertions. Author bio with credentials, case studies they ran, screenshots of results — these are the proof layers AI systems use to establish trust.
  3. Is your content independently corroborated by external sources? Content that earns zero external citations signals to AI systems that its claims are uncorroborated. Backlinks are now simultaneously ranking signals and AI corroboration signals. A page with no external references is a page an AI has no reason to trust — rank notwithstanding.
  4. Does the page contain something AI cannot replicate from existing training data? Unique proprietary data. An original study with a named sample and methodology. A first-person experience from a specific project. Something that requires you specifically to have produced it. If not, the AI simply won’t cite you because there’s nothing uniquely yours to cite.
  5. Is your direct answer in the first 30% of the text? 44.2% of all LLM citations come from a page’s opening third. Setup paragraphs, context-building, and “in this article we will cover” structures push the citable content below the threshold where AI systems focus. Answer first. Build context second.
  6. When was this page last updated? Content not refreshed within 90 days suffered 20–40% traffic losses after the March 2026 update. Freshness is no longer a nice-to-have signal. Build it into the content ops calendar as a non-negotiable cadence.

Run this against your 20 highest-traffic informational pages. The ones with the biggest gap between impressions and clicks are your most exposed — and most likely to be reassessed in the next update cycle. Run a full E-E-A-T audit at ContentEvaluator.online →

Section 07 What Comes Next: Three Evidence-Based Scenarios for the Rest of 2026

The three plausible scenarios for the next 12 months don’t differ in direction — pressure on informational traffic continues in all three. They differ in speed and severity. That distinction determines how aggressively to reprioritise investment now.

Base Case · Most Likely

AI Overview expansion with steady quarterly click compression

AI Overviews move toward 60–65% of all queries by year-end. Informational content faces 3–5% quarterly traffic erosion. Zero-click searches reach 70% of all queries (Bain projects this trajectory). Manageable with the right adaptation strategy, but 2022–23 click volumes on educational queries are structurally unrecoverable.

Condition: Google maintains current rollout cadence. The commercial incentive — ads in 25.5% of AIO results — makes a significant pullback unlikely absent a major quality failure.

Upside Case · Possible

Expert-led and transactional content gains disproportionately

If Google’s quality detection keeps pace with synthetic content volume, sites with genuine expertise, proprietary data, and transactional capability gain sharply. The Bruce Clay case study (367% organic growth over 17 months, deep technical content) is the proof-of-concept. This path exists and is already delivering for early movers.

Condition: Google’s quality systems improve faster than synthetic content production scales. The 86.5% of top-ranking pages using AI assistance creates a differentiation opportunity for pages with genuine first-hand insight layered on top.

Downside Case · Monitor

AI Mode goes mainstream; click bifurcation deepens sharply

AI Mode already reports 93% zero-click sessions. If it crosses from early adopter (current: 75M daily users) to default research tool for majority users, the viable SEO opportunity narrows dramatically to branded, commercial-intent, and local queries. GEO becomes the primary discovery channel for informational content — at 1/34th current traffic volumes.

Condition: AI Mode international expansion hits a mainstream adoption threshold. It’s already in 200+ countries. Watch the US mobile usage data over Q2–Q3 2026 as the leading indicator.

The one action to take before conditions shift further

Pull your Search Console data and segment top-traffic pages by intent. For each informational page driving meaningful sessions, ask: does this page give Google a specific, citable reason to reference it inside an AI Overview — unique data, a direct first-hand answer, a proprietary methodology? If the answer is no, you don’t have a ranking problem. You have a citation strategy problem. They require different fixes. Start the citation audit at ContentEvaluator.online →

Section 08 The 2026 Action Checklist — What to Do in the Next 30 Days

Based on the patterns that consistently separate the March 2026 update winners from losers, here’s what to prioritise. Not everything. Not “build better content.” Specific, sequenced actions tied to evidence.

  • Add llms.txt to your site. Tell AI crawlers what to prioritise. This is a baseline technical requirement in 2026, not an advanced tactic.
  • Set up AI referral tracking in GA4 — filter for chatgpt.com, perplexity.ai, gemini.google.com, claude.ai. You cannot optimise what you can’t measure.
  • Identify your 20 highest-impression, lowest-CTR pages in Search Console. These are your AI Overview exposure candidates. The gap between impressions and clicks quantifies your citation problem.
  • Move your direct answer to the first 30% of the text on any page you want cited. Stop warming up. Start answering.
  • Add structured data (FAQ, HowTo, Article schema) to every page that lacks it. Structured formatting earns 18–27% more AI citations. Schema markup is the machine-readable layer that AI crawlers parse first.
  • Flag any page not updated in 90+ days. Content with no freshness signal suffered 20–40% traffic losses. Build a refresh calendar and treat it as non-negotiable.
  • Audit author bios for verifiable credentials. 72% of top-ranking post-March 2026 pages include detailed author biographies. Claimed expertise is not enough. Evidence of it must be visible on the page.
  • Identify one data asset you own that nobody else can replicate — a survey, internal benchmark, client dataset, original test. Build a page around it. This is the highest-leverage investment in AI citation strategy available to most teams right now.

Frequently Asked Questions

Why does the rankings-to-clicks decoupling matter so much right now?

Because the 25-year assumption underlying SEO investment — that ranking well produces proportional traffic — is no longer reliable for informational queries. Pew Research’s controlled study found CTR drops 46.7% when an AI Overview appears. Ahrefs now reports a 58% decline at position one, nearly double what they measured earlier in 2025. Teams still measuring primarily through rankings are systematically underreporting how much organic value is being intercepted before a user ever reaches their result. The gap between ranking and performance is the defining diagnostic challenge in 2026.

The directional trend is well-evidenced across multiple large-sample studies: Pew Research (68,000 real queries from 900 users), Ahrefs (300,000+ keywords), Seer Interactive (2.43 billion impressions across 53 brands), JetDigitalPro (600,000+ pages). Specific percentages should be treated as approximations — Google doesn’t provide direct AI Overview attribution in Search Console, and appearance rates vary significantly by geography and query type. The direction is reliable. The precise figure for any individual site requires first-party measurement in your own Search Console data, segmented by intent and AI Overview exposure.

Most exposed: mid-tier publishers, affiliate operations, SaaS content blogs, and B2B educational hubs whose traffic depends on broad informational keywords. HubSpot’s 70–80% organic traffic loss is the clearest large-scale example. Affiliate sites saw a 71% negative impact rate in the March 2026 core update analysis. B2B technology queries now trigger AI Overviews in 82% of cases — the highest exposure rate of any category tracked. Least exposed: transactional e-commerce sites (4% AI Overview rate), local businesses, brands with strong direct navigation, and organisations with proprietary data no AI can replicate.

GEO (Generative Engine Optimization) is optimising content to appear in AI-generated answers rather than traditional ranked results. GEO traffic grew 527% year-over-year in 2025–2026 — but it is still only approximately 1/34th the size of traditional SEO traffic. The answer to “instead of” is: no. The strongest 2026 strategy combines both channels. Traditional rankings still drive the vast majority of discovery traffic. AI citation optimisation captures the increasingly high-value top-of-SERP real estate and delivers 35% more clicks for cited brands. Neither alone is sufficient. You need both surfaces covered simultaneously, and the good news is that many of the signals that drive AI citation — depth, freshness, structured content, verifiable expertise — also correlate with stronger traditional rankings.

For algorithm analysis with live SISTRIX data: Lily Ray’s Substack and Amsive’s insights blog provide the most detailed post-update breakdowns. Search Engine Land and Search Engine Roundtable (Barry Schwartz) publish confirmed update timelines with minimal speculation. For AI Overview-specific CTR research: Ahrefs, Seer Interactive, and BrightEdge publish periodic large-sample studies. For first-party AI citation tracking: Semrush’s AI Visibility Toolkit and GA4 referral filtering for AI platforms. For original E-E-A-T analysis and content audit tools: ContentEvaluator.online.

Sources & Further Reading

Find Out Which of Your Pages Can Actually Earn a Citation

Run the E-E-A-T citation audit on your top informational pages. See exactly where the gap between impressions and clicks is — and what’s fixable before the next core update cycle.

Start Free Audit at ContentEvaluator.online → See how it works