Branded web mentions correlate 0.664 with AI Overview appearances. Backlinks correlate 0.218. — Position Digital Research

Research from Position Digital found that branded web mentions correlate 0.664 with AI Overview appearances. Backlinks correlate 0.218. This single data point explains why businesses with strong Google rankings are disappearing from AI answers — and why some newer brands are showing up everywhere.

SEO isn't dead. But the authority signal changed. If you're still optimizing exclusively for backlinks, you are optimizing for the wrong game.

What Changed: From Ranking to Citation

Google's search algorithm spent two decades teaching the internet one lesson: earn links from authoritative sources and you will rank. Agencies built workflows around it. Entire software categories exist to track it. It worked.

The problem is that AI models — ChatGPT, Gemini, Perplexity, Claude — don't rank pages the way Google does. They synthesize answers from sources they've indexed and deemed credible. And credibility, to an LLM, looks different than it does to a traditional crawler.

An LLM determines whether your brand is a real, trustworthy entity based on signals like:

None of these are directly measured by a domain authority score.

The Numbers Behind the Shift

Zero-click searches climbed from 56% to 69% in the 12 months following Google's AI Overviews launch (May 2024 to May 2025). When an AI Overview is active on a query, 83% of users never click a result. Organic click-through rates dropped from 1.62% to 0.61% — a 61% decline — when AI Overviews are present.

The traffic that's dying is undifferentiated informational traffic. Users who would have clicked your blog post to get an answer they could get from an AI Overview.

What's growing is more valuable. LLM-referred traffic grew 527% year-over-year, from roughly 17,000 sessions to 107,000 sessions (January–May 2024 vs. January–May 2025). AI platforms sent 1.13 billion referral visits in June 2025 alone — a 357% year-over-year increase. And visitors arriving from AI citations convert at 23 times the rate of traditional organic visitors.

The businesses experiencing the collapse are the ones building content for Google's algorithm and not showing up in AI answers. The businesses experiencing the upside are the ones agents can find, read, and cite confidently.

How AI Agents Actually Read Your Site

Google's SAGE research (Steerable Agentic Data Generation for Deep Search) documented four shortcuts AI agents use to avoid deep research. The most common: Information Co-Location (35% of cases). Agents prefer pages that consolidate related information in one place over sites that scatter content across many URLs requiring cross-reference.

The second most documented shortcut: Overly Specific Questions (31% of cases). When a query is narrow enough, agents pull from a single authoritative source rather than synthesizing across multiple. Being that source requires being both findable and comprehensive on the specific topic.

Critically, SAGE found agents pull from the top 3 ranked pages per query. Traditional SEO ranking still matters as a filtering step. But once inside that set, citation preference shifts to machine-readability, information density, and entity clarity — not link count.

ChatGPT's crawler now operates at crawling levels equivalent to Google Desktop. Grok grew over 1,000% in a single month (July to August 2025). ChatGPT overall agent activity doubled in that same period. These agents are actively indexing your site. The question is whether what they find is useful enough to cite.

The Amazon vs. Perplexity Lawsuit Is Not a Side Story

On March 10, 2026, Amazon won a court injunction against Perplexity's Comet AI browser for bypassing robots.txt and scraping product data without authorization. Cloudflare separately documented Perplexity using stealth crawling to disguise its identity from web servers.

This matters to every business with a website.

The AI agent ecosystem is fracturing. Some agents will follow access rules. Others won't until courts force them to. Your robots.txt is no longer just a directive for Google — it's a policy document for an expanding universe of AI crawlers, each with different behavior and different legal standing.

Brands that haven't defined their agent access policy are making a decision by default. They're either blocking crawlers that would benefit them, or welcoming ones with unclear terms.

The SEO war of 2026 isn't only about rankings. Part of it is about which AI agents can read your site, and under what terms.

What Agentic Engine Optimization (AEO) Actually Requires

GEO (Generative Engine Optimization) is already the established term for optimizing content to appear in AI-generated answers. AEO goes further: it's optimization for AI agents that don't just answer questions but take actions — browsing, comparing, and transacting on behalf of users.

OpenAI's Operator already completes purchases on eBay, Instacart, and Etsy without the user visiting a product page. 51% of Gen Z starts product research in ChatGPT or Gemini rather than Google.

An agent buying on someone's behalf doesn't find you through keyword ranking. It finds you because your product data is structured, machine-readable, and consistent with the entity profile the agent has built for your brand.

Optimizing for this requires five things most SEO programs aren't doing:

1. Entity markup, not just page markup. Schema.org basics are table stakes. AEO requires entity-level structured data that tells AI models what your business is, who it serves, what makes it distinctive, and where it operates — consistently, across your site.

2. Information co-location on key pages. Your pricing, methodology, differentiated claims, and proof points should appear together on the pages you want agents to cite — not distributed across your site architecture where agents won't piece them together.

3. Branded mention strategy. Not guest posting for links. Strategic placement of your brand name as a cited entity in trade publications, industry roundups, expert comparisons, and review platforms that AI models index as authoritative. This is the 0.664 correlation signal. It needs a dedicated workflow.

4. Volatility monitoring. AI citation rates shift 15–45% week-to-week. Monthly SEO reporting is too slow to catch this. Brands need AI-specific monitoring at a cadence that reflects how frequently LLMs update their knowledge and ranking signals.

5. Agent access policy. Review your robots.txt for named AI crawlers. Decide which agents you want indexing your site. Update your policy proactively — the Perplexity litigation shows that "no one told me" is not a defense once agent behavior is established.

The Window Is Still Open

AI recommendations currently shift dramatically week-to-week because the space is young and unsettled. That volatility cuts both ways: it means early movers can establish citation presence before competitors catch on, because the systems haven't yet locked in default answers.

The agentic AI market was valued at $5.4 billion in 2024 and is projected to reach $50.31 billion by 2030. The businesses that establish entity clarity and branded mention footprints now will be the default answers AI agents surface in 2027 and 2028, when the market is large enough to matter enormously.

The ranking game rewarded the most authoritative site.

The citation game rewards the most legible one.

Those are different things. Most brands haven't made the translation.

Frequently Asked Questions

What is Agentic Engine Optimization (AEO)?

Agentic Engine Optimization (AEO) is the practice of structuring your web presence so that autonomous AI agents — systems that browse, compare, and transact on behalf of users — can find, read, and act on your business information. Unlike GEO (Generative Engine Optimization), which focuses on appearing in AI-generated text answers, AEO specifically addresses optimization for agents that take actions like purchasing, booking, or comparing products without human navigation.

How is AI visibility different from traditional SEO ranking?

Traditional SEO optimizes for Google's ranking algorithm, which weights backlinks and page authority heavily. AI visibility optimization targets how language models and AI agents determine which entities to cite or interact with. Research shows branded web mentions correlate 0.664 with AI Overview appearances versus 0.218 for backlinks — a fundamentally different authority signal that most SEO programs aren't measuring.

Why is organic CTR dropping so sharply?

Zero-click searches have climbed from 56% to 69% since Google launched AI Overviews in May 2024. When an AI Overview appears on a query, 83% of users get their answer without clicking any result. Organic click-through rate drops from 1.62% to 0.61% when an AI Overview is active — a 61% reduction. Traffic that previously drove clicks is now absorbed by AI-generated summaries.

What is the Google SAGE research and why does it matter for SEO?

Google's SAGE (Steerable Agentic Data Generation for Deep Search) research documented how AI agents navigate information rather than reading everything available. Agents pull from the top 3 ranked pages per query and use documented shortcuts — the most common being Information Co-Location (35% of cases), where they prefer pages that consolidate related information rather than requiring cross-site synthesis. This research shows why information architecture matters differently for AI agents than it does for traditional search crawlers.

How should brands respond to the Amazon vs. Perplexity lawsuit?

The Amazon court injunction against Perplexity's Comet AI browser (March 10, 2026) establishes that AI crawlers are subject to the same robots.txt rules as traditional crawlers. Brands should audit their robots.txt to explicitly address named AI crawlers, decide which agents they want indexing their content, and update their access policies proactively. The litigation signals that agent behavior is moving into legal territory — passive policies are no longer sufficient.

How do I start improving AI citation rates for my business?

Three high-priority steps: (1) Implement entity-level structured data that clearly defines your business as a named entity with consistent attributes. (2) Audit your branded mention footprint across third-party sources — trade publications, review platforms, industry comparisons — and build a strategy to increase citation frequency in sources AI models treat as authoritative. (3) Consolidate your key claims — pricing, methodology, proof points — on primary pages rather than distributing them across your site architecture.

Sources: Similarweb/SE Roundtable · ALM Corp · BrightEdge Research (2025) · Search Engine Land · Elementor · Position Digital · Google SAGE via Search Engine Journal · CNBC (March 10, 2026) · Fast Company · Logicbroker · Tryprofound · Search Engine Land Guide to Agentic AI in SEO