Branded web mentions correlate 0.664 with AI Overview visibility. Backlinks correlate 0.218. — Ahrefs, 75,000-brand study

Ahrefs analyzed 75,000 brands and found that branded web mentions correlate 0.664 with AI Overview visibility. Backlinks correlate 0.218 (Ahrefs). 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:

  • How often your brand name appears as a named entity in third-party sources
  • Whether those sources are ones the model has indexed as authoritative
  • Whether your site provides structured, machine-readable information that agents can parse without guessing
  • Whether your core claims are verifiable and consistent across the web

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

The Numbers Behind the Shift

Zero-click behavior now dominates Google: less than one third of searches still send a click to the open web (SparkToro, 2026). And when AI Overviews are present, organic click-through rates drop from 1.76% to 0.61% — a 61% decline (Seer Interactive, September 2025).

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. Traffic to U.S. retail sites from generative AI tools grew 693% year-over-year during the 2025 holiday season (Adobe, January 2026). And the visitors who do arrive from AI convert better — 38% more likely to convert on Black Friday 2025 than other traffic (Adobe via Digital Commerce 360).

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 with Execution feedback) documented the shortcuts AI agents use to avoid deep research (Search Engine Journal, 2026). 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.

AI crawlers and indexers are actively crawling your site right now — every major lab runs them at scale. 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 preliminary injunction against Perplexity's Comet AI browser. The ruling's basis matters: Judge Maxine Chesney found Amazon showed strong evidence that Comet accessed Amazon's site at the user's direction but without Amazon's authorization — including password-protected areas — while masking its bot activity (CNBC, March 2026). Cloudflare separately documented Perplexity using stealth, undeclared crawlers to evade websites' no-crawl directives (Cloudflare, August 2025).

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 agent tooling already completes purchases on partner sites without the user ever visiting a product page, and younger shoppers increasingly start 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 dramatically week-to-week — we see it in our own citation probes. 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.

Morgan Stanley estimates agentic shoppers could account for 10-20% of U.S. e-commerce spending by 2030 — up to $385 billion (Morgan Stanley via Digital Commerce 360, December 2025). 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. Ahrefs' 75,000-brand study found branded web mentions correlate 0.664 with AI Overview visibility versus 0.218 for backlinks (Ahrefs) — a fundamentally different authority signal that most SEO programs aren't measuring.

Why is organic CTR dropping so sharply?

Less than one third of Google searches now send a click to the open web (SparkToro, 2026). When an AI Overview is active on a query, organic click-through rate drops from 1.76% to 0.61% — a 61% reduction (Seer Interactive, September 2025). 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 with Execution feedback) research documented how AI agents navigate information rather than reading everything available (via Search Engine Journal). 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 injunction against Perplexity's Comet browser (March 10, 2026) established that an agent acting at a user's direction still needs the website's authorization — the court focused on Comet accessing password-protected areas of Amazon while masking its bot activity, not on robots.txt (CNBC, March 2026). The lesson for brands: agent access is becoming a legal and policy question. Audit your robots.txt for named AI crawlers, decide which agents you want, and write your access policy deliberately — 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: Ahrefs — 75,000-brand AI Overview correlation study · SparkToro (2026) · Seer Interactive (September 2025) · Adobe (January 2026) · Adobe via Digital Commerce 360 (January 2026) · Google SAGE via Search Engine Journal (2026) · CNBC (March 10, 2026) · Cloudflare (August 2025) · Morgan Stanley via Digital Commerce 360 (December 2025)