Search has split in two. One path still delivers ten blue links. The other hands users a synthesized answer drawn from multiple sources, often without a click. Marketers who mastered the first now scramble to appear in the second.

Google’s AI Overviews and AI Mode, ChatGPT, Perplexity, Gemini and Claude pull passages, cite domains and shape perceptions at scale. Data from Hustle Marketers shows only 38% of AI Overview citations in March 2026 came from pages ranking in Google’s organic top 10, down from 76% in July 2025. The rest surface from deeper results or pages absent from the traditional list entirely.

But rank still matters. Kevin Indig’s analysis with AirOps found that position one on fan-out subqueries captured 58% of citations, four times the share of position ten. Classic visibility opens the door. What happens after depends on something else.

That something is citability. AI systems retrieve candidates through search infrastructure, then select which passages to quote or synthesize. Retrieval relies on proven signals: relevance, authority, crawlability. Selection favors clarity, specificity and corroboration from elsewhere on the web.

Google itself draws a firm line. Its guide to optimizing for generative AI features, updated July 10, 2026, states that terms like Answer Engine Optimization and Generative Engine Optimization describe common online ideas, yet many suggested tactics lack support in how Search operates. From Google’s perspective, optimizing for these features remains search optimization. The company explicitly cautions against over-focusing on structured data, noting no special schema is required.

A controlled study of 1,885 pages cited by Pepper Content found schema markup produced a 4.6% drop in AI Overview citations and no meaningful lift elsewhere. Similarly, llms.txt files saw near-zero requests across 137,210 domains in one analysis. Publishing volume alone doesn’t move the needle either. Among cited pages, 75% had been updated in the last year while only 42% were newly published.

The strongest signal? What others say about you. Pepper’s review concluded that 84% of AI citations trace to earned media. Paid placements contributed 0.3%. Third-party coverage builds entity strength that AI systems weigh heavily when deciding whom to trust.

HubSpot’s own examination of the topic echoes this shift. The company frames AI search optimization, which it also calls answer engine optimization, as the work of increasing the odds a brand appears in responses from ChatGPT, Perplexity, AI Overviews and Gemini. It builds on traditional search foundations rather than replacing them. AI-referred visitors convert at rates far higher than standard organic traffic. One Microsoft Clarity study across 1,200 publisher sites showed AI-driven users signed up at roughly 11 times the rate of typical search visitors.

Yet volume remains small. Semrush data indicated AI traffic accounted for just 0.14% of visits in 2025 despite 66% growth. The implication is clear. These mentions function less as traffic funnels and more as influence channels that shape consideration before buyers ever reach a website.

Technical access forms the first gate. AI crawlers such as GPTBot, Google-Extended and others must reach content. Major models largely ignore JavaScript-rendered text, demanding answers appear in raw HTML. Sites blocking these agents or hiding content behind logins simply vanish from consideration. Google Search Console’s Search generative AI control, rolled out in 2026, lets publishers opt out of AI features without harming traditional rankings. Most leave it enabled.

Content structure determines what gets extracted. Answer-first writing helps. Placing the direct response in the opening sentences of a section, followed by supporting detail, aligns with how models chunk and retrieve information. Question-shaped headings, concise lists, specific numbers and clear boundaries between concepts improve passage-level relevance. Titles matter disproportionately. Cited pages showed higher title similarity to the user’s prompt than non-cited ones.

Originality counts. Google stresses unique perspectives grounded in expertise over commodity summaries. Scaled AI-generated content without human insight or added value triggers spam filters. A Search Engine Journal report from October 7, 2026, predicted tighter scrutiny on self-promotional listicles and undisclosed paid mentions in AI responses. Early signs suggest such tactics already lose effectiveness.

Measurement has evolved too. Search Console’s generative AI performance reports, introduced in June 2026, track impressions in AI Overviews and AI Mode separately from classic results. Tools that simulate prompts across multiple models provide competitive benchmarks. Citation rate, share of voice across key buyer questions and accuracy of how a brand is described now sit alongside traditional ranking metrics.

Traffic patterns confirm the split. Pew Research Center data revealed users clicked traditional results 8% of the time when an AI summary appeared, versus 15% without one. Ahrefs measured a 58% decline in click-through rate for the top organic position on queries triggering Overviews. Informational content suffers most. Commercial and transactional queries remain less affected so far.

Yet visibility still drives downstream value. Brands cited in AI answers gain recognition that influences later direct visits or branded searches. Kevin Indig noted that more than 70% of AI-assisted journeys may appear as direct traffic in analytics, rendering last-click attribution incomplete. Incrementality testing and self-reported attribution become essential.

Practitioners who treat this as an extension of search rather than a replacement gain ground. They audit for crawler access first. They earn strong rankings on core topics. They craft passages that stand alone as answers. They secure mentions on authoritative external sites. They refresh existing assets more than they launch new ones. And they track AI-specific metrics continuously.

Google AI Mode crossed one billion monthly users by May 2026, with queries more than doubling each quarter. Perplexity, Claude and others expand the surfaces. The volume of zero-click searches climbed to 68% of U.S. Google queries in early 2026 per SparkToro analysis.

Success belongs to organizations that deliver clear, evidence-based information trusted across the web. No single hack delivers citations. No acronym supplants fundamentals. The systems reward depth, clarity and earned authority. Those qualities always did. They simply matter in new places now.

Teams ignoring the change watch traffic erode on informational queries while competitors appear in the answers buyers read first. Teams adapting discover that strong search foundations, sharpened for extractability and reinforced by genuine external validation, sustain relevance in both worlds.