AI-driven search is reshaping how audiences discover information online, with major implications for how publishers and brands measure visibility and reach. New data shows AI traffic expanded rapidly in 2025 even as it remained a small slice of overall visits, and editors and marketers are reassessing how to appear in responses generated by ChatGPT, Perplexity, Google’s Gemini, and other tools. For organizations operating in fast-moving sectors such as cryptocurrency and blockchain, where timely market context and technical explainers are central to audience needs, the mechanics of earning AI citations now matter as much as traditional search engine optimization.

Semrush’s analysis of more than 50,000 websites found AI traffic grew 66% in 2025 compared with 2024, but still accounted for under 0.15% of total visits. That mix points to a structural shift: AI platforms increasingly act as both the starting point and endpoint for user queries, pulling relevant passages into a synthesized answer and often satisfying intent without a click-through. The result is a noticeable decoupling between search rankings and measurable inbound traffic. Sites may hold position in conventional results yet quietly lose visits as users receive concise, answer-first summaries inside AI products.

This behavior does not render SEO obsolete. Longstanding signals—backlinks, domain credibility, site performance, and high-quality writing—continue to matter. What has changed is how generative systems interpret those signals. Rather than rewarding comprehensive page-level authority alone, AI tools prioritize precise, self-contained passages that directly solve a user’s question. Good material from a lower-authority site can surface if it neatly addresses the query, while broad but meandering pages can be overlooked.

That reordering has two immediate consequences. First, even strong sites built on traditional SEO may not be prioritized the same way by AI systems. Second, visibility can exist without a referral: a model may cite a brand, author, or post in-line, exposing the name to readers without producing a session in analytics. For teams publishing research notes, explainers, and technical guides—including those serving crypto audiences—brand recognition from repeated AI citations can matter, even when direct traffic lags.

AI Integration

Generative platforms draw from a growing ecosystem of apps, models, and tools, including Proton AG’s Lumo, Meta AI, Mistral Vibe (formerly Le Chat), xAI’s Grok, Microsoft’s Copilot, Google’s Gemini, Anthropic’s Claude, Perplexity, Deepseek, OpenAI’s Chat GPT, Google’s Notebook LLM, and Suno. For web publishers, the practical interface with that ecosystem begins at the crawler level. Many content management systems ship with restrictive defaults that can inadvertently block AI bots. Ensuring robots.txt does not exclude GPTBot, OAI-SearchBot (OpenAI), PerplexityBot, ClaudeBot (Anthropic), or Google-Extended is a foundational housekeeping step; blocking them can suppress appearance in AI answers, regardless of a site’s conventional ranking strength.

Once crawl access is set, the structure of on-page writing becomes pivotal. AI systems tend to extract passages rather than ingest and quote entire pages. Leading each section with a clear, declarative sentence that resolves the core question—followed by nuanced context—maps well to how models assemble citations. Explicit numbers and source attributions also help: vague claims are more likely to be deprioritized during a model’s verification process.

Technology Use Case

Measuring whether AI engines are citing your work can be done with minimal cost. A straightforward starting point is qualitative testing: in ChatGPT, Perplexity, or Google AI Mode, run 10 to 15 queries where your content should be relevant and note which sources appear and how often. For a basic quantitative view, free utilities such as HubSpot’s AEO Grader offer quick evaluations of brand presence in AI search. Site owners using Google Analytics can also filter referral sources for chatgpt.com, perplexity.ai, and gemini.google.com to track patterns over time, watching for directional improvement or decline.

Beyond free checks, paid AEO tools consolidate tracking and benchmarking. Platforms such as Semrush One and Otterly.AI provide structured monitoring, with entry plans available at lower monthly costs. These services build on familiar SEO dashboards while focusing on where and how content is cited by AI products, which can be useful as teams expand their assessment beyond click-based metrics.

Market Impact

Research examined at the 2024 KDD Conference—conducted by Princeton, Georgia Tech, and the Allen Institute for AI—tested nine tactics across 10,000 queries and reported visibility gains of up to 40%. The findings align with the practical changes outlined above and translate into several actionable steps. First, confirm that AI crawlers are not blocked by default. Second, draft sections so that the opening line cleanly answers the question models are likely to ask. Third, embed numbers and citations to boost credibility signals during model verification. Fourth, consider a simple llms.txt file placed at the site root. This is a low-cost experiment proposed by Answer.AI’s Jeremy Howard; while it has not been validated by major AI vendors or AEO platforms, it is easy to test.

Fifth, broaden brand presence beyond a single domain. LLMs tend to assemble answers from a mix of formats and sources, including YouTube and Reddit. Publishing summaries, walkthroughs, and clarifications across those channels increases the chances that an AI system will encounter and reuse a publisher’s material when constructing an answer. That distribution principle is sector-agnostic and is particularly relevant wherever complex topics require synthesis for non-expert readers.

Industry Response

For teams focused on traditional SEO, much of the required toolkit already exists. Suites such as Semrush or Ahrefs provide the baseline for technical fixes, content audits, and competitor reviews. Dedicated AEO analytics can add depth, but they are not a prerequisite for getting started. The overarching directive is to keep creating useful, accurate explanations and to present them in a form that aligns with passage-level extraction. In practice, that means leading with the answer, supporting it with sourced details, and formatting sections so models can lift the essentials without ambiguity.

As AI products become the terminal surface for more queries, the visibility metric shifts from click counts to citations. That shift can be unsettling because it obscures traditional traffic attribution. Yet it also gives smaller, authoritative voices a path to surface if they write with precision and clarity. For subject areas where readers want fast, reliable synthesis—whether they are scanning for definitions, regulatory context, or how-to steps—the brands that are cited consistently inside AI responses will remain familiar, even when users do not open a new tab.

Publishers should also recognize the operational backdrop. Alongside product updates to AI search experiences—including interface changes and agent features—industry stakeholders continue to debate data access and usage. (Disclosure: Ziff Davis, ZDNET’s parent company, filed an April 2025 lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

The core message holds across topical domains. Traditional ranking factors still contribute to discoverability, but generative systems favor content that is immediately useful and easily quotable at the passage level. By allowing responsible crawler access, structuring sections for direct answers, grounding claims with numbers and sources, and maintaining a multi-platform presence, publishers can improve the odds of being cited in AI outputs. That approach preserves the value of established SEO practice while acknowledging a new reality: increasingly, the reader’s journey may begin and end inside the AI interface, and visibility will be earned sentence by sentence.