Is the URI Fetch Request the New 'Click' for an Agentic Web?
Clicks to a Website Are Not the Metrics for Content Performance Anymore
Nearly 60% of all Google searches now end without a single click (SparkToro/Similarweb, 2024). When AI Overviews appear, that number jumps to roughly 80% (Similarweb, 2025). In Google’s dedicated AI Mode, 93% of searches produce zero clicks (Semrush, 2025). And AI Overviews reduce click-through rates on the top-ranking result by 58% (Ahrefs, 2026). The zero-click world isn’t coming - it’s here, and it’s accelerating.
But here’s what most brands are missing: the click didn’t disappear. It changed form.
When a user asks ChatGPT to recommend a product, compare services, or research a purchase, the AI doesn’t generate an answer from thin air. It sends a URI fetch request to your website - a real HTTP request that hits your server, retrieves your content, and uses it to build the answer the user sees. That fetch request is the new click. It’s the moment your content was selected from the entire internet to answer a specific question. The user may never visit your website, but your content was chosen, retrieved, and presented to them.
In a zero-click world, the URI fetch request is the unit of engagement that matters. Brands need to start tracking it as a performance metric.
What Is a URI Fetch Request?
A URI fetch request is an HTTP request made by an AI platform to retrieve a specific page from your website in real time, during a live user conversation. It is not a crawl for training data. It is not an index build. It is a targeted retrieval: an AI system determined that your content is relevant to what a user just asked, and it fetched your page to use in generating the response.
This distinction matters because AI platforms use different bots for different purposes, and they leave different signatures in your server logs:
A training crawl is like a library acquiring books for its collection. A URI fetch request is like a librarian pulling a specific book off the shelf because someone just asked a question. The crawl builds the knowledge base. The fetch answers the question. Both hit your server, but they mean fundamentally different things for your brand.
Why the Fetch Request Is the New Click
In traditional search, the click was the signal that mattered. A user saw your result, decided it was relevant, and clicked through to your site. That click was the unit of engagement - the moment a human chose your content over everything else on the page. Click tracking isn’t going away - it’s still a valuable signal when users do visit your site. But it now has a sibling: the URI fetch request.
The URI fetch request is the AI equivalent of the click. Here’s why:
1. Selection - The AI evaluated your content against every other source on the internet and chose to fetch yours. This is not random. It’s a relevance decision made by the model in real time. Your page was selected.
2. Retrieval - The AI sent an HTTP request to your server and received your content. This is a measurable event in your server logs. It happened. You can count it.
3. Presentation - Your content was synthesized into the answer the user received. The user may never click through to your site, but they consumed your content - and it shaped their decision.
Clicks still matter when they happen. But as zero-click interactions grow, relying on clicks alone means missing the majority of moments where your content reaches consumers. The fetch request fills that gap - it’s the new sibling metric that tells you your content is being selected and served through AI, even when no click follows.
The AI Bots Making Fetch Requests to Your Site
Every major AI platform uses distinct user agents for fetch requests versus training crawls. These are the fetch bots triggered by real user queries - the ones that represent your content being selected to answer a question in real time.
| User Agent | Platform | Purpose |
|---|---|---|
| ChatGPT-User | OpenAI | On-demand fetch. Fires when a user or Custom GPT requests live web content during a conversation. |
| Claude-User | Anthropic | On-demand fetch. Retrieves content in real time when Claude needs current information to answer a user’s question. |
| Perplexity-User | Perplexity | On-demand fetch during live user queries. Content is used immediately and not stored for training. |
Each of these user agents represents a moment where a real person asked an AI a question, and the AI chose your page to answer it. That’s the fetch request - the new sibling to the click.
The Fetch-to-Crawl Ratio: Your AI Content Quality Score
Your server logs contain two categories of AI bot activity: crawl requests (training data collection) and fetch requests (real-time content retrieval). The ratio between them is the most diagnostic metric in your AI visibility toolkit.
A high crawl rate with a low fetch rate (ratio closer to 0) means your content made it into the training data but isn’t being pulled into live conversations. The AI knows your content exists, but it doesn’t find it useful enough to cite when users ask relevant questions. That’s a content quality signal, not a technical problem.
A balanced or high fetch-to-crawl ratio (ratio trending upward) means your content is both accessible and authoritative. The AI not only knows about your content - it actively retrieves it to answer questions. That’s the signal that your content is doing its job.
Think of it this way:
- Crawl rate = the AI has read your content
- Fetch rate = the AI is recommending your content
- Fetch-to-crawl ratio = how useful the AI thinks your content is
We introduced this metric in our AI Performance Metrics framework as KPI #2 (AI Fetch Rate). In the OpenAI-specific context, the ChatGPT Fetch Rate distinguishes between GPTBot training crawls and ChatGPT-User fetch requests. This article explains why that distinction is the most important measurement shift brands need to make in 2026.
What Makes Content Fetchable
Being crawled is a technical requirement. Being fetched is a content quality achievement. The AI decided your page was the best answer to a specific question - and that decision is based on signals you can influence.
Structured data and schema markup. AI systems use structured data to understand what your page is about, what entities it references, and how authoritative it is. Pages with comprehensive schema markup are easier for AI to parse and more likely to be fetched as authoritative sources.
Content depth and specificity. Thin content gets crawled but not fetched. AI systems retrieve content that provides substantive, specific answers. A 200-word product description loses to a competitor’s page with specifications, comparisons, use cases, and structured data. The depth of your content directly correlates with fetch frequency.
Freshness and accuracy. Real-time fetch requests are triggered by user questions about current topics. If your content is outdated, AI systems will fetch a competitor’s more current page instead. The content that gets fetched is the content that’s been maintained.
Technical accessibility. If your content is locked behind JavaScript rendering that AI bots can’t execute, it doesn’t matter how good it is. Server-side rendering, fast response times, and clean HTML structure are prerequisites for fetchability. A page that takes 5 seconds to render will be abandoned by the fetch request.
Multi-modal content coverage. AI platforms increasingly favor content-rich pages with multiple content types - text, images with alt text, video, structured specifications. Products with comprehensive multi-modal content are fetched more frequently than those with text-only descriptions.
The Scale of What’s Happening
The numbers paint a clear picture of how fast this shift is moving:
- 50 billion crawler requests per day from AI bots hit the web by late 2025 (Cloudflare)
- 305% growth in GPTBot traffic year-over-year (Cloudflare)
- 15x increase in AI “user action” (fetch) crawling in 2025 (Cloudflare Radar)
- 4.2% of all HTML page requests now come from AI-oriented bots (Cloudflare Radar)
- ~60% of all Google searches end without a click (SparkToro/Similarweb); ~80% when AI Overviews are present (Similarweb)
- 93% of AI Mode searches produce zero clicks (Semrush)
Every one of those fetch requests is a moment where an AI system chose someone’s content to answer a user’s question. If your site isn’t being fetched, your competitors’ sites are.
What This Means for Your AI Visibility Strategy
The URI fetch request reframes how brands should think about AI visibility. It’s not enough to ask “Can AI bots crawl my site?” The question that matters in 2026 is “Is AI fetching my content to answer user questions?”
This connects directly to the measurement framework we’ve built at WISLR:
- AI Bot Crawl Rate tells you whether AI can access your content (infrastructure)
- AI Fetch Rate tells you whether AI is citing your content (the metric this article is about)
- AI Referral Traffic tells you when users click through from AI (the traditional click, still valuable)
- Revenue from AI tells you what it’s all worth
The fetch rate sits at the center of this funnel. It’s the bridge between technical accessibility and business impact. Without fetch, crawl is just storage and referral traffic never materializes.
When to Work With an AI Visibility Consultant
Building fetch monitoring infrastructure requires server log analysis, user agent segmentation, and a measurement system that doesn’t exist in any off-the-shelf tool. Most brands have the raw data in their server logs but lack the expertise to extract, segment, and interpret it.
The WISLR team helps brands build their AI visibility measurement infrastructure - from server log pipelines to fetch-to-crawl ratio dashboards to full AI channel reports. If you want to know whether your site is being fetched and what to do about it, we can get you from zero to a working measurement system.
Want to see how fetch rate fits into the full AI metrics framework? Our AI Performance Metrics: Seven KPIs Every Brand Should Track covers the complete measurement system from crawl to revenue. For OpenAI-specific KPIs including ChatGPT fetch rate and Operator transaction tracking, see our OpenAI KPIs and Success Metrics breakdown.
Frequently Asked Questions
What is a URI fetch request from an AI bot?
A URI fetch request is an HTTP request made by an AI platform to retrieve a specific page from your website in real time, during a live user conversation. Unlike training crawls (which collect content on a schedule for model training data), fetch requests are triggered on demand when an AI system determines your content is relevant to what a user just asked. The AI sends a request to your server, retrieves your page content, and uses it to generate the response the user sees. Each major AI platform uses distinct user agents for fetch requests: ChatGPT-User (OpenAI), Claude-User (Anthropic), and Perplexity-User (Perplexity). These fetch requests are identifiable in your server logs and represent the moment your content was selected to answer a specific question.
How is a URI fetch request different from an AI crawl?
Training crawls and URI fetch requests serve fundamentally different purposes and use different user agents. Training crawls (GPTBot, ClaudeBot, PerplexityBot, Bytespider) collect content on a schedule to build or update an AI model’s training data - like a library acquiring books for its collection. URI fetch requests (ChatGPT-User, Claude-User, Perplexity-User) retrieve content in real time during a live user conversation - like a librarian pulling a specific book to answer a question someone just asked. Both appear in your server logs as HTTP requests, but crawl activity tells you whether your content is in the AI’s knowledge base, while fetch activity tells you whether your content is being actively cited in user conversations.
Why is the fetch request called “the new click”?
In traditional search, the click was the unit of engagement - the moment a user saw your result and chose to visit your site. Click tracking isn’t going away - it’s still valuable when users do visit your site. But in a zero-click world where nearly 60% of all searches end without a click, roughly 80% of AI Overview searches are zero-click, and 93% of AI Mode searches produce zero clicks, relying on clicks alone means missing most of the moments where your content reaches consumers. The URI fetch request is the new sibling metric to the click. When an AI platform fetches your content, it means the system evaluated your page against every other source on the internet and selected yours to answer a user’s question. The user consumed your content even though they may never have visited your site. The fetch request represents the same selection moment that a click does - it’s just happening between the AI and your server instead of between the user and your URL.
What is the fetch-to-crawl ratio and why does it matter?
The fetch-to-crawl ratio is calculated by dividing the number of real-time fetch requests your site receives from AI platforms by the number of training crawl requests over the same period. This ratio functions as an AI content quality score. A high crawl rate with a low fetch rate means your content is in the AI’s training data but isn’t being cited in live conversations - indicating a content quality problem, not a technical one. A balanced or improving fetch-to-crawl ratio means your content is both accessible and authoritative enough that AI systems actively retrieve it to answer questions. Tracking this ratio per platform (OpenAI, Anthropic, Perplexity) gives you granular insight into which AI ecosystems find your content most useful.
How do you track AI fetch requests on your website?
AI fetch requests can only be reliably tracked through server log analysis. No off-the-shelf analytics tool captures this data. Filter your server logs for fetch-specific user agents: ChatGPT-User (OpenAI), Claude-User (Anthropic), and Perplexity-User (Perplexity). Separate these from training crawl agents like GPTBot, ClaudeBot, and PerplexityBot. Track total fetch volume, trending direction, which specific pages are being fetched, and calculate your fetch-to-crawl ratio per platform. AI user-action crawling increased more than 15x in 2025, so if your fetch volume is flat while the industry is growing, your content is losing relevance to AI platforms.
What makes a page more likely to be fetched by AI bots?
Pages that get fetched share several characteristics: comprehensive structured data and schema markup that helps AI systems understand the content, substantive depth with specific answers rather than thin descriptions, current and accurate information that reflects real-time reality, fast server-side rendering that allows bots to access content without executing JavaScript, and multi-modal content including images with alt text, video, and structured specifications. Thin content with minimal structured data gets crawled for training but rarely fetched for real-time answers. The pages AI platforms choose to fetch are the ones that provide the best, most structured, most current answer to the question a user just asked.