Most businesses treat technical SEO as a housekeeping task — something that keeps Google happy but has no bearing on whether ChatGPT, Perplexity, or Gemini will ever mention their brand. New data suggests that assumption is wrong.
Key Takeaways
- A Semrush study of 5 million AI-cited URLs found that pages cited by AI platforms implement schema markup at significantly higher rates than the web average, with Organization schema present on 25-34% of cited pages.
- URL slugs between 17 and 40 characters receive the highest AI citation volume, with the peak range of 21-25 characters accounting for roughly 87,000 citations in the dataset.
- Page speed indirectly affects AI citation rates through user engagement metrics — pages cited in top positions by AI platforms had longer session durations and more pages per visit.
- In AI search, there is no ranked list of results — you are either cited or completely absent, making technical deficiencies far more costly than in traditional search.
- The recommended priority order for AI visibility is: schema markup, page speed, URL structure, mobile experience, crawlability, and security.
A Semrush study analysing 5 million URLs cited by AI platforms found clear patterns between technical SEO implementation and AI citation rates. Pages cited by ChatGPT Search and Google AI Mode consistently demonstrated stronger technical foundations than non-cited pages — from schema markup adoption to URL structure and user engagement signals tied to page performance.
The study draws correlations, not causal claims. But the pattern is unmistakable: the technical foundations that have driven traditional search performance for years are now showing up as distinguishing factors in AI search visibility. Here is what the data reveals, and what it means for your website.
Schema Markup: The Factor AI Models Rely On Most
Schema markup is no longer optional. It has become the primary way AI systems identify what your page is about, who published it, and whether the information is trustworthy.
The Semrush study found that pages cited by AI platforms implement schema markup at significantly higher rates than the web average:
| Schema Type | ChatGPT Search | Google AI Mode |
|---|---|---|
| Organization | 25% | 34% |
| Article | 20% | 26% |
| BreadcrumbList | 15% | 20% |
| FAQ | 3% | 5.5% |
Google AI Mode showed higher schema implementation rates across every type measured. This makes sense — Google has been encouraging structured data adoption for over a decade, and its AI system naturally gravitates toward pages that speak its preferred language.
The practical risk of poor schema implementation is what some researchers call the unattributed citation problem. When an AI model cannot confidently link information back to a specific brand — because Organization schema is missing or malformed — it may attribute the insight to a competitor or present it without attribution entirely. You did the work; someone else gets the credit.
If your pages lack Organization, Article, and BreadcrumbList schema at minimum, you are making it harder for every AI platform to understand and cite your content. Our complete SEO audit checklist walks through the full structured data verification process.
URL Structure: Shorter and Clearer Wins
URL structure is a technical factor most businesses set once and never revisit. The data suggests it deserves more attention.
The Semrush study found that URL slug length follows a clear citation curve. Pages with slugs between 17 and 40 characters received the highest citation volume, with the peak range of 21-25 characters accounting for roughly 87,000 citations in the dataset. Very short slugs (1-5 characters) and very long slugs (56+ characters) appeared significantly less frequently among cited pages.
This aligns with how AI retrieval systems work. A URL like /technical-seo-ai-search immediately signals topic relevance, while /p?id=4829 or /2026/03/26/how-do-technical-seo-factors-impact-ai-search-visibility-complete-guide gives the model either too little or too much to parse.
The practical takeaway: use descriptive, hyphenated slugs that capture the page's primary topic in 3-5 words. Remove dates, categories, and filler words from your URL path.
Page Speed and Core Web Vitals
AI platforms do not directly measure your page speed the way Google's Core Web Vitals do. But the Semrush data shows a strong indirect relationship: pages cited in top positions by AI platforms had higher engagement metrics — longer session durations, more pages per visit, and higher conversion rates.
Page speed is the invisible driver behind all of those metrics. According to Google's web performance research, pages that load their main content within 2.5 seconds (the LCP threshold) see measurably higher engagement across every metric. Slow pages lose visitors before they can engage, which means less signal for AI models that factor in user behaviour patterns.
Google AI Mode in particular showed notably higher engagement metrics among its cited pages compared to ChatGPT Search — suggesting that Google's AI system weighs engagement signals more heavily, likely because it has direct access to Chrome and Search Console data.
If your Largest Contentful Paint exceeds 2.5 seconds, your Cumulative Layout Shift is above 0.1, or your Interaction to Next Paint exceeds 200 milliseconds, those numbers are not just hurting your Google rankings — they are reducing the engagement signals that AI models associate with citable content.
Mobile Optimisation
Every major AI platform crawls from a mobile-first perspective. Google switched to mobile-first indexing years ago. ChatGPT's web search, Perplexity, and other AI crawlers follow similar patterns — evaluating the mobile version of your page as the primary representation.
A page that renders poorly on mobile — broken layouts, unresponsive elements, text that requires horizontal scrolling — creates two problems for AI visibility. First, the mobile render is what the AI crawler actually sees, so layout issues can obscure content. Second, poor mobile experience drives down the engagement metrics that correlate with AI citation rates.
Mobile optimisation in 2026 is not about having a responsive template. It is about ensuring that every interactive element, every image, and every content block delivers the same quality experience on mobile that it does on desktop. AI crawlers and the users whose engagement data feeds AI models are both evaluating the mobile version.
Site Architecture and Crawlability
AI systems need to discover your content before they can cite it. The rise of what BrightEdge CEO Jim Yu calls "agentic crawlers" — AI agents that traverse websites to gather information — means your site architecture matters more than ever.
Clean heading hierarchies (a single H1, logical H2-H3 structure) help AI models parse your content's structure. Clear internal linking helps them discover related pages. A well-maintained XML sitemap ensures nothing important is hidden behind JavaScript rendering or orphaned from the main navigation.
The fundamentals of crawlability have not changed — they are covered in depth in our analysis of SEO principles that still apply in the AI era. What has changed is the number of systems crawling your site. Beyond Googlebot, you now have ChatGPT's crawler, PerplexityBot, ClaudeBot, and dozens of agentic crawlers that need clean, parseable HTML to do their job.
If your robots.txt blocks AI crawlers, your content cannot be cited. If your pages rely on client-side JavaScript rendering without server-side fallbacks, AI crawlers may see empty shells instead of content.
HTTPS and Security Signals
HTTPS has been a Google ranking signal since 2014. In AI search, it functions more as a trust baseline — a necessary condition rather than a differentiator. The vast majority of pages cited by AI platforms in the Semrush study used HTTPS, which reflects the web's overall migration toward encrypted connections.
The more important security consideration for AI visibility is whether your site actively blocks AI crawlers through aggressive WAF configurations, CAPTCHA walls, or bot-detection systems that treat AI user agents as threats. Some businesses inadvertently block the very crawlers they want to be cited by.
Review your security configuration to ensure you are blocking malicious bots without excluding legitimate AI crawlers. ChatGPT, Perplexity, Google AI, and Claude all have documented crawler user agents — your server should recognise and allow them.
What This Means for Your AI Visibility Strategy
The Semrush data reinforces a principle that experienced SEOs already understand: technical foundations are not glamorous, but they are load-bearing. A beautifully written page with no schema, slow load times, broken mobile experience, and a cryptic URL is working against itself in both traditional and AI search.
The difference in 2026 is the stakes. In traditional search, poor technical SEO might cost you a few positions in a list of ten results. In AI search, there is no list — you are either cited or absent. Technical issues that cause a minor ranking drop in Google can cause complete invisibility in AI-generated answers.
The priority order based on current data:
- Schema markup — implement Organization, Article, and BreadcrumbList as a minimum
- Page speed — get LCP under 2.5 seconds and fix layout shift
- URL structure — aim for descriptive slugs of 17-40 characters
- Mobile experience — ensure full feature parity with desktop
- Crawlability — verify AI crawlers can access your content
- Security — HTTPS baseline plus AI-crawler-friendly bot management
If you want to see exactly where your site stands on these technical factors, SwingIntel's free AI Readiness Scan checks 15 signals across structured data, content clarity, and technical signals in under a minute. For the complete picture — including live AI citation testing across eight platforms and competitive benchmarking — the AI Readiness Audit covers all 24 checks plus the AI research that shows whether these technical foundations are translating into actual AI visibility.
Frequently Asked Questions
How does schema markup affect AI search visibility?
Schema markup is the primary way AI systems identify what a page is about, who published it, and whether the information is trustworthy. The Semrush study found that AI-cited pages implement Organization schema at 25-34% rates and Article schema at 20-26% rates — significantly higher than the web average. Without schema, AI platforms may attribute your content to a competitor or present it without attribution entirely.
Does page speed directly affect whether AI platforms cite your website?
AI platforms do not directly measure page speed, but there is a strong indirect relationship. Pages cited in top positions by AI platforms showed higher engagement metrics — longer sessions, more pages per visit, and higher conversion rates. Page speed drives these engagement signals, and AI models that factor in user behaviour patterns use them as trust indicators.
What URL slug length is optimal for AI citations?
Pages with URL slugs between 17 and 40 characters received the highest AI citation volume in the Semrush study, with the sweet spot at 21-25 characters. Descriptive, hyphenated slugs that capture the page's primary topic in 3-5 words perform best. Very short slugs and very long slugs appear significantly less frequently among AI-cited pages.
Can blocking AI crawlers in robots.txt hurt my visibility?
Yes. If your robots.txt blocks AI crawlers like GPTBot, ClaudeBot, PerplexityBot, or Google-Extended, your content cannot be cited by those platforms. Some businesses also inadvertently block AI crawlers through aggressive WAF configurations or CAPTCHA walls that treat AI user agents as threats.
The businesses that get technical SEO right for AI search are not learning new skills. They are applying existing skills to a new set of evaluators — evaluators that happen to be faster, more thorough, and less forgiving than the ones that came before. To see exactly where your site stands on these technical factors, run a free AI Readiness Scan or explore the full AI Readiness Audit for complete AI visibility research.
