AI visibility: earned media, tech SEO and agents

Organic rankings in Google no longer guarantee that AI will mention you. Research from Marketing Land shows that 88% of citations in AI Mode come from pages outside the organic top 10, and earned media accounts for 84% of all AI citations. If your visibility strategy has been built purely on ranking, you are working with half the picture.

Key takeaways

• Ranking does not guarantee an AI citation: 88% of AI Mode sources are not in the organic top 10.
• Earned media holds an 84% share of AI citations; paid content accounts for just 0.3%.
• Author schema nearly triples the chance of appearing in AI-generated answers.
• AI Overviews cause a 39.8% drop in organic clicks, and the lost visits are not low-quality.
• ChatGPT Thinking mode cites a different set of sources than Instant mode, with only 25.6% overlap.

Why AI skips you even when you rank

Traditional SEO and AI visibility are now two separate channels with very little overlap. According to Marketing Land data, only 16.5% of sources cited in AI Overviews also appear in the organic top 10. AI systems have learned to trust editorially independent sources because those sources have passed third-party fact-checking. Brand-owned content simply does not carry the same weight as journalist coverage.

The practical consequence: PR and media relations have become a direct tool for AI discoverability, not just a brand-building add-on. Investment in earned coverage in relevant publications now has a measurable impact on whether AI mentions you when your customers run comparison queries.

Named authors as verifiable entities

Sites with author schema implemented have nearly 3x higher odds of appearing in AI answers. The logic is straightforward: AI systems map information graphs, and a consistent byline across multiple publications creates a cross-verifiable authority. When your expert publishes under the same name on your own blog, in a trade publication, and in an industry journal, AI reads that as a strong trust signal. Author schema is a technically simple change with a disproportionate impact on AI visibility.

Two ChatGPT modes, two different source sets

A Semrush study produced a result that will surprise many marketers: ChatGPT in Thinking mode (high reasoning) cites a completely different set of sources than in Instant mode. Domain overlap is just 25.6%. Thinking mode generates an average of 4.5 citations per answer versus 2.6 in Instant mode, and triggers 4.6x more internal searches.

What does this mean in practice? Reddit and UGC sites lose half their share in Thinking mode, while government and academic sources grow fourfold. If you are targeting B2B buyers who use ChatGPT with deep reasoning for complex decisions, you need visibility in authoritative sources and your own original research, not just community threads.

There is also a compounding effect: a brand cited in the early Problem stage in Thinking mode tends to persist all the way to the final Selection stage. Name your products and methodologies explicitly in top-of-funnel content, not just at the bottom of the funnel.

Technical foundations that determine visibility on both channels

Technical SEO now serves two systems at once. A mistake in robots.txt or a broken redirect chain can cause invisibility in both classic search and AI answers. The Semrush technical checklist highlights fundamentals that get overlooked in practice: Core Web Vitals (LCP under 2.5s, INP under 200ms, CLS under 0.1), clean HTML without JavaScript dependencies for critical content, and correct domain canonicalization.

Two points deserve particular attention right now:

• LCP element: John Mueller flagged that the browser can lock onto the wrong element, especially in templated and carousel layouts. Before you optimize, confirm in DevTools which element the browser is actually measuring.
• AI agents and robots.txt: Google has warned against blanket-blocking agentic browsers. AI retrieval bots now represent real users, so blocking them means losing visibility. Robots.txt lets you control AI retrieval bots separately from training scrapers, so use that granularity.

The Siteline report also identified a specific problem for B2B sites: if an AI agent cannot read pricing from the initial HTML (because it is hidden behind JavaScript or a form), it will automatically turn to third parties and may cite outdated or incorrect data about your products. Pricing and key product information must be readable in raw HTML.

How to measure AI visibility alongside classic SEO

Organic traffic dropped 39.8% due to AI Overviews, according to a randomized experiment cited by Search Engine Journal. Critically, the lost visits were not low-quality: they had the same bounce rate, dwell time, and return-to-search rate as clicks that happened without an AI Overview. This is a real loss, not a filtering out of bad traffic.

For reporting, Search Engine Journal recommends a three-layer structure: financial outputs at the top (pipeline, revenue, CAC), performance drivers in the middle (traffic trends, conversion rates), and operational details at the bottom. The key metric is Organic CAC compared to paid CAC. Flat traffic with growing revenue is not a problem; it is evidence of better targeting.

AI Mode and AI Overviews now cover roughly 50% of searches, up 58% year-over-year. Without separate monitoring of AI citations, you are working with half your data. Charlie's AI agents can track your brand's mentions across AI answers and earned media sources, giving you the full picture rather than just organic positions.

What to do this week

Prioritization depends on your starting point, but three steps make sense for almost every team:

1. Check author schema on all key pages and ensure consistent bylines for your experts across publications.
2. Review robots.txt and confirm you are not blanket-blocking AI retrieval bots. Add rules for specific training scrapers if that is your intent.
3. Audit your pricing and product detail pages in raw HTML: open the page source and verify that critical information is readable without JavaScript.

If you want to see what your AI visibility actually looks like in numbers, take a look at Charlie or explore options for agencies.