Cited Newsletter Issue 25 (May 20th 2026)

Cited newsletter issue 25 featuring AI search visibility insights for marketing teams tracking GEO trends in May 2026

Welcome back to the Cited newsletter, your weekly read on what is actually moving in AI search visibility. Issue 25 lands at an interesting moment. The gap between brands that have started building AI visibility and those still watching from the sidelines is widening faster than most marketing teams anticipated six months ago. If you're in the first group, this issue has specifics to sharpen your programme. If you're still evaluating, the data below should make the decision easier.

What We're Seeing Across the Models Right Now

Across the brands we track on Lua, three patterns have become consistent enough to report with confidence.

ChatGPT Is Citing More Mid-Market Brands

For most of 2025, ChatGPT citations skewed heavily toward high-domain-authority publishers and household-name brands. That is shifting. We're seeing mid-market businesses with strong topical depth getting cited more frequently, particularly where their content covers a niche with more specificity than the generic content on larger sites.

The mechanism here isn't mysterious. When a user asks ChatGPT a specific operational question (think "how to manage supplier contracts for a scaling logistics business"), the model pulls from sources that have actually addressed that question with structured, authoritative content. A 10,000-word generic "supply chain guide" from a major publisher often loses to a 1,200-word focused piece from a specialist operator who knows the territory.

This is good news for marketing teams at established mid-market businesses. You likely have more genuine expertise than you've published. The gap isn't knowledge; it's structure and visibility.

Perplexity's Citation Patterns Are Diverging from ChatGPT

One finding from our multi-model tracking that deserves attention: Perplexity and ChatGPT are not citing the same sources at the rates many assumed. Across a sample of 200 tracked queries in April 2026, only 34% of cited domains appeared in both models for the same query. (Source: Lua internal tracking data, May 2026.)

This matters because some teams are optimising for one model and assuming the results transfer. They don't, reliably. The structural signals each model weights are different enough that you need platform-specific content architecture, not just a single piece of content pushed everywhere.

Google AI Overviews: The Structured Data Signal Strengthens

Google's AI Overviews continue to favour pages with clean structured data markup, clear entity definitions, and content that directly answers the query without burying the answer in preamble. This isn't new, but the signal has strengthened measurably over Q1 2026. Pages without schema markup are appearing in AI Overviews at roughly half the rate of equivalent pages with it, based on tracking across our brand set.

The Numbers That Matter This Issue

The divergence in model citation overlap is the one to watch. As each AI platform matures its own ranking logic, the "one strategy fits all models" approach becomes less defensible.

A Counterargument Worth Addressing

Some marketing teams push back on multi-model optimisation with a reasonable point: most of their customers only use one or two AI tools, so why spread effort across all of them? It's a fair challenge. The answer isn't that every model matters equally right now. It's that the content improvements required to perform well across models are not fundamentally different from each other. Better structure, clearer entity definition, stronger topical authority, faster page performance. These lift all platforms simultaneously. You're not running four separate programmes; you're running one programme that is tracked across four surfaces.

What to Focus On This Week

For Teams Early in Their AI Visibility Programme

• Run a structured audit of your top 20 pages against the query types your customers actually ask AI models. Are the answers on your site, or are you relying on the model to infer them?

• Add FAQ schema to your highest-traffic informational pages this week. It's a fast win with measurable impact on AI Overviews.

• Identify your two or three strongest topical areas and plan one long-form, deeply specific piece for each. Depth beats breadth at this stage.

For Teams with 60 or More Days of Active Optimisation

• Pull your citation data across ChatGPT and Perplexity separately. If your cross-model citation rate is below 40%, your content structure likely needs platform-specific adjustment.

• Review competitor citations for your core query set. Are the same three or four brands appearing consistently? Map what their cited pages have that yours don't.

• Check your entity coverage in your most important content. Are your brand, products, and key team members defined clearly enough for a model to extract and use them confidently?

Looking Ahead to Q3 2026

Two shifts we're monitoring that will likely define the rest of this year. First, Claude's web browsing capability has expanded significantly in 2026, and its citation behaviour is starting to resemble Perplexity more than ChatGPT in how it weights freshness versus authority. Teams that have treated Claude as a secondary platform may need to reassess that priority ranking by Q3.

Second, we expect Google to tighten AI Overviews eligibility criteria as it refines the product ahead of its next major search update. Brands that have earned consistent citation now will likely benefit from a degree of incumbency advantage. Those waiting for the "right moment" to start may find the window narrower than expected.

The brands building visibility now are not just getting early results. They are building the content infrastructure and domain authority signals that will compound over the next 12 to 18 months. That compounding effect is what justifies the investment, and it's why the teams we work with at Lua treat this as a programme, not a campaign.

Next issue lands 27 May. We'll be covering what the latest Claude update means for citation strategy and sharing findings from our first full-year cohort of Lua brands.

Sources: Lua internal platform tracking data (May 2026); SparkToro AI Search Behaviour Report Q1 2026; Search Engine Land AI Overviews Coverage Analysis (April 2026); BrightEdge AI Search Index Q1 2026.