Schema Markup and AI Citations: What the Data Actually Shows

The slide deck version of this story writes itself: AI-cited pages are nearly three times more likely to use JSON-LD than non-cited pages. Screenshot that, add a gradient, post to LinkedIn. But Ahrefs’ Louise Linehan ran the actual experiment — 1,885 pages, before and after schema implementation — and the citation rate barely moved. That gap between correlation and causation is where a lot of search strategy goes wrong in 2026.

What the Schema Data Is Actually Telling You

Ahrefs’ study found that while AI-cited pages do skew heavily toward structured data, adding schema to a previously unstructured page produced negligible improvement in AI citation rates. The implication is uncomfortable but important: the pages that get cited by AI systems already had something schema alone can’t manufacture — demonstrable topical authority, clear entity relationships, and content structured around answering a specific question well.

Schema is a signal amplifier, not a signal creator. If a large language model is parsing your page and finding thin, ambiguous content wrapped in clean JSON-LD, the markup is doing nothing. Think of it less like a ranking factor and more like proper grammar — necessary for being taken seriously, but not sufficient to make you worth quoting.

For teams in Southeast Asia operating across multilingual content environments — Thai, Bahasa, Vietnamese, Tagalog — this distinction matters more than most. Schema implementation on content that hasn’t first resolved its entity clarity (who is this brand, what does it definitively do, in which market?) is putting the cart before the horse.

The Keyword Is Not Dead — It’s Just Working Differently

Semrush’s foundational explainer on keywords published this week is easy to dismiss as basics content, but there’s a strategic thread worth pulling. A keyword in 2026 is no longer just a traffic signal — it’s a classification system that tells AI models how to categorise your content within a knowledge graph.

The distinction between head terms, long-tail queries, and conversational intent phrases has always existed, but answer engines treat these categories with fundamentally different retrieval logic. A page optimised for “digital marketing agency Bangkok” is competing in one kind of ecosystem. A page that answers “how do brands in Thailand structure their agency relationships for performance marketing” is feeding a different machine — the one that generates cited responses in ChatGPT, Perplexity, and Google’s AI Overviews.

The practical shift: keyword research for GEO (Generative Engine Optimisation) should prioritise question-shaped, entity-rich phrases that map to discrete answers — not just search volume. A 200-monthly-search query that frames a specific problem your content resolves cleanly is often more valuable for AI citation than a 2,000-volume head term where your page is one of forty competing perspectives.

---

Affiliate and Partnership Models Are Being Repriced by AI Discovery

Search Engine Journal’s Adam Riemer raises a pointed observation about affiliate marketing’s structural vulnerability: when AI systems synthesise product recommendations without surfacing the affiliate link that would have completed a traditional conversion path, the revenue model breaks down. This isn’t a hypothetical — it’s already affecting commission structures across verticals.

For SEOs who have built brand or publisher strategies around affiliate traffic, the adaptation Riemer recommends isn’t to abandon the model but to reposition what value you’re actually selling. If AI is going to answer the “which product is best” question, the brand that gets cited is the one with the clearest, most structured, most authoritative answer — not necessarily the one with the highest affiliate commission rate.

In Southeast Asian e-commerce, where Shopee and Lazada’s internal search algorithms already mediate a significant portion of product discovery, this dynamic is accelerating. Brands that invest in structured content for AI retrieval while simultaneously optimising platform-native listings are hedging intelligently. Those treating SEO and platform content as entirely separate workstreams are leaving citation surface area on the table.

Building for Citation, Not Just Ranking

Pulling these threads together: the search visibility playbook in mid-2026 has a clear hierarchy. Content authority comes first — does your page actually contain a complete, defensible answer to a specific question? Entity clarity comes second — does every machine parsing your content know unambiguously who you are, what you do, and in which context? Structured data comes third — once the first two conditions are met, schema helps AI systems extract and attribute information correctly.

Teams that skip to step three are the ones adding JSON-LD to pages and then wondering why their AI citation rate didn’t move. Ahrefs’ data confirms what should have been intuitive: you cannot mark up your way to authority.

The implementation path for most marketing teams is to audit existing high-traffic content for entity ambiguity before touching a single line of schema. Can an AI model read your “About” page and definitively categorise your business, its geography, its specialisation, and its relationship to adjacent entities? If not, that’s the first fix.

Key Takeaways

• Schema markup correlates with AI citation but adding it to weak content produces no measurable citation lift — fix content authority first, then implement structured data.
• Keyword strategy for GEO should prioritise question-shaped, entity-rich phrases over high-volume head terms, particularly for AI Overview and Perplexity retrieval.
• Affiliate and partnership content models need to evolve toward citation-worthy authority, not just conversion-path optimisation, as AI increasingly intermediates product discovery.

The deeper question this data provokes: if AI systems are already sophisticated enough to distinguish between pages that use schema correctly and pages that use it well, how long before they can distinguish between brands that have genuine authority and brands that have simply learned to perform it? That gap — between signal and substance — is where the next wave of search strategy will be won or lost.

---

At grzzly, we work with marketing teams across Southeast Asia who are navigating exactly this shift — from traditional SERP optimisation to building content architectures that earn AI citation across Google, Perplexity, and emerging regional discovery surfaces. If your content strategy was built for a search landscape that no longer exists, that’s a conversation worth having. Let’s talk