AI Search Integration: Why 78% of Marketers Are Already Behind

Only 22% of marketers have fully integrated AI search into their SEO workflows. That’s not a laggard statistic — that’s a compounding authority gap opening up in real time.

A Semrush survey of 481 marketers confirms what most teams already sense but haven’t acted on: AI search is no longer a parallel track to SEO. It’s rewriting the rules of discoverability, and the brands treating it as a future-state concern are already playing catch-up.

The Operational Gap Is the Actual Problem

The Semrush study doesn’t just flag that adoption is low — it identifies where the breakdown happens. Most teams understand the shift intellectually. The failure is operational: workflows, content briefs, measurement frameworks, and internal sign-off processes were all designed for a world where ranking in the ten blue links was the primary objective.

Generative engines don’t return a ranked list. They synthesise an answer — and they source that answer from entities, brand signals, and structured semantic context they’ve learned to trust. That means your content strategy needs to be engineered for citation probability, not just keyword density. In Southeast Asia, where mobile-first consumption means AI-assisted answers often replace a browse session entirely, this shift is already more acute than in Western markets.

The 22% who’ve closed this gap aren’t working harder — they’re briefing differently, measuring differently, and governing content with entity authority in mind from the first draft.

What Google’s GEO Guidelines Actually Tell You

Moz’s Dr. Peter J. Meyers published a sharp breakdown of Google’s official GEO guidelines this week, and one finding cuts through the noise: structured data is not the shortcut everyone hoped it would be. Google’s guidelines make clear that schema markup assists interpretation but doesn’t substitute for genuine topical authority or content quality.

What the guidelines do reinforce is that SEO isn’t dead — it’s foundational. The signals that help Google’s traditional crawlers understand your content (clear entity relationships, consistent NAP data for local, authoritative backlink context) are the same signals feeding generative retrieval. The GEO layer sits on top of a functioning SEO substrate, not beside it.

For regional teams managing multilingual content across markets like Thailand, Vietnam, and the Philippines, this has an immediate implication: entity consistency across language variants matters. A brand that has clean entity signals in English but fragmented ones in Thai or Filipino is invisible to a generative engine answering a query in those languages.

---

LLMs.txt Is Not the Protocol You’re Waiting For

If your team has been watching the LLMs.txt conversation and wondering whether to implement it — Google’s John Mueller has a clear answer: don’t hold your breath. Search Engine Journal reports that Mueller has publicly confirmed LLMs.txt has no current implementation at Google, describing it as purely speculative. He’s indicated a preference for WebMCP, a Google-backed alternative that remains in early stages.

This matters strategically because several agencies and tooling vendors have been positioning LLMs.txt as a near-term dial you can turn to improve AI discoverability. That framing is premature at best. The more durable move is investing in the fundamentals that do feed generative retrieval today: clean semantic structure, strong entity associations, and the kind of cited-source reputation that comes from genuine editorial presence.

Brands in Southeast Asia chasing LLMs.txt implementations are spending energy on a protocol that may never ship in its current form, while the underlying authority architecture that actually drives AI citations gets another quarter behind.

Domain Authority Still Matters — But the Frame Has Shifted

Ahrefs published updated analysis this week on what constitutes a meaningful Domain Rating, and the finding has a useful application to GEO strategy. DR is inherently relative — your score reflects competitive positioning, not an absolute quality threshold. A DR of 50 means something very different in a hypercompetitive finance vertical than in a regional niche.

For generative engine optimisation, the parallel logic applies: entity authority is contextual. A mid-sized e-commerce brand in Indonesia with deep, consistent content across its product verticals can achieve stronger AI citation rates within its domain than a high-DR generalist site that never commits to topical depth. The generative retrieval model rewards coherent expertise signals, not raw link equity.

The practical implication: stop benchmarking against global DR averages and start auditing whether your entity footprint — across your own site, earned media, and platform presence on Shopee, Lazada, or regional news outlets — tells a coherent, citable story about what your brand knows and owns.

---

If the question keeping growth leads up at night is “will our brand be cited or ignored by AI search,” the honest answer is that the gap is already forming. The 22% who’ve integrated AI search into their workflows aren’t smarter — they started treating entity authority and semantic structure as infrastructure, not optimization. The question worth sitting with: what would it actually take to rebuild your content operations around citation probability rather than click probability?

At grzzly, this is the exact terrain we’re navigating with growth teams across Southeast Asia — auditing entity authority, restructuring content workflows for generative retrieval, and helping brands build the kind of semantic presence that AI engines learn to trust. If your team is sensing the gap but unsure where to start, Let’s talk.