Google AI Mode & GEO: What SEO Teams Miss in 2026

AI Mode is not coming. It’s already here — and the traffic data is starting to tell an uncomfortable story.

What Google AI Mode Is Actually Doing to Click Volumes

SEO.com’s analysis of Google AI Mode confirms what a lot of us suspected but couldn’t yet quantify: AI-generated answer surfaces are intercepting informational queries before users ever reach a blue link. The mechanism is straightforward — when Google’s LLM synthesises a confident answer from multiple sources, the incentive to click through evaporates. For brands whose organic traffic relied heavily on top-of-funnel, definition-style content, the exposure is real.

But here’s the nuance that most post-mortems miss: AI Mode doesn’t kill all traffic equally. Queries with high purchase intent, local specificity, or genuine ambiguity still route users to websites. The sessions that disappear are the ones that probably weren’t converting anyway — the “what is” searches, the broad comparisons. The strategic question isn’t whether AI Mode hurts traffic. It’s whether the traffic you’re losing was ever worth chasing.

For Southeast Asian markets, where mobile-first users on Shopee or Lazada are already conditioned to get answers inside platforms rather than bouncing to external sites, this shift is less a disruption than an acceleration of an existing pattern.

GEO: The Discipline That Replaces Classic Link-Building Logic

Generative Engine Optimisation is the practice of structuring your brand’s information so that LLMs — Google’s AI Mode, ChatGPT search, Perplexity — cite you accurately and confidently. It is not about stuffing pages with keywords. It is about entity authority: being the clearest, most corroborated source for a specific claim in your domain.

The practical implication is a shift in editorial strategy. Content that performs in GEO tends to be opinionated, specific, and well-attributed. A fintech brand in Bangkok that publishes a detailed breakdown of QR payment adoption rates across Thai provinces — with named data sources, a clear methodology note, and structured markup — is far more likely to surface in an AI-synthesised answer than a competitor’s generic “digital payments in Thailand” overview.

Moz’s Whiteboard Friday on vibe-coding custom SEO tools is quietly relevant here. Gus Pelogia’s entity-tracking automation — built in Google Colab using ChatGPT — is exactly the kind of lightweight tooling that lets lean SEO teams monitor whether their brand entities are being correctly understood and cited by AI systems. If you’re not tracking entity mentions in LLM outputs, you’re flying blind on roughly half your discoverability picture.

---

Low-Competition Keywords Still Matter — But the Strategy Has Shifted

Semrush’s guide on finding low-competition keywords remains tactically sound, but the strategic context has changed. In a pre-AI-Mode world, low-competition keywords were a volume play — find the gap, rank quickly, harvest traffic. In 2026, they’re better understood as entity-building opportunities.

A low-competition keyword cluster around a specific topic tells you where the knowledge graph has thin coverage. That’s where a well-structured, authoritative piece can become the canonical source that an LLM reaches for when constructing an answer. The goal isn’t just the ranking — it’s becoming the entity that gets cited.

For multilingual markets across Southeast Asia, this opens an underexplored advantage. Thai, Bahasa Indonesia, and Vietnamese query spaces are significantly less saturated in LLM training data than English. Brands that produce structured, semantically rich content in local languages — properly marked up with hreflang, schema, and entity disambiguation — are staking claims in knowledge graph territory where competition is still thin. The window won’t stay open indefinitely.

The Spam Policy Shift You Probably Overlooked

Search Engine Journal flagged an update that reads like a footnote but matters for competitive intelligence workflows: Google has updated its spam reporting tool to reject submissions containing personal information. Reports that include named individuals or personal identifiers will simply not be processed.

The practical consequence for SEO teams doing competitive monitoring is that any spam-flagging strategy that relied on detailed attribution of manipulative behaviour to specific actors is now less viable as a formal channel. More broadly, it’s a signal that Google is tightening the data it ingests through community reporting — which shifts the burden of quality signalling back onto algorithmic signals and structured data. Another nudge toward building legitimacy through entity clarity rather than attempting to game or police rankings through external channels.

For brands operating across Southeast Asia’s diverse regulatory environments — where content moderation standards and data privacy laws vary sharply between Singapore, Indonesia, and Vietnam — this is also a reminder that compliance-aware content architecture is no longer optional. Clean entity structure and clearly attributed information are both an LLM ranking signal and a risk management posture.

Three things worth acting on this week:

• Audit your entity footprint: Search your brand name and key product claims in AI Mode, Perplexity, and ChatGPT. Note what gets cited, what gets wrong, and what’s missing. That gap is your GEO brief.
• Reframe your keyword research: Treat low-competition clusters as knowledge graph beachheads, not just traffic plays. Structure content to be the canonical answer, not just a ranking page.
• Build lightweight entity tracking: Even a basic Google Colab script monitoring brand mentions in LLM outputs will tell you more about your generative visibility than most enterprise SEO dashboards right now.

The brands that will own generative search visibility in Southeast Asia over the next 18 months are the ones building entity authority quietly, right now, while the majority of the market is still debating whether AI Mode is a threat. The question worth sitting with: if an LLM had to describe your brand’s core expertise in two sentences, what would it actually say — and would you be comfortable with that answer?

---

At grzzly, we work with growth teams across Southeast Asia who are navigating exactly this transition — from classic SEO playbooks toward entity authority, structured semantics, and generative discoverability. If you’re trying to figure out where your brand sits in the LLM knowledge graph and what to do about it, we’re thinking about this every day. Let’s talk