AI Overviews Cut CTR 61%—But Your Traffic Isn't Dead Yet

A 61% drop in click-through rate sounds like a building on fire. The building is still standing — but the floor plan has definitely changed.

Seer Interactive’s analysis of brands cited in Google’s AI Overviews found that CTR fell 61% as impressions scaled dramatically faster than clicks. Search Engine Journal reports this clearly: more people are seeing brand mentions inside AI-generated answers, but fewer of them are clicking through to the source. Before you treat this as a traffic apocalypse, you need to understand what’s actually being measured — and what it means for how you invest in search.

Impressions Are Lying to You (In a Useful Way)

The CTR collapse isn’t a signal that AI Overviews are cannibalising your audience. It’s a signal that the metric itself has shifted meaning. When your brand gets cited in an AI Overview, you’re receiving an impression in a context where the user’s intent is already being partially satisfied. The click was never guaranteed — but the brand exposure is real and attributable.

Think of it less like a billboard that no one stopped for, and more like a trusted recommendation whispered into a conversation. The person may not call you today. They may already know what they needed. But they heard your name at the right moment.

For brands in Southeast Asia, where consumers frequently research on mobile before converting in-app or in-store — particularly through ecosystems like Shopee, Grab, or Lazada — this ambient brand reinforcement matters more than a direct click would suggest. The conversion path was never purely linear here.

Where Traffic Is Actually Moving

Semrush’s analysis of billions of visits across more than 50,000 sites reveals something more structurally interesting than a single CTR stat: the overall channel mix is shifting. Direct traffic and branded search are strengthening as referral traffic from organic SERPs softens across multiple industries. This isn’t Google breaking SEO — it’s the search experience compressing the discovery-to-decision journey.

What this means tactically: the brands winning in this environment are those that made themselves memorable at the awareness layer before the user ever typed a query. In markets like Thailand and Vietnam, where LINE and TikTok serve as de facto search interfaces for younger consumers, this cross-channel brand presence is already table stakes. Google’s AI layer is catching up to behaviour that Southeast Asian marketers have been navigating for years.

For local and hyperlocal search specifically, proximity signals and Google Business Profile completeness are becoming even more load-bearing. An AI Overview citing a business for “best laksa near Orchard Road” is doing work that a rank-3 organic result never could — it’s answering with confidence, not a list.

---

Why Some Sites Don’t Appear in AI Mode At All

SEO.com’s breakdown of AI Mode exclusions is a useful diagnostic. The seven reasons brands fail to appear in AI-generated responses cluster around three root causes: technical accessibility (crawlability, structured data gaps), authority signals (thin link profiles, low E-E-A-T indicators), and content architecture (answers buried in prose rather than surfaced clearly for extraction).

The last one is the most fixable and the most overlooked. AI systems — whether Google’s Overview engine or any generative retrieval layer — are optimised to extract discrete, confident answers. If your content buries a direct response inside four paragraphs of preamble, you’re effectively hiding the answer from the system doing the citing.

For local businesses across Southeast Asia managing multilingual content — English and Bahasa, Thai and Mandarin on the same domain — this extraction problem compounds. A structured FAQ in Bahasa that clearly states business hours, service area, and pricing is infinitely more citable than a beautifully written brand story that gestures at the same information. Clarity serves both users and AI retrieval.

Content Engineering Is the New Production Advantage

Ahrefs’ Ryan Law recently documented his content engineering workflow using Claude Code — a process that compresses multi-day production cycles into hours by systematically structuring research, brief creation, and draft generation. The strategic implication isn’t about speed for its own sake. It’s about being able to produce the volume and specificity of content required to be citable across a wider range of query types.

For teams managing search across five Southeast Asian markets with different languages and intent patterns, this kind of engineered content production isn’t a nice-to-have. It’s the only realistic path to maintaining presence across AI-cited, organic, and local search simultaneously. A single pillar piece isn’t enough when AI systems are pulling from dozens of distinct, specific sources to build a single answer.

The brands that will be cited consistently are those producing content that is structured like evidence — specific, attributed, extractable — not content that reads like it was written to impress a reader.

Key takeaways:

• Optimise for AI citation by structuring content with clear, extractable answers — especially for local queries where proximity intent is high.
• Treat AI Overview impressions as brand-building metrics, separate from direct traffic KPIs, and build measurement frameworks that capture downstream branded search lift.
• Invest in multilingual content engineering workflows that can produce structured, citable content at scale across your Southeast Asian markets.

The deeper question worth sitting with: if AI systems increasingly mediate what users discover and believe, does “ranking” even remain the right frame — or are we actually in the business of building citation authority, more like PR than SEO?

---

At grzzly, we work with regional brands navigating exactly this shift — building search strategies that account for AI citation mechanics, local pack visibility, and the messy reality of multilingual, mobile-first audiences across Southeast Asia. If your search investment is optimised for a 2023 SERP that no longer exists, we should talk. Let’s talk