GEO, AEO, and the Chatbot That Talks Buyers Out of Buying

Generative AI was supposed to help people buy things faster. Turns out, it’s doing the opposite — and that changes the entire calculation for search strategy in Southeast Asia.

Semrush’s survey of 2,338 US consumers found that 57.5% of AI chatbot users were talked out of a purchase by the chatbot itself. Meanwhile, 65% of AI users have already replaced some product-related Google searches with chatbot conversations. If your GEO and AEO strategy isn’t built around this friction, you’re optimising for a funnel that increasingly doesn’t exist in its original form.

When the Chatbot Becomes the Objection Handler (Against You)

The Semrush data deserves a slower read. AI chatbots aren’t just replacing the search bar — they’re inserting a new, opinionated layer between brand and buyer. When a user asks a chatbot whether they should buy a particular product and the bot surfaces comparisons, caveats, or alternatives, the brand that isn’t cited authoritatively in that response has effectively lost the moment.

Four in ten AI users also said they dislike chatbot ads, which means paid interception isn’t going to paper over the gap. The play here is entity authority — ensuring your brand is cited as a credible, substantive source in AI training corpora and live retrieval contexts. For Southeast Asian brands with Shopee or Lazada storefronts, this is especially pointed: product discovery increasingly starts in a chat interface before it ever reaches a marketplace search bar.

Search Console AI Reports: The Diagnostic Tool You’re Not Using Yet

Google’s rollout of Search Console AI Overview reports to all markets globally — confirmed in early September 2026 — finally gives SEO teams in Southeast Asia a direct window into how their content surfaces inside AI-generated answers. Search Engine Journal reported the expansion alongside commentary from John Mueller clarifying two things that matter operationally.

First, Mueller confirmed that markdown formatting is genuinely useful for AI crawlers — not just a stylistic choice. Structured headers, concise definitions, and bulleted logic help LLMs parse and cite content accurately. Second, on cache-busting sitemaps: Mueller noted this can help accelerate re-crawling after significant content updates, which is relevant if you’re iterating fast on GEO-optimised content.

The practical implication for regional teams: if your CMS is still outputting walls of unstructured HTML copy, you’re making it harder for AI systems to extract and surface your expertise. A content audit filtered through the lens of AI parsability — not just keyword density — is now a legitimate Q4 priority.

---

The Harvard Provocation No One Wants to Say Out Loud

A Harvard study, covered by Search Engine Journal’s Greg Jarboe, scored 940 occupations on how morally objectionable the public finds automating them. Search marketing landed at 2.31 out of 7 — near the bottom of public concern. The profession, in the public’s estimation, is among the least ethically fraught to automate.

This isn’t a career threat article, and the research isn’t predictive of what will actually happen. But it is a signal worth sitting with strategically. If AI systems are already reshaping how brands get discovered and evaluated — and the broader culture is broadly unbothered by AI handling this work — then the competitive advantage shifts decisively toward marketers who understand how AI systems make decisions, not just how to feed them keywords.

In the context of Southeast Asia, where multilingual content, fragmented platforms, and mobile-first discovery patterns already create complexity, the human edge isn’t in execution volume. It’s in the structural thinking: which entities does my brand need to own, in which languages, across which platforms, so that when a Grab user in Jakarta or a LINE user in Bangkok asks a chatbot for a recommendation, the answer includes us?

What GEO Actually Requires Now

Generative Engine Optimisation, at its most functional, is about making your brand easy to cite correctly. That means three things working in parallel.

Entity clarity: Your brand, products, and subject-matter expertise need to be unambiguously defined across structured data, Wikipedia-adjacent sources, and high-authority publications. If an LLM can’t triangulate what your brand actually is, it won’t cite you confidently.

Content that answers, not just attracts: The chatbot-as-objection-handler finding from Semrush suggests that AI systems are comfortable contradicting brand claims when better information exists elsewhere. Publishing genuinely useful, specific, verifiable content — the kind that would survive a fact-check — is now a defensive GEO move, not just good content practice.

Platform-specific signals: For Southeast Asia, this means considering how content is structured for TikTok search, Shopee’s onsite search algorithm, and regional AI assistants that may have different training biases than Western LLMs. One-size-fits-all content architecture won’t hold.

---

Key Takeaways

• Structure content for AI parsability now — markdown formatting and clear semantic hierarchy directly influence how LLMs cite and surface your brand in generative results.
• AI chatbots are actively reducing purchase intent for brands that aren’t cited authoritatively; entity authority is your first line of defence, not ad spend.
• Southeast Asian teams should use Google’s newly global Search Console AI Overview reports as a diagnostic baseline before building any GEO content roadmap.

---

The real question isn’t whether AI is changing search — it clearly is, and faster than most editorial calendars have adapted. The more interesting provocation is this: if chatbots are now talking nearly 6 in 10 users out of buying, and your brand has no strategy for becoming the source those chatbots cite approvingly, what exactly is your search investment doing right now?

---

At grzzly, we work with brands across Southeast Asia to build search strategies that account for generative AI — not as a future consideration, but as the present reality reshaping how buyers discover, evaluate, and decide. If your GEO and AEO architecture hasn’t been pressure-tested against the way LLMs actually surface information, that’s a conversation worth having. Let’s talk