Google's personalization push and ChatGPT's split reasoning modes are reshaping local search visibility. Here's what Southeast Asian brands need to do now.
Google’s head of Search, Liz Reid, recently floated a claim that personalization could be a lifeline for smaller publishers inside AI-powered search. Meanwhile, Semrush quietly published research showing that ChatGPT’s two reasoning modes — Thinking and Instant — share only 25% of their cited sources for identical prompts. For local and hyperlocal search teams, both signals point to the same uncomfortable truth: the search landscape is fracturing, and proximity alone won’t save you.
Google’s Personalization Bet — Promise or Deflection?
At a recent industry event, Google’s Liz Reid suggested that as AI Overviews become more personalised — surfacing content from sources users have previously engaged with — smaller, specialist publishers stand to benefit. The logic is intuitive: a user who regularly reads a Kuala Lumpur food blog might start seeing it cited in AI-generated answers about local dining.
The catch? Reid offered no supporting data. Zero. For local SEO practitioners who’ve watched neighbourhood businesses lose map-pack real estate to aggregators, this is a familiar pattern: a plausible mechanism with no empirical floor underneath it. Search Engine Journal’s Matt G. Southern flagged the absence of evidence directly.
The strategic implication isn’t to dismiss personalization — it’s to actively engineer it. That means building return-visit behaviour through email, LINE OA broadcasts, or Grab merchant loyalty signals that drive users back to your content repeatedly. You can’t wait for Google’s algorithm to discover your audience; you have to build it first and let the algorithm follow.
ChatGPT Is Not One System. Stop Treating It Like One.
The Semrush study is the most structurally important research I’ve seen this quarter for anyone running an AEO or GEO strategy. The finding: when the same prompt is run through ChatGPT’s Thinking mode versus Instant mode, only 25% of cited sources overlap. These modes don’t just answer differently — they draw from fundamentally different content pools and favour different brand recommendations.
Thinking mode, which applies extended reasoning chains, tends to favour longer-form, heavily cited, authoritative content. Instant mode gravitates toward concise, directly answerable content — the kind that mirrors a well-structured FAQ or a tightly written Google Business Profile description.
For a retail brand operating across, say, five cities in Thailand with both a corporate content hub and individual store pages, this bifurcation matters enormously. Your in-depth category guides may win citations in Thinking mode while your store-level pages — if structured correctly — capture Instant mode responses for “near me” and transactional prompts. These aren’t competing strategies; they require parallel execution.
Meta Descriptions: The Unglamorous Signal That Still Moves the Needle
Google confirmed this week, via Search Engine Journal’s Roger Montti, that meta descriptions carry no direct ranking weight. This will surprise no one who’s been paying attention. What’s worth dissecting is why they remain worth writing — and it’s not the reason most guides will tell you.
Google rewrites meta descriptions roughly 70% of the time when serving search results. But when it does use yours, it’s typically because your description matched the user’s query intent more precisely than anything it could generate from the page body. In other words, a well-written meta description is a signal of content clarity, not a ranking lever.
For local SEO specifically, this has direct application. A Google Business Profile description and a page meta description are doing similar jobs: communicating relevance in a constrained character window to an audience with high purchase intent. Brands operating multilingual storefronts across Bahasa Indonesia, Thai, and Vietnamese should treat meta description localisation with the same rigour as ad copy localisation. A machine-translated description that reads unnaturally will be overridden by Google — and won’t survive in AI citation environments either.
The implementation step most teams skip: audit which pages Google is already rewriting your descriptions for, using Search Console’s performance data cross-referenced against your CMS. If Google rewrites a page’s description consistently, that’s a signal the page body isn’t clearly communicating its core value — fix the content, not just the meta tag.
Local Search in 2026: Parallel Signals, Not a Single Funnel
The through-line across all three developments is this: search is no longer a single system you optimise for. It’s a collection of overlapping signal environments — Google’s personalised AI layer, ChatGPT’s mode-dependent citation logic, and the traditional local pack — each with different content preferences and different proximity-weighting mechanisms.
For Southeast Asian brands, this complexity is compounded by platform fragmentation. A restaurant group in Manila might capture strong Instant-mode citations on ChatGPT for dish-level queries while simultaneously losing ground in Google’s local pack because their GBP photo cadence has slipped. These aren’t the same problem, and they don’t have the same fix.
The teams winning in this environment aren’t publishing more — they’re publishing with more structural precision. Short, answerable content for Instant mode. Authoritative long-form for Thinking mode. Localised meta content to survive Google’s rewrite threshold. And return-visit mechanics to seed the personalization signals that Liz Reid says are coming.
Proximity is still a strategy. But in 2026, it needs a content architecture behind it. The question worth sitting with: if AI search modes are already bifurcating citation behaviour, how many more layers of fragmentation are eighteen months away — and is your content infrastructure built to adapt, or just to react?
At grzzly, we spend a lot of time inside exactly this problem — mapping how local and hyperlocal search signals interact across Google, AI platforms, and Southeast Asia’s unique platform ecosystem. If your team is wrestling with how to structure content for parallel search environments without doubling your production overhead, that’s a conversation worth having. Let’s talk
Sources
Written by
Dusty GrizzlyDeep in the weeds of Google Business Profiles, local pack mechanics, and neighbourhood-level search intent. Believes proximity is a strategy, not a coincidence.