AI Citations Are the New Backlinks: How to Get Cited

AI answers don’t appear from thin air. They’re assembled from specific pages — and the brands on those pages get mentioned. The brands that aren’t? They don’t exist in AI’s version of the market.

GEO Is Now a Source Selection Problem

Generative Engine Optimisation has moved past the theory stage. Semrush has documented a concrete workflow: use their tools to identify which third-party pages AI models consistently cite when answering questions in your category, then run structured outreach to get your brand’s facts embedded on those pages. The mechanic is closer to PR and link-building than traditional on-page SEO — you’re not optimising your own site, you’re colonising the sources that AI trusts.

For brands operating in Southeast Asia, this has a specific texture. AI answers about, say, ride-hailing options in Bangkok or e-commerce platforms in the Philippines are being assembled from a narrower pool of English-language authority sources — regional tech publications, review aggregators, and platform documentation. That concentration is actually an opportunity: the citation landscape is less competitive than in mature Western markets, and a focused outreach campaign to five or six key regional publishers can shift your brand’s AI visibility meaningfully.

The tactical starting point is mapping which URLs appear as citations when you query AI tools about your product category. That list becomes your outreach target sheet.

The Slop Problem Is Real, and It’s a GEO Risk

Here’s where strategy meets execution risk. Shopify CEO Tobi Lütke recently called unreviewed AI output “slop grenades” — content that looks finished but lands as someone else’s problem to clean up. Search Engine Journal reports that freelance marketplaces are seeing rising demand specifically for AI content remediation work. This isn’t just an internal productivity concern; it’s a GEO liability.

If your brand’s citation outreach produces AI-generated content placed on third-party sites without human review, you’re injecting inaccurate or generic brand facts into the exact sources that AI models will later cite. The compounding effect is brutal: bad information gets embedded in authority sources, AI pulls from those sources, and your brand description in AI answers becomes a distorted echo of whatever slop got placed six months ago.

The fix isn’t avoiding AI in your outreach workflow — it’s treating human review as a non-negotiable production step, not an optional quality check. For Southeast Asian markets specifically, this matters even more: product details, pricing structures, and regulatory contexts vary enough across markets that a Bangkok-sourced brand fact placed on a regional outlet may actively mislead AI answers serving users in Jakarta or Kuala Lumpur.

---

What This Means for Local and Hyperlocal Search

Local SEO teams have spent years understanding that proximity signals — Google Business Profile completeness, review velocity, local citation consistency — are what push brands into the local pack. GEO citation strategy is the same mechanic operating at a different layer. Instead of convincing Google’s local algorithm that you’re the most relevant result for a neighbourhood search, you’re convincing a language model that your brand is the most citable source of accurate facts about your category.

The implementation parallel is instructive. Local SEO requires consistent NAP (name, address, phone) data across dozens of directory listings. GEO citation strategy requires consistent brand fact data — founding year, product descriptions, market positioning, pricing tiers — across the specific authority pages AI models trust. Both break down when there’s no single owner of the source-of-truth data.

For multi-location brands in Southeast Asia — a regional F&B chain, a fintech with country-specific products — this means building a brand fact sheet per market, not a single global document. What’s true for your Singapore operation may be materially different from your Vietnam one, and AI answers are increasingly localised enough to notice the difference.

Building the Workflow Without Breaking the Team

Semrush’s documented approach pairs their keyword and citation data with Claude to automate the initial research and outreach drafting phases. That’s a reasonable division of labour: AI handles volume and first-draft generation, humans handle accuracy verification and relationship judgment. The risk of inverting that — letting AI make the judgment calls while humans just approve at speed — is exactly what Lütke is warning about.

A practical three-step workflow for Southeast Asian growth teams: First, query five to ten AI tools with your core category questions and export every cited URL. Second, cross-reference that list against your existing media relationships — warm outreach converts faster than cold. Third, produce placement content with a mandatory human fact-check against your regional brand fact sheets before anything goes live. Budget roughly two to three weeks per citation placement when you factor in outreach, negotiation, and production. This is not a quick-win channel; it’s a compounding asset that builds AI share-of-voice over a six-to-twelve month horizon.

The question worth sitting with: if AI answers are already shaping how your potential customers understand your category — and they are — who at your organisation is accountable for what those answers say about your brand?

---

At grzzly, we work with Southeast Asian brands to map their AI citation exposure, identify the authority sources that matter for their specific markets, and build outreach strategies that hold up under human review. If your team is starting to think seriously about GEO and what it means for your search presence, Let’s talk