Our Python scraping pipeline hits Google SERP, Pantip, Reddit, app stores, and AI assistants every night across 10,000+ Thai and English queries. We publish what we find — partly because the writing forces us to think clearly, partly because "thought leadership" is a moat that compounds.
We tracked 4,200 prompts across ChatGPT, Claude, Perplexity, and Gemini in Thai and English over 90 days. Here's what gets cited, what gets ignored, and what changed since 2025.
aeo chatgpt perplexity thai-marketForum content has E-E-A-T advantages most brands ignore. We scraped 2,800 SERPs where Pantip threads outrank brand pages — and identified the 6 patterns brands can replicate.
thai-seo pantip long-tailMost "topic cluster" advice is qualitative. We built a graph-theoretic model trained on 74 client clusters and identified the link density that consistently transfers authority to pillar pages.
topic-clusters internal-linking graph-theory90% of TH/EN bilingual sites we audit have hreflang configuration that hurts rankings. Common failure modes, the correct setup, and a free validator.
hreflang bilingual technical-seoOpen data drop. 50 search queries growing 200%+ YoY in Thai market across SaaS, finance, beauty, hospitality, and education verticals. With volume estimates, intent classification, and SERP-feature analysis. Updated quarterly.
thai-market keyword-trends scraper-report open-data1. Specificity wins. Pages with explicit numerical claims ("reduces churn by 23%") get cited 3.2x more often than pages with hedged claims ("can reduce churn"). AI models prefer commitable statements.
2. FAQPage schema is necessary but not sufficient. Schema gets you parsed; clear answer-first writing gets you cited. Lead with the answer in 1-2 sentences, then explain.
3. Thai-language pages get cited disproportionately when the query is Thai. Models route queries by language family. Don't translate — author natively in Thai. A poorly-translated TH page loses to a well-written original even from a smaller domain.
4. Citations cluster around 8-15 domains per topic. Once you're in the cluster, you compound. The barrier-to-entry is high, but so is the moat once you cross it.
5. Update frequency matters. Models de-weight content older than ~14 months for time-sensitive topics. Refresh schedules built into your editorial system, not as ad-hoc projects.
Quarterly trend report. No spam. Unsubscribe in one click.