Heiso 部落格
看 AEO、網站開發、AI 行銷的文章最怕只有理論。阿育是 Heiso 的 blog 寫手，他把研究過、實作過的東西——哪些方法有效、哪些沒效——一次交給你，讓你不只看懂，還能照著做。
Modern B2B forms only need 4 fields. The rest — source page, view history, subscriber status — comes from backend visitor tracking, not from forcing a CV.
AI keeps driving down the cost of code. What to outsource changed: what you can do internally with AI, what to outsource while keeping data yours, and how to gradually agentify your workforce.
Organization schema tells AI "who you are". FAQPage tells AI "when a user asks X, your answer is Y". Here's how we restructured our B2B service pages into the kind of structure AI loves to quote — with full Next.js implementation.
AI is trained mostly on English. If you want AI to understand who you are and what you do — whether you write in Chinese, English, or anywhere in between — follow these three principles.
After shipping AEO changes, how do you know AI is actually citing you? Three layers — manually asking ChatGPT/Perplexity, bot-log analysis, AI referral traffic — from quickest to hardest, one at a time.
SEO brings people; your site introduces them. AEO brings buyers who already evaluated you in ChatGPT. Your site's job shifted from introduction to closing.
The AEO to-do list can run to 20 items, but 80% of the effect lives in three of them. Here's what we did on our own B2B site, with Next.js code.
SEO isn't evolving — it's switching tracks. AEO (Answer Engine Optimization) isn't the next SEO; it's a different game. Here's how B2B sites should get started, based on what we learned.
When B2B product lines multiply, customers can't find the right model. Most sites try to solve this with deep tree navigation + a single page dumping every spec — buyers get lost five layers in, decision-makers drown in 30 spec fields, no one knows where to start reading. The actual problem is two tangled together: search can't find it + the content has no structure. One navigation can't fix both.