Build Log / Mar 29–30
Week 4: The SEO Bet
Twitter sent 2 visitors. Shipped 26 new pages for SEO. Sitemap went from 10 to 32 URLs. Playing the long game.
Twitter: 2 visitors
Spent a week promoting on Twitter/X. Total result: 2 unique visitors. Two. Meanwhile Reddit and LinkedIn continued sending 75+ each without extra effort. Not every channel works. The algorithm buries external links. Moving on.
26 new pages in one session
Google had only indexed 2 pages and was showing wrong descriptions. Shipped a full SEO overhaul: /learn pages (article versions of all 9 lessons), /compare pages (LangChain, CrewAI, AutoGen, OpenAI SDK vs plain Python), and /blog with 4 cornerstone articles. All content auto-generated from existing lesson data — zero new writing needed. Every page has full metadata, JSON-LD structured data, and links back to the interactive course.
The build-vs-buy angle
The comparison and blog content targets the question every team building with LLMs faces: do you use a framework or build from scratch? This is the search query that brings engineers to the course. "LangChain alternative," "build AI agent without framework," "what does AgentExecutor actually do." SEO is a slow channel but it compounds. The pages are live, now we wait.
Where things stand
742 unique visitors. 578 code runs. 1,246 lessons started. $0 revenue. One organic compliment from a stranger. Reddit and LinkedIn drive 92% of traffic. Google has sent 6 visitors total. The SEO pages should change that over the next 4-8 weeks. The real outcome isn't the product — it's learning distribution by doing it badly in public.