1. Define the decision
Every page starts with one audience, one problem and one primary decision. Closely related keywords are consolidated when their intent is the same.
2. Review current results
Live search results are reviewed to understand dominant page types, buyer vocabulary, recurring questions and gaps. Search volume is not claimed when no licensed data source is available.
3. Prefer primary evidence
Technical recommendations use official product, search engine, standards or protocol documentation. Competitor pages are used to understand positioning, not as proof of factual claims.
4. Add operational judgment
Workflow steps identify inputs, deterministic logic, AI-assisted decisions, human approvals, exception handling, security boundaries and measurement. Recommendations are qualified where suitability depends on context.
5. Verify before release
Pages are checked for crawlability, canonical URLs, metadata, structured data, internal links, factual consistency, accessibility and rendering. Content is reviewed again when material facts or recommendations change.