1. Opportunity and intent screening
Before research begins, the system checks whether the proposed topic is genuinely distinct from existing coverage. A wording variation should expand or merge into an existing page rather than create a near-duplicate URL.
2. Research and fact pack
Research is collected into structured source records and claims. Material claims such as specifications, compatibility, manufacturer behavior, maintenance intervals, electrical details, pressure figures, or safety instructions require supporting evidence.
3. Content brief
The brief defines the primary user situation, questions that must be answered, sections that are needed, sections that should not be added, internal-link targets, visual candidates, and any safety considerations.
4. Draft and independent review
The draft is evaluated by separate factual, technical, English, SEO, and visual checks. Review issues become structured revision tasks instead of vague requests to make the article longer.
5. Final quality gate
Publication requires all hard gates to pass. A high aggregate score cannot compensate for a known critical safety issue, unsupported material technical claim, or failed technical review.
6. Search feedback and refresh
After publication, search performance can identify weak titles, new query clusters, decaying pages, orphaned content, or possible cannibalization. Refresh decisions use sustained evidence rather than reacting to every small ranking change.