mikefutia/seo-consultant/content-from-data

Content From Data

Data picks the topic, the cluster picks the headings, and existing rankings veto duplicates. That's the entire difference between this and "write me a blog post about X."

Process

1. Take the target cluster (from the user, or the highest-opportunity GAP in the latest runs/intent-map-*.md; run /seo-os:intent-map first if none exists).
2. Cannibalization check before drafting: query GSC for pages already earning impressions on this cluster's queries. If a page already targets it, REFUSE to draft a duplicate — recommend upgrading that page instead, and say which queries it already ranks for.
3. Draft the piece:

• H1 + headings that mirror the actual GSC query phrasings (quote the real queries you used)
• The entities a page on this topic must cover
• Meta title under 60 chars containing the head term + meta description under 155
• Internal-links plan: 3+ links FROM existing relevant pages TO this post (name the source pages and anchor text), and links from this post to the site's money pages

1. Match the brand voice if a CLAUDE.md / brand file defines one; otherwise write clean and direct, zero AI-slop phrases ("in today's fast-paced world," "game-changer," "delve").
2. Save brief + draft to runs/content/[slug].md.

Rules

• Every heading decision cites the query data behind it.
• Never fabricate stats or sources in the draft. A claim that needs a number gets a [NEEDS SOURCE] marker instead of an invented figure.
• One piece per run. Depth beats volume.