candidate-selection

Select one publishable SEO opportunity from verified Ahrefs evidence.

SKILL.md

GitHub

Candidate Selection

Inputs

Current Ahrefs evidence packet; content inventory; product truth; current site authority; country; existing-URL/cannibalization check.

Method

  1. Reject incomplete, conflicting, or stale evidence. Broad research may be 30 days old; the selected overview and SERP packet must be within 14 days.
  2. Establish traffic potential from parent topic and top-page footprint, not one keyword’s volume.
  3. Classify intent from actual SERP page types and reader job.
  4. Score Business Value: 3 natural direct customer path; 2 meaningful product help; 1 support/nurture; 0 no credible path.
  5. Assess link reality using difficulty, top-ten referring domains, weak-link winners, site ceiling, internal links, and the promotion/linkable-asset plan.
  6. Test AI-overview risk: a topic fails when the answer removes any differentiated reason to click.
  7. Test cannibalization against existing pages. Prefer the canonical page or a relaunch when appropriate.
  8. Name weak ranking URLs, their weakness, and the exact counter-asset.
  9. Declare the Article Promise: reader job, portfolio role, uniqueness asset, authority source, conversion path, and internal-link/promotion plan.

Decisions

  • GO: all evidence is present, Business Value is at least 2, intent and link reality fit, and the Article Promise is specific.
  • KILL: no fit, no click reason, no realistic link plan, unresolved cannibalization, or cosmetic uniqueness.
  • DATA_REQUIRED: identify the exact report, country, fields, filters, freshness, and accepted attachment format.

Output

Candidate brief with evidence citations, the Course Formula (Business Value × Traffic Potential ÷ max(1, Estimated Links Needed)), vulnerability verdict, and decision.