Search Console Quick Wins: Striking-Distance + CTR Fixes

Differentiation (read this first) Unlike keyword-cannibalization or content-decay audits, this pack finds the quick wins in a Google Search Console export: striking-distance queries (average position 4–20 with real impressions) and pages whose CTR is far below what that position earns on your own site. It ranks pages by estimated extra clicks, then drafts titles/metas from the page's real queries plus on-page actions. It never invents metrics, and every click figure is labelled as an estimate.

Sample output (real script run on the included fictional Harborline Outdoor export, 76 rows — 2026-10-19)

rows=73 | brand_excluded=4 | opportunities=13 | striking=10 | ctr_gap=1 | both=2 | do_not_touch=3 | est_click_upside ~5374 (estimate)

Sample rewrite (#1): Title "How to Choose a Kayak Paddle: Length Chart + Size Guide" (55 chars) · On-page top query in H1 + first H2, answer "what size kayak paddle do i need" in the first 100 words, FAQ for "best kayak paddle for beginners", 2 internal links with query-matched anchors · Needs check SERP features, whether the page really has a length chart.

The problem GSC shows thousands of rows and no priorities. Asking a chatbot "find my quick wins" gets you generic advice or made-up CTR benchmarks. Cannibalization and decay tools answer different questions.

What's inside

• scripts/gsc_quick_wins.py — stdlib Python: reads query+page exports or the standard Queries.csv / Pages.csv (Top queries, Clicks, Impressions, CTR as "3.2%", Position; localized headers; comma/semicolon/tab; UTF-8/UTF-16)
• Normalizes URLs (utm_*, gclid, srsltid, fragments, trailing slash) and merges duplicates
• Expected-CTR curve derived from your own rows per position bucket; a rough default (clearly labelled planning assumption, not a statistic) only where your data is thin; or pass your own curve
• Flags striking-distance / CTR-gap / both, brand queries excluded with --brand, a do-not-touch list of over-performing pages
• CSV + query-level CSV + Markdown report; SKILL.md method for title/meta drafts (≤60 / ≤155 chars), on-page actions and "Needs check" flags
• Template + fictional example input + the real outputs above

How to use

• Claude Code / Cursor / Codex: unzip into your skills folder (e.g. ~/.claude/skills/gsc-quick-wins/), drop your GSC export in the project and say: "Find the quick wins in this Search Console export and rewrite titles for the top 5 pages. Brand: acme."
• ChatGPT or any chat agent without Python: paste SKILL.md plus your rows; the method includes a manual path (median CTR per bucket, flags, scored list, show working).

Trust No invented data: every number comes from your export · click upside is always labelled an estimate, never a forecast · SERP features and anything missing from the export are marked "Needs check" · stdlib Python only, no network · v1.0.0, Oct 2026 · questions answered within 24h · Cygnix Labs · More tools: https://apify.com/cygnix_labs

Not this skill's job (use these instead): Fix Keyword Cannibalization Using Google Search Console · Find Decaying Pages: GSC Triage + Content Refresh Plan. Pairs with: Featured Snippet Optimizer — Win Position Zero · SEO Meta Tags Generator · SERP Intent Mapper.