Squeeze Max Traffic

Squeeze Max Traffic Skill

Strategy anchor: STRATEGY.md §3 (full traffic potential of the top page, not a single keyword's volume) + the Lesson 5 row ("squeeze max traffic; wide kw family") + Lesson 3 (one strong page ranks for ~1,000 related keywords). A page that nails its parent topic should pull traffic from the whole family of related queries, not just its head term. This step makes sure the drafted article actually reaches for that family — without ever crossing into the keyword-stuffing line (anti-pattern #10) or padding (the empty-paragraph test in /draft).

This is Lesson 5, part 3 of the Ahrefs method, operationalized as a discrete pipeline step: after the article is drafted, audit it against (a) the keywords this page can realistically rank for, and (b) the Content Gap — keywords competitors rank for that we don't — and weave the worthwhile ones in as genuine added value. Coverage we add must read like the writer always meant to cover it; if a keyword can't be served with a real paragraph a reader benefits from, it doesn't go in.

Where this sits in the pipeline

/research → /outline → /product-mentions → /draft → [ /squeeze-max-traffic ] → /quality-check → /verify-claims → /generate-visuals → /format-for-publish

After /draft, before /quality-check. It reads the drafted body at content-pipeline/5-drafts/{slug}.md, expands it in place, and re-saves to the same path so the quality gate judges the squeezed article (including the uniqueness gate — added paragraphs must still be original, never page-1 clones). Run it once per article; re-run only if /research is re-run and the keyword family shifts.

Tool calls: this step reuses the Ahrefs MCP exactly as /research and /content-gap-analysis do. Read ../research/references/ahrefs-mcp-cheatsheet.md first — params are comma-separated strings, not JSON arrays (keywords:"ai girlfriend app", not ["ai girlfriend app"]); select + country are required on most endpoints; call doc {tool:"..."} for any tool you haven't used this run; never invent tool names. Ahrefs outage policy is the same as /research: MCP first, REST fallback (surface it loudly), and if both are down, do the keyword-family expansion from the data already in 1-research/{slug}.md rather than blocking the pipeline — note the degradation in the report.

Input

For slug {slug}:

• content-pipeline/5-drafts/{slug}.md (required — the drafted body this step expands; byline comment stays the first line, untouched)
• content-pipeline/1-research/{slug}.md (the dossier — primary keyword, parent topic, the BEAT SPEC's "Secondary keywords" + consensus/partial/gap topics; the matching/related-terms pool already lives here)
• content-pipeline/3-outlines/{slug}.md (coverage map — what the article already deliberately covers)
• content-pipeline/0-keywords/keyword-ideas.csv + cache/competitors.json (if present — the resolved competitor set and the gap pool from /content-gap-analysis)
• brand-config.md (audience, products, forbidden phrases — added prose obeys the same voice/forbidden rules as /draft)

Process

1. Map what the page already covers. Read the draft and the outline's coverage map; list the H2/H3 topics and the head + secondary keywords already worked in. This is the baseline — we expand the family, we don't re-add what's there.

2. Pull the full keyword family for THIS page (Lesson 3 — the page ranks for ~1,000 related keywords, not one). From the dossier's primary keyword and parent topic:

   • keywords-explorer-related-terms ("also rank for" / "also talk about") and keywords-explorer-matching-terms (match_mode:"phrase", then match_mode:"terms" for multi-word) → the variation pool for the parent topic. select to carry keyword,volume,difficulty,traffic_potential,parent_topic,intents.
   • site-explorer-organic-keywords on the top-ranking page in this SERP (limit ~30, order_by traffic_potential:desc) → the secondary keywords the winning page already earns traffic from. These are the proven sub-queries a strong page on this topic captures (the dossier's step-5 authority check already started this; go wider here).
   • Keep only same-intent, same-parent keywords — a query that needs a different article shape is a separate cluster member for the queue (route it to /keyword-prioritization, do NOT bolt it on here; mismatched intent never ranks — STRATEGY §6, anti-pattern #9).

3. Pull the Content Gap (Lesson 5 part 3 — keywords competitors rank for but we don't). Reuse /content-gap-analysis: if content-pipeline/0-keywords/keyword-ideas.csv already holds a fresh gap (gap_mode=missing) from a recent keyword-research run, filter it to rows whose parent topic / cluster matches this article and use those. If there's no fresh gap for this topic, derive a quick page-level gap directly: take the resolved competitors (cache/competitors.json, or auto-discover via site-explorer-organic-competitors), pull each competitor's site-explorer-organic-keywords for terms in this parent topic, and keep the ones they rank for and our drafted article doesn't cover. Tag these source=content_gap.

4. Merge, dedupe, and TRIAGE — keep only the worthwhile. Combine the family pool (step 2) + the gap pool (step 3); drop anything already covered (step 1). Then triage each remaining keyword — most candidates do NOT make it in. A keyword earns a place only if all hold:

   • Same searcher intent as the article (no intent mismatch — STRATEGY §6).
   • On-topic for the parent — a reader of this article would genuinely expect it covered (not a tangent that belongs to another cluster member).
   • Servable with real value — you can add a true paragraph/sub-section a reader benefits from (a specific, a step, a comparison, an answer), not a sentence whose only job is to contain the phrase.
   • Worth the words — meaningful volume/traffic_potential, or it closes a consensus/gap a competitor covers and we don't. Trivial near-duplicates of a phrase already present get merged into existing prose, not added as new sections. Output a short squeeze plan: each kept keyword → the section it strengthens (or the new sub-section it justifies) → the value the added prose delivers.

5. Weave it in as natural prose — never stuff (anti-pattern #10; the /draft empty-paragraph test). Apply the plan:

   • Add real paragraphs / sub-sections where a kept keyword justifies genuinely new coverage — written in the article's persona/voice (the byline persona already chosen in /draft; keep its craft) and to the same specificity bar as /draft (named things, numbers cited with [link] for verify-claims, steps, first-hand product detail). A new sub-section gets a real H3 and earns its length.
   • Fold near-variants into existing sentences where a separate section would be redundant — phrase the existing point using the searcher's actual language so the page is matchable for that variant without a new block.
   • Keep the authority + uniqueness intact: added coverage must preserve the dossier's product-led authority element and must not turn a unique angle into a page-1 clone (the quality gate's Gate 1.5 still has to pass). New paragraphs are original — never lifted/paraphrased from a competitor we found in the gap step.
   • Respect the existing structure and components: don't bloat the article past what its content earns; don't duplicate a native ::: component; don't stack or duplicate visuals. If a kept keyword maps to a comparison/FAQ the article should have, extend the existing table/:::faq rather than starting a parallel one.
   • Give a genuinely new section its visual (the squeeze→visuals seam — DON'T leave squeezed content visual-less). The outline planned visuals BEFORE this expansion, and /generate-visuals runs AFTER it — so a brand-new H3/sub-section you add here would otherwise ship with no visual. If a new section would genuinely benefit from one, add ONE typed [VISUAL:...] placeholder inside it per templates/visual-strategy.md — value-first (a real screenshot, a chart from resolvable research.<key> data, a diagram, or an external competitor/SERP shot), and never a PNG of a stat/quote/table/callout a native ::: component already renders. PRESERVE every existing [VISUAL:] marker untouched — only ADD, never edit or delete the ones /draft and /outline already placed. (Externals you add resolve via /generate-visuals → /capture-visuals automatically.)
   • Stay light on on-page SEO (Lesson 5 part 2): intent-match and useful coverage, NOT keyword density. No exact-match repetition for its own sake; if a sentence only exists to hold a phrase, cut it.

6. Depth + forbidden-phrase re-check. Re-read top to bottom: every added paragraph passes the empty-paragraph test (a concrete noun/number/step/example), no forbidden phrase entered, no crutch word now used 3+ times because of the additions, rhythm still varied. The article total may grow, but only by content that earns its place — never pad to a number.

7. Save the expanded body back to content-pipeline/5-drafts/{slug}.md (overwrite; byline comment still the first line). Write the squeeze plan + the kept/rejected keyword lists to content-pipeline/5-drafts/{slug}-squeeze.md for the audit trail and so /quality-check (and a human) can see what was added and why.

Output

• content-pipeline/5-drafts/{slug}.md — the drafted article, expanded to capture the worthwhile keyword family + content gap, woven in as natural prose. Same byline, same persona voice, same structure where possible.
• content-pipeline/5-drafts/{slug}-squeeze.md — the squeeze plan: kept keywords (→ section → value added) and rejected keywords (→ reason), plus a note on any Ahrefs degradation.

Quality checklist

Before saving, confirm:

• Pulled the page's keyword family (related/matching terms on the parent topic + the winning page's organic keywords) — same-intent, same-parent only
• Pulled the Content Gap (reused /content-gap-analysis's missing rows for this topic, or derived a page-level gap from competitors) — keywords competitors rank for that the draft didn't cover
• Triaged: every kept keyword is same-intent, on-topic, servable with real value, and worth the words; off-cluster / intent-mismatch keywords routed to /keyword-prioritization, not bolted on
• Added coverage is natural prose — real paragraphs/sub-sections (or folded variants), to the /draft specificity bar; zero keyword stuffing (anti-pattern #10), no sentence that exists only to hold a phrase
• Persona voice preserved (byline unchanged, first line intact); structure/components not bloated or duplicated; no stacked visuals
• Every existing [VISUAL:] marker left untouched; any genuinely-new section that warrants one got exactly ONE value-first [VISUAL:] added (no native-component duplication) so squeezed content isn't visual-less
• Uniqueness + authority preserved — additions stay original (no page-1/competitor clone; Gate 1.5 still passes) and the dossier's product-led authority element survives
• Empty-paragraph test passes on every addition; no new forbidden phrase; no crutch word now ≥3×; rhythm still varied
• {slug}-squeeze.md written (kept → section → value; rejected → reason; any Ahrefs-outage degradation noted)

When to skip / when it's a no-op

• Skip if the article already covers the full same-intent family and there's no matching content gap (a tight, narrow topic). Write a one-line {slug}-squeeze.md saying "no worthwhile family/gap expansion — already complete" and pass through. A no-op is a valid, honest outcome; never invent thin sections to look busy.
• Don't use this step to chase a higher-volume but different-intent keyword (that's anti-pattern #9) or to push the article above its winnable scope — those are queue decisions for /keyword-prioritization, not draft-time additions.