SEO Strategy

Produce a written SEO strategy the user can execute immediately: where they can win (topical authority map + winnable keywords), what to publish (content pipeline), how to earn authority (links), and what technical floor must hold. Prime directive: only recommend fights the site can win. A "low difficulty" score is not the same as a winnable keyword — every recommendation must survive a manual SERP check.

When to use / when not to

• Use for site-level strategy: keyword targeting, content roadmaps, link building, SEO audits, "grow my organic traffic."
• If the user needs one article written, hand off to skills/marketing/seo-content (bring the target keyword and intent with you).
• If the user only needs a keyword list, not a full strategy, use skills/marketing/keyword-research.
• If the user's real problem is messaging or conversion, not traffic, route to skills/marketing/positioning-angles or skills/marketing/direct-response-copy.

Intake

Ask in one batch, only what's missing:

1. Site URL and what the business sells (and to whom).
2. Current state — roughly how much organic traffic, any existing content, domain age. Access to Google Search Console data if available.
3. Topics of genuine expertise — what could they credibly write 20+ articles about?
4. Constraints — content capacity (articles per month they can realistically produce/review) and budget for tools or outreach.

Infer from the site itself whenever you can (crawl/fetch it, inspect existing pages). If the user gives a URL and a one-line goal, state your assumptions about the rest and proceed — don't stall.

Workflow

1. Situate the site

Classify the site, because it changes everything downstream:

• New/weak site (little content, few links) → strategy is long-tail winnable keywords only (KGR-qualified), heavy topical clustering, foundational links.
• Established site with content that doesn't rank → strategy leads with a content audit: fix intent mismatches, consolidate cannibalizing pages, internal linking, then expand.
• Ranking site that plateaued → strategy leads with link building and topical gap-filling against the top competitor.

Identify the 2–3 organic competitors actually occupying the SERPs the user wants (not who the user thinks competes with them).

2. Build the topical authority map

Google ranks sites that demonstrate comprehensive expertise in a topic, not isolated pages. Define 3–5 core topics where the user has genuine expertise (E-E-A-T: experience, expertise, authoritativeness, trust), and for each, 10–30 supporting subtopics. Each core topic = one pillar page + supporting cluster articles interlinked to it. If the user can't credibly cover a topic in depth, cut it — a thin cluster is worse than no cluster.

3. Qualify winnable keywords

For each cluster, expand candidates (Google Autocomplete, People Also Ask, Reddit/forum questions, Google Keyword Planner, Search Console queries the site almost ranks for). Then qualify — never trust a tool's difficulty score alone:

• Manual SERP check: forums, Quora, thin listicles, or outdated pages on page 1 = opportunity. All big brands with exact-match titles = skip.
• KGR (Keyword Golden Ratio) for new/weak sites: allintitle count ÷ monthly search volume. KGR < 0.25 = prime target. (Heuristic, not law — still eyeball the SERP.)
• Intent: classify each keyword as informational / commercial / transactional / navigational and note the content format Google is rewarding. Wrong format = no ranking regardless of quality.
• Score each surviving keyword 1–5 on: intent match to the business, winnability, business value, SERP opportunity. Shortlist those scoring 14+/20; the shortlist goes in the output document.

4. Design the content pipeline

Match pipeline ambition to the user's stated capacity — never plan more than they can review for quality:

• Editorial pace (≤8 articles/month): brief each article individually; execution goes through skills/marketing/seo-content.
• Programmatic scale (data-backed page sets: locations, integrations, comparisons, templates): design one template (meta title/description, H1, intent-matching intro, H2 sections, FAQ block, internal links, schema), a data source per page, and a mandatory human editorial review step. AI drafts; a human owns strategy, fact-checking, and quality control — flag this split explicitly.

Every article slots into a cluster and links to its pillar. Specify the internal-linking rule in the plan (each new article links to its pillar + 2–4 siblings).

5. Plan link acquisition

Prioritize in this order:

1. Linkable assets — 1–2 pieces built to attract links: original research/data, free tools, comprehensive canonical guides, templates. Pick the asset type that fits the niche (data journalism niches → original research; practitioner niches → tools/templates).
2. Targeted outreach — build a list of ~50 sites that already link to comparable content; personalized pitch referencing their specific article; make linking frictionless (direct URL, suggested anchor, preview); one follow-up after 5–7 days, then stop.
3. Digital PR — HARO/journalist requests, podcast guesting, guest posts on relevant sites only.

Hard rule: never recommend paid link schemes or exchanges with low-quality/irrelevant sites — one relevant, authoritative link beats a pile of junk links, and junk links carry penalty risk.

6. Set the technical floor

Don't produce a full technical audit unless the site is established and underperforming (Step 1). Minimum bar for every strategy: pages indexable and in the sitemap, Core Web Vitals passing (LCP < 2.5s, CLS < 0.1), mobile-friendly, appropriate schema markup (FAQ/Article/Product/LocalBusiness) on templated and pillar pages. Anything failing this floor goes in Phase 1 of the plan before content scales.

7. Assemble the strategy document

Sequence the plan by dependency, not by calendar: foundation (audit fixes + topical map + keyword shortlist) → content production (pillars first, then clusters) → link acquisition (needs content to link to) → review-and-expand (needs ranking data to act on). Define the measurement plan: Search Console (impressions, clicks, average position per cluster), index coverage, referring domains. Set expectations honestly — new content typically needs months, not days, to rank; say so rather than promising traffic numbers.

Required output format

Deliver the strategy as this document (markdown, concise):

# SEO Strategy — [site]

## Situation
- Site class: [new/weak | established-not-ranking | plateaued] — [evidence]
- Real SERP competitors: [2–3, with one-line strength notes]
- Biggest blocker to organic growth: [one sentence]

## Topical Authority Map
| Core topic (pillar) | Why credible (E-E-A-T) | Supporting subtopics (count) |
| ... | ... | ... |

## Winnable Keyword Shortlist
| Keyword | Intent | Volume (est.) | Winnability evidence (SERP note / KGR) | Score /20 | Cluster |
| ... | ... | ... | ... | ... | ... |
(Top 10–20 only; full candidate list in an appendix if generated)

## Content Pipeline
- Mode: [editorial | programmatic] — [why]
- Cadence: [n articles/pages per month, matched to user capacity]
- Order of production: [pillar/cluster sequence with rationale]
- Internal-linking rule: [rule]
- [If programmatic] Template spec + data source + editorial review step

## Link Plan
- Linkable asset(s): [what + why this type fits the niche]
- Outreach: [target profile, list size, pitch angle]
- Digital PR: [1–2 channels worth the effort, or "skip for now"]

## Technical Floor
- [ ] Pass/fail per item: indexability, sitemap, CWV, mobile, schema
- Fixes required before scaling content: [list or "none"]

## Phased Plan (by dependency)
1. Foundation: [tasks]
2. Content: [tasks]
3. Links: [tasks]
4. Review & expand: [what data triggers what decision]

## Measurement
- Track: [GSC metrics per cluster, index coverage, referring domains]
- Review cadence: after each content batch; kill/fix/scale rule: [rule]

Quality bar

Before delivering, verify:

• Every shortlisted keyword has SERP-check evidence written down, not just a tool score.
• Every keyword's intent is classified and the planned content format matches it.
• Every planned article belongs to a cluster in the map — no orphan topics.
• Content cadence ≤ the user's stated review capacity.
• No traffic guarantees, no invented benchmark statistics; rules of thumb (e.g., KGR < 0.25) labeled as heuristics.
• No paid-link or link-scheme recommendations.
• Plan is sequenced by dependency with no calendar-week scaffolding.

Integration

• skills/marketing/keyword-research — feeds this skill a raw candidate keyword list; this skill qualifies and prioritizes it.
• skills/marketing/seo-content — consumes the keyword shortlist + cluster map; hand each keyword over with its intent and SERP notes.
• skills/marketing/content-atomizer — consumes published articles as seed content for social distribution.
• skills/marketing/newsletter — consumes SEO traffic as its subscriber-acquisition source; note the capture mechanism in the strategy if audience-building is a goal.
• skills/marketing/orchestrator — routes here when a user's growth problem is organic acquisition.

References

• references/Complete_SEO_Playbook.md — the full agency playbook: detailed keyword-research walkthroughs, programmatic SEO build-out, outreach scripts, technical checklists, and case detail. Read it when you need step-by-step depth on any single workflow step (e.g., building the programmatic template, running the outreach campaign) or when the user asks "how exactly do I do X".