The end-to-end content SEO playbook: keyword research, topic clusters, content briefs, writing standards, on-page optimization, and the audit-and-update loop that compounds over years.
Most content SEO programs fail at the research stage by chasing volume over intent. The framework we use:
1. **Map the customer journey first.** What does someone search before they even know they need you? When they are evaluating options? When they are ready to buy? Each stage gets different content. 2. **Build the seed list from your sales calls.** The five questions every prospect asks become five pages. Real demand, no guessing. 3. **Layer in keyword data** from Search Console (already-ranking, opportunity to deepen), Ahrefs/Semrush (competitor coverage and difficulty), and AlsoAsked / AnswerThePublic (question expansion). 4. **Cluster by topic, not by keyword.** A topic cluster is a single deep page (the *pillar*) that links to many narrower pages (the *spokes*). This guide is a pillar; our tool roundups and glossary are spokes. 5. **Validate with the SERP.** Open every target query and look at the top 10. If they are all videos, you are wrong about the format. If they are all listicles, do not write a long-form essay.
Measure topic difficulty using the SERP itself, not aggregate domain scores. A query with 10 weak result pages is winnable regardless of the head term's KD.
A great brief is the difference between a page that ranks and a page that costs $1,500 and dies. Every brief we ship to a writer includes:
- **Target query and search intent classification** (informational / navigational / transactional / commercial investigation). - **Target word count range** based on the top-3 SERP results, not a default. - **Required H2s** with required sub-points under each. - **Required entities to mention** (people, products, places, concepts) for topical completeness. - **Required citations** (primary sources, government data, peer-reviewed research). - **Schema requirements** (FAQ, HowTo, Article, etc.). - **Internal links to include** with anchor text suggestions. - **Author and reviewer assignment.** - **Snippet target** — what we want to win in AI Overviews / featured snippets.
Use our keyword density tool and meta tag analyzer to validate as you brief.
Across thousands of client pages we have settled on these as the standards that compound:
- **Direct first paragraph.** Tell the reader what they are about to learn in the first 60 words. AI engines and humans both reward this. - **Short paragraphs.** 1–4 sentences. Wall-of-text underperforms in 2026 dwell-time data. - **Active voice.** Passive constructions read more slowly and get cited less by AI engines. - **Concrete examples and numbers.** "Three of our 41 clients saw…" beats "some clients see…" - **Named author with credentials.** Pen names rank, but they don't earn AI citations. - **No filler.** Every paragraph either answers a question, makes an argument, or shows evidence. If a paragraph does none of those, cut it. - **Reading level matched to audience.** Our readability checker defaults to Grade 9–11 for B2B and Grade 7–9 for consumer. - **Genuine first-party experience.** Where you have done the thing, say so. Where you have not, cite who has.
Modern on-page SEO is about completeness, not density. The checklist:
- **Title tag** under 60 characters, includes the focus keyword once near the front, and reads like a result a human would click. Test in our SERP preview tool. - **Meta description** under 160 characters, written as ad copy not a summary. - **Single H1** matching the focus keyword's intent (does not have to match exactly). - **H2s structured as the questions the page answers.** AI Overviews lift heavily from this structure. - **First paragraph** — direct answer to the H1's implicit question. - **Internal links** in-context, with descriptive anchor text. 3–10 per page is typical. - **External links** to primary sources where you cite them. - **Images** with descriptive alt text — not keyword-stuffed alt text. - **Schema** appropriate to the page type (Article, FAQ, HowTo, LocalBusiness, etc.). - **Authorship signals** — visible byline, link to author bio, last-updated date.
Validate density and overuse with our keyword density tool. The thing to avoid is the *appearance* of keyword stuffing; modern Google reads as a human would.
New content is the headline; updating old content is the compound interest. The cadence we run:
**Quarterly content audit:**
- Pull every URL with at least one Search Console click in the last 90 days. - Tag each by performance trend (rising, flat, declining). - For declining: investigate (lost a link? new competitor? stale data?). Fix or sunset. - For flat: identify the next-step opportunity — better internal linking, better SERP-match, schema upgrade. - For rising: protect with internal links from related content.
**Annual content prune:**
- Identify pages with zero clicks in 12 months. - For each: improve, consolidate (301 to a stronger page), or remove. - Thin / outdated content actively suppresses the rest of the site under modern Google quality signals.
This loop is what separates a site that grows linearly from a site that compounds. We document the process in detail in our methodology.
**Free tools we use daily** — many are on this site:
- Google SERP Preview — title and meta validation - Keyword density — overuse check - Readability checker — grade-level fit - Meta tag analyzer — full meta audit - Lorem ipsum generator — for layout drafts
**Paid tools that earn their fee** for content programs:
- **Ahrefs or Semrush** — for competitive content gap analysis. - **AlsoAsked** — for People Also Ask expansion. - **Frase, Surfer, or Clearscope** — for topic-completeness scoring (use with judgment, not as a rigid rubric). - **Notion or Airtable** — for the editorial calendar and content briefs.
See our complete tool roundups for content SEO, keyword research, and link building.
As long as it needs to be to fully cover the query, no longer. The top-3 SERP results are the data point — not a default 2,000 words. We see top-ranking 600-word answers and top-ranking 6,000-word guides; both are correct for their query.
Cadence matters less than quality. One excellent piece per month outperforms four mediocre pieces. Most clients we run ship 4–8 substantive pieces per month, plus updates to existing pages.
The best content for SEO is written or substantially edited by a subject-matter expert with operating experience. Outsourcing fully to a freelance generalist is the single most common reason a content program does not work.
Google's stated position is that production method does not matter — quality does. In our experience, the actual issue with AI content is not 'AI penalty' but accuracy. AI without expert review ships inaccurate content. That is what fails.
Start with the questions your sales team is tired of answering. Layer in Search Console queries you already rank for in positions 5–20 (huge opportunity to deepen). Use AlsoAsked to expand each into a topic cluster.
A pillar page covering a broad topic exhaustively, plus 5–30 narrower spoke pages each covering a sub-topic. The pillar links to every spoke; every spoke links back to the pillar. This guide is a pillar.
Track clicks, impressions, and average position in Search Console — by URL and by query, monthly. Track conversions from organic in GA4. Track AI Overview citation rate with a purpose-built tool.
Update if the page has any clicks and the topic has not fundamentally changed. Write new if you would be writing a substantially different angle, or if the existing page is so outdated a redirect is cleaner.