Content and On-Page SEO skill 01 of 13

Content SEO Audit SKILL.md

The Content SEO Audit skill makes Claude, Codex or another AI agent review every indexable page on a site or section with real data. It joins Search Console, analytics, crawl and backlink data, scores each page on demand, performance, intent match, quality and business value, and gives every page one outcome: keep, improve, merge, redirect or remove, with a prioritised plan.

content-seo-audit/SKILL.md, version 1.0.0, 1,310 words.

Most sites carry more pages than they can keep in good shape. Some earn traffic, some used to, some never did, and a few compete with each other for the same searches. A content audit looks at every indexable page with real data and decides what to do with each one, so effort goes where it will move traffic and leads.

This skill runs that audit end to end and hands each finding to the specialist skill that can go deeper.

Operating principles

  1. Data before opinion. Every decision rests on clicks, impressions, conversions, links and a reading of the page, not on its age or how it looks.
  2. Judge pages against the query, not in isolation. A page is good if it satisfies the searches it can rank for better than the alternatives.
  3. Five outcomes only: keep, improve, merge, redirect or remove. Every page gets one.
  4. Protect what earns. Never remove or merge a page that brings links, conversions or steady traffic without a plan to keep that value.
  5. Prioritise by impact and effort, quick wins first.

When to use, and when not to

Use it for a site or section with more than a few dozen pages, after a traffic decline across many pages, before a content strategy reset, before a migration, or when nobody knows which pages still matter.

Don't use it for a single page (use on-page-seo or content-refresh), for crawling and indexing faults (use the Technical SEO skills) or for keyword research on new topics (keyword-research).

Inputs to ask for, in one message

  1. Search Console Performance by page and by page plus query, 16 months if available.
  2. Analytics landing page data for organic traffic: sessions, engagement, conversions, revenue, for 12 months.
  3. A crawl with status, indexability, title, meta description, H1, word count, publish and modified dates, author, inlinks and canonical.
  4. Backlink data by target URL.
  5. The business goals: which products, services or conversions matter most.
  6. Any rules about pages that must stay (legal, partner, product pages).

Raw SKILL.md file

.md

This is the exact file agents read, including the credit header and licence notice. Save it as content-seo-audit/SKILL.md.

---
name: content-seo-audit
description: "Run a complete content and on-page SEO audit of a site or section: inventory every indexable page, join Search Console, analytics and crawl data, score each page on intent match, quality, E-E-A-T, on-page elements and performance, and decide keep, improve, merge, redirect or remove, with a prioritised action plan. Use whenever someone asks for a content audit, content inventory, blog audit, on-page SEO audit, wants to know which pages to update, merge or delete, or sees organic traffic falling across many articles."
license: CC-BY-4.0
metadata:
  author: Afeez Adebayo
  author_url: https://afeezadebayo.com
  author_linkedin: https://www.linkedin.com/in/afeez-adebayo/
  author_x: https://x.com/za_icon
  source: https://afeezadebayo.com/skills-md/marketing/content-seo/content-seo-audit/
  category: marketing
  subcategory: content-seo
  version: "1.0.0"
  created: "2026-10-10"
  updated: "2026-10-10"
---

# Content SEO Audit

Most sites carry more pages than they can keep in good shape. Some earn traffic, some used to, some never did, and a few compete with each other for the same searches. A content audit looks at every indexable page with real data and decides what to do with each one, so effort goes where it will move traffic and leads.

This skill runs that audit end to end and hands each finding to the specialist skill that can go deeper.

## Operating principles
1. **Data before opinion.** Every decision rests on clicks, impressions, conversions, links and a reading of the page, not on its age or how it looks.
2. **Judge pages against the query, not in isolation.** A page is good if it satisfies the searches it can rank for better than the alternatives.
3. **Five outcomes only:** keep, improve, merge, redirect or remove. Every page gets one.
4. **Protect what earns.** Never remove or merge a page that brings links, conversions or steady traffic without a plan to keep that value.
5. **Prioritise by impact and effort,** quick wins first.

## When to use, and when not to
Use it for a site or section with more than a few dozen pages, after a traffic decline across many pages, before a content strategy reset, before a migration, or when nobody knows which pages still matter.

Don't use it for a single page (use `on-page-seo` or `content-refresh`), for crawling and indexing faults (use the Technical SEO skills) or for keyword research on new topics (`keyword-research`).

## Inputs to ask for, in one message
1. Search Console Performance by page and by page plus query, 16 months if available.
2. Analytics landing page data for organic traffic: sessions, engagement, conversions, revenue, for 12 months.
3. A crawl with status, indexability, title, meta description, H1, word count, publish and modified dates, author, inlinks and canonical.
4. Backlink data by target URL.
5. The business goals: which products, services or conversions matter most.
6. Any rules about pages that must stay (legal, partner, product pages).

## Workflow

### Step 1. Build the inventory
Join crawl, Search Console, analytics and backlink data on URL. Keep only indexable, canonical pages; list non-indexable pages separately for a technical check. Add page type, topic and the main query each page ranks for.

### Step 2. Add trend data
Compare the last 3 months with the same period a year earlier and with the previous 3 months. Flag pages losing more than a set share of clicks as decaying.

### Step 3. Score each page
| Dimension | What to check | Score 0 to 2 |
|---|---|---|
| Demand | Impressions for queries the page targets | 0 none, 1 some, 2 strong |
| Performance | Clicks, position, CTR against similar pages | 0 poor, 1 average, 2 strong |
| Intent match | Does the format and angle match the current results? | 0 mismatch, 1 partial, 2 match |
| Quality and E-E-A-T | Accuracy, depth, first-hand evidence, author, freshness | 0 weak, 1 fair, 2 strong |
| On-page | Title, H1, headings, meta description, internal links, images | 0 weak, 1 fair, 2 strong |
| Business value | Conversions, assisted conversions, link value, strategic fit | 0 none, 1 some, 2 high |

### Step 4. Decide the outcome
| Pattern | Outcome |
|---|---|
| Strong performance and value | Keep; light maintenance |
| Demand exists, page underperforms | Improve (`content-refresh`, `on-page-seo`) |
| Two or more pages split the same query | Merge (`keyword-cannibalisation`) |
| No demand, no value, but has links or traffic history | Merge or redirect to the closest page |
| No demand, no value, no links | Remove with 410, or noindex if users still need it (`content-pruning`) |

### Step 5. Read a sample by hand
Numbers miss quality. Read at least the top 20 pages by traffic and 20 random pages from each outcome group. Note patterns: outdated facts, thin sections, missing authors, broken examples.

### Step 6. Prioritise
Score each action by expected impact (traffic and business value at stake) and effort (hours to fix). Order quick wins first: high impact, low effort.

### Step 7. Plan the work
Group actions into batches by owner and template. Set a re-check date per batch, usually 6 to 12 weeks after release.

## What good looks like
- Every page has one main purpose and one main query group.
- Titles and H1s match what searchers want.
- Content shows first-hand experience where it matters: data, screenshots, examples, named authors.
- Important pages are linked from related content.
- No two pages compete for the same intent.

## Worked example
A SaaS blog had 410 posts. The inventory showed 38 posts brought 81% of organic clicks over 12 months (Search Console), 96 posts had no clicks at all, and 22 pairs of posts targeted the same queries. The plan kept and maintained the top 38, refreshed 54 posts with demand but falling clicks, merged the 22 pairs into 22 stronger posts, redirected 31 old posts with backlinks to their closest replacements, and removed 65 thin posts with no links. Twelve weeks after the first two batches, clicks to refreshed posts were up and the blog had 140 fewer pages to maintain.

## Output format
```
# Content SEO audit: <site or section>, <date>
## Summary (pages audited, share of clicks from top pages, outcome counts)
## Inventory (spreadsheet): URL, type, topic, main query, clicks, trend, conversions, links, scores, outcome, owner
## Patterns found in the manual review
## Prioritised action plan
| # | Action | Pages | Impact | Effort | Skill to use | Owner | Re-check |
```

## Edge cases and judgement calls
| Situation | What to do |
|---|---|
| Seasonal pages with no clicks out of season | Judge on the same season last year, not the last three months |
| New pages under six months old | Mark as "too early" unless clearly thin or duplicate; re-audit later |
| Pages with links but no traffic | Never remove; merge or redirect to keep the link value |
| Pages that convert but get little traffic | Keep and improve visibility; they are worth more than traffic suggests |
| Legal, policy or support pages | Keep regardless of search data; exclude from pruning decisions |
| Pages ranking for queries unrelated to the business | Low value even with traffic; consider refocusing or removing |
| Programmatic or templated sections | Audit by template and sample, then apply decisions to the whole group |
| Multilingual sites | Audit each language separately; a page weak in one market may be strong in another |

## Scoring at scale
For sites with thousands of pages, score automatically first: demand and performance from Search Console, value from analytics, links from backlink data. Then read a stratified sample from each score band. Adjust the automatic rules where the reading disagrees, and re-run. Record the rules so the next audit is comparable.

## Handoffs to specialist skills
- Pages to improve: `content-refresh` for decaying pages, `on-page-seo` for single-page gaps
- Overlapping pages: `keyword-cannibalisation`
- Pages to remove, merge or redirect: `content-pruning`
- Quality and trust gaps: `eeat-content-quality`
- Weak titles and descriptions: `title-tags-meta-descriptions`
- Missing topics: `topical-map` and `keyword-research`
- Indexing or rendering faults found on the way: the Technical SEO skills, starting with `technical-seo-audit`

## Quality bar: what an excellent audit includes
- One outcome for every indexable page, backed by data and a reason.
- A manual reading of samples, with the patterns it found.
- Protection rules for pages with links, conversions or legal value, applied before any removal.
- A prioritised plan with owners, batches and re-check dates.
- Honest limits: missing data, seasonality and anything the numbers can't show.

## Common mistakes
- Deleting pages on age or word count alone.
- Ignoring backlinks and conversions when pruning.
- Treating every traffic drop as a content problem when it is seasonal or technical.
- Auditing without reading pages.
- Producing a spreadsheet with no owners or dates.

## Questions to ask before you recommend changes
- Which pages must stay for legal, product or partner reasons?
- What conversions matter most, and how are they tracked?
- Who can update, merge and redirect pages, and how fast?
- Has anything technical changed recently that could explain drops?

## Self-check before you answer
- Does every page have an outcome backed by data?
- Did I read a sample of pages, not only the numbers?
- Are pages with links or conversions protected?
- Does every action have an owner, a skill to use and a re-check date?

## Official documentation
- Creating helpful, reliable, people-first content: https://developers.google.com/search/docs/fundamentals/creating-helpful-content
- SEO Starter Guide: https://developers.google.com/search/docs/fundamentals/seo-starter-guide
- Search Console Performance report: https://support.google.com/webmasters/answer/7576553
- Google Search Status Dashboard: https://status.search.google.com/

---
Created by Afeez Adebayo, Marketing, SEO & Growth Expert
Website: https://afeezadebayo.com
LinkedIn: https://www.linkedin.com/in/afeez-adebayo/
X: https://x.com/za_icon
Source: https://afeezadebayo.com/skills-md/marketing/content-seo/content-seo-audit/

Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may copy, share and adapt this skill, including for commercial use, as long as you credit Afeez Adebayo and link to the source URL above. Keep this notice in any copy or adapted version.

Workflow

Step 1. Build the inventory

Join crawl, Search Console, analytics and backlink data on URL. Keep only indexable, canonical pages; list non-indexable pages separately for a technical check. Add page type, topic and the main query each page ranks for.

Step 2. Add trend data

Compare the last 3 months with the same period a year earlier and with the previous 3 months. Flag pages losing more than a set share of clicks as decaying.

Step 3. Score each page

DimensionWhat to checkScore 0 to 2
DemandImpressions for queries the page targets0 none, 1 some, 2 strong
PerformanceClicks, position, CTR against similar pages0 poor, 1 average, 2 strong
Intent matchDoes the format and angle match the current results?0 mismatch, 1 partial, 2 match
Quality and E-E-A-TAccuracy, depth, first-hand evidence, author, freshness0 weak, 1 fair, 2 strong
On-pageTitle, H1, headings, meta description, internal links, images0 weak, 1 fair, 2 strong
Business valueConversions, assisted conversions, link value, strategic fit0 none, 1 some, 2 high

Step 4. Decide the outcome

PatternOutcome
Strong performance and valueKeep; light maintenance
Demand exists, page underperformsImprove (content-refresh, on-page-seo)
Two or more pages split the same queryMerge (keyword-cannibalisation)
No demand, no value, but has links or traffic historyMerge or redirect to the closest page
No demand, no value, no linksRemove with 410, or noindex if users still need it (content-pruning)

Step 5. Read a sample by hand

Numbers miss quality. Read at least the top 20 pages by traffic and 20 random pages from each outcome group. Note patterns: outdated facts, thin sections, missing authors, broken examples.

Step 6. Prioritise

Score each action by expected impact (traffic and business value at stake) and effort (hours to fix). Order quick wins first: high impact, low effort.

Step 7. Plan the work

Group actions into batches by owner and template. Set a re-check date per batch, usually 6 to 12 weeks after release.

What good looks like

  • Every page has one main purpose and one main query group.
  • Titles and H1s match what searchers want.
  • Content shows first-hand experience where it matters: data, screenshots, examples, named authors.
  • Important pages are linked from related content.
  • No two pages compete for the same intent.

Worked example

A SaaS blog had 410 posts. The inventory showed 38 posts brought 81% of organic clicks over 12 months (Search Console), 96 posts had no clicks at all, and 22 pairs of posts targeted the same queries. The plan kept and maintained the top 38, refreshed 54 posts with demand but falling clicks, merged the 22 pairs into 22 stronger posts, redirected 31 old posts with backlinks to their closest replacements, and removed 65 thin posts with no links. Twelve weeks after the first two batches, clicks to refreshed posts were up and the blog had 140 fewer pages to maintain.

Output format

# Content SEO audit: <site or section>, <date>
## Summary (pages audited, share of clicks from top pages, outcome counts)
## Inventory (spreadsheet): URL, type, topic, main query, clicks, trend, conversions, links, scores, outcome, owner
## Patterns found in the manual review
## Prioritised action plan
| # | Action | Pages | Impact | Effort | Skill to use | Owner | Re-check |

Edge cases and judgement calls

SituationWhat to do
Seasonal pages with no clicks out of seasonJudge on the same season last year, not the last three months
New pages under six months oldMark as "too early" unless clearly thin or duplicate; re-audit later
Pages with links but no trafficNever remove; merge or redirect to keep the link value
Pages that convert but get little trafficKeep and improve visibility; they are worth more than traffic suggests
Legal, policy or support pagesKeep regardless of search data; exclude from pruning decisions
Pages ranking for queries unrelated to the businessLow value even with traffic; consider refocusing or removing
Programmatic or templated sectionsAudit by template and sample, then apply decisions to the whole group
Multilingual sitesAudit each language separately; a page weak in one market may be strong in another

Scoring at scale

For sites with thousands of pages, score automatically first: demand and performance from Search Console, value from analytics, links from backlink data. Then read a stratified sample from each score band. Adjust the automatic rules where the reading disagrees, and re-run. Record the rules so the next audit is comparable.

Handoffs to specialist skills

  • Pages to improve: content-refresh for decaying pages, on-page-seo for single-page gaps
  • Overlapping pages: keyword-cannibalisation
  • Pages to remove, merge or redirect: content-pruning
  • Quality and trust gaps: eeat-content-quality
  • Weak titles and descriptions: title-tags-meta-descriptions
  • Missing topics: topical-map and keyword-research
  • Indexing or rendering faults found on the way: the Technical SEO skills, starting with technical-seo-audit

Quality bar: what an excellent audit includes

  • One outcome for every indexable page, backed by data and a reason.
  • A manual reading of samples, with the patterns it found.
  • Protection rules for pages with links, conversions or legal value, applied before any removal.
  • A prioritised plan with owners, batches and re-check dates.
  • Honest limits: missing data, seasonality and anything the numbers can't show.

Common mistakes

  • Deleting pages on age or word count alone.
  • Ignoring backlinks and conversions when pruning.
  • Treating every traffic drop as a content problem when it is seasonal or technical.
  • Auditing without reading pages.
  • Producing a spreadsheet with no owners or dates.

Questions to ask before you recommend changes

  • Which pages must stay for legal, product or partner reasons?
  • What conversions matter most, and how are they tracked?
  • Who can update, merge and redirect pages, and how fast?
  • Has anything technical changed recently that could explain drops?

Self-check before you answer

  • Does every page have an outcome backed by data?
  • Did I read a sample of pages, not only the numbers?
  • Are pages with links or conversions protected?
  • Does every action have an owner, a skill to use and a re-check date?

Official documentation

When to use this skill, and when not to

Use it when organic traffic is falling across many pages, when a blog or resource centre has grown without a plan, before a content strategy reset or migration, or when nobody can say which pages still matter. It is the starting point for the whole Content and On-Page SEO set.

Don’t use it for a single page (use On-Page SEO Optimisation), for crawling and indexing faults (use the Technical SEO Audit) or for choosing new topics (use Keyword Research).

Example prompts

  • “Audit our blog. Search Console, GA4 and Screaming Frog exports attached. What should we keep, update, merge or delete?”
  • “Organic clicks fell 25% this year across our guides. Find where and why.”
  • “We are migrating next quarter. Which pages are worth moving?”
  • “Score our 300 product guides on intent match and quality.”
  • “Which 20 pages would give us the biggest gain if we updated them?”

Example output

The skill returns a fixed structure. A shortened example for an imaginary site:

# Content SEO audit: example.com/blog, 10 October 2026

## Summary
410 posts audited. 38 posts bring 81% of organic clicks (Search Console, 12 months). 96 posts had no clicks.

## Outcomes
| Outcome | Pages | Next skill |
| Keep | 38 | maintenance |
| Improve | 54 | content-refresh |
| Merge | 44 (22 pairs) | keyword-cannibalisation |
| Redirect | 31 | content-pruning |
| Remove | 65 | content-pruning |

## Quick wins
| # | Action | Pages | Impact | Effort |
| 1 | Refresh posts at positions 5-15 with falling clicks | 18 | High | Medium |

Every page gets one outcome backed by data, and pages with links or conversions are protected before anything is merged or removed.

What the agent needs from you

The skill asks for Search Console Performance by page and by page plus query (ideally 16 months), analytics landing page data with conversions, a crawl with titles, H1s, word counts, dates, authors and inlinks, backlinks by URL, the business goals, and any pages that must stay for legal or partner reasons. It reads a sample of pages by hand, because numbers alone miss quality.

How this skill hands off to related skills

The audit decides what to do; specialist skills do it. Pages to improve go to Content Refresh and On-Page SEO Optimisation, overlapping pages to Keyword Cannibalisation, pages without a future to Content Pruning, quality and trust gaps to E-E-A-T and Content Quality, and weak titles to Title Tags and Meta Descriptions. New topics found along the way go to Topical Maps.

How to run the Content SEO Audit step by step

  1. Install the skill as content-seo-audit/SKILL.md, ideally with the other 12 Content SEO skills so handoffs work.
  2. Export Search Console data by page and by page plus query, for 16 months if available.
  3. Export analytics landing page data for organic traffic: sessions, engagement, conversions and revenue for 12 months.
  4. Add a crawl and backlinks: titles, H1s, word counts, dates, authors, canonicals and inlinks, plus links by URL.
  5. Tell the agent your goals and which pages must stay for legal, product or partner reasons.
  6. Review outcomes and the plan, then run each batch with the specialist skill named for it and re-check after six to twelve weeks.

The skill always reads a sample of pages by hand, because numbers miss quality. Expect it to ask about pages it can’t judge from data alone.

Situations this skill handles

Every content library has its own mess. These are the situations the audit handles most often:

SituationWhat the skill does
A blog with hundreds of posts and no planInventories every page, scores it and gives one outcome per page
Traffic falling across many articlesSeparates decay, cannibalisation, intent shifts and technical causes
Preparing for a migrationIdentifies pages not worth moving and how to retire them safely
Seasonal contentJudges pages against the same season last year
Pages with links but no trafficProtects their link value through merges or redirects
Multilingual contentAudits each language separately, because performance differs by market

What makes this skill different

Most content audits are spreadsheets of traffic numbers with a “delete” column. This skill scores every page on demand, performance, intent match, quality and business value, reads samples by hand, and protects pages with links, conversions or legal value before anything is removed.

It also connects the audit to action. Every outcome names the specialist skill that does the work (refresh, merge, prune, rewrite titles), with an owner and a re-check date, so the audit turns into a plan the team can actually run.

Practical tips from real projects

  • Start with your top pages. In most libraries a small share of pages brings most clicks; protect and improve those first.
  • Look for decay early. Pages losing clicks year on year are usually the cheapest wins.
  • Merge before you delete. Two weak pages on one topic often make one strong page.
  • Record your scoring rules so next year’s audit is comparable.
  • Audit at least once a year, with a quarterly check for decaying pages.

A common pattern in the first audit of an established blog: around a tenth of posts bring most of the traffic, a quarter have demand but are slipping, a fifth overlap with other posts, and a large group brings nothing at all. The skill turns that pattern into batches, starting with the refreshes and merges that recover traffic fastest, and leaving pruning for last, once every page with value has been protected.

The audit also gives you a defensible story for leadership. Instead of “we deleted old posts”, you can show how many pages were improved, merged or retired, which ones kept their links through redirects, and how clicks to the improved groups changed after twelve weeks. That makes the next round of content investment much easier to approve.

Frequently asked questions

Usually a few dozen or more. Below that, review pages one by one with the on-page skill. Above a few hundred pages, the audit saves weeks of guesswork.

Only pages with no demand, no links, no conversions and no user need. Everything with value is kept, improved, merged or redirected, so traffic and links aren’t thrown away.

With exports ready, an agent returns a first inventory and outcomes in one session for most sites. Reading samples and agreeing actions with the team usually takes longer than the analysis.

No. Search Console, analytics and a crawler cover most of it. Backlink data from a paid tool helps protect pages with links, and the skill labels any third-party estimates.

A full audit once or twice a year is enough for most sites, with a lighter quarterly check for decaying pages using the refresh skill.

Changelog

  • 1.0.0, 10 October 2026: first release.

Use, adapt and share this skill under CC BY 4.0: credit Afeez Adebayo and link to this page.