Claude Cowork for On-Page Optimisation: Auditing and Improving Existing Content

On-page SEO is foundational. But auditing your actual pages—comparing them against ranking competitors, checking technical elements like meta descriptions and heading structure, identifying keyword opportunities within existing content—is tedious and time-consuming. Screaming Frog crawls can generate thousands of rows of data that take hours to manually review and synthesize into actionable recommendations.

Claude Cowork transforms this workflow. Upload your Screaming Frog crawl data, and Cowork analyzes each page against on-page best practices, identifies specific optimization opportunities, and generates an actionable prioritized list organized by estimated impact. This article covers the complete process: preparing your crawl data, configuring the audit skill, understanding what signals Cowork checks, and how to decide which fixes to implement first.

Teams running these audits report finding 50-100 high-impact optimization opportunities per 1,000 pages crawled—opportunities that would have remained hidden in spreadsheet rows. For context on how on-page optimization fits into your broader Cowork SEO strategy, see Claude Cowork for SEO Specialists: The Complete Playbook.

The Problem: Manual Crawl Analysis Doesn't Scale

Most teams run Screaming Frog crawls regularly. But the insight extraction is manual:

• Download the CSV
• Filter for status codes 200-299 (live pages)
• Scroll through thousands of rows looking for patterns
• Manually spot errors: duplicate meta descriptions, missing H1 tags, long title tags, thin content
• Create a spreadsheet of findings
• Estimate impact ("high", "medium", "low") based on intuition
• Hand off to developers with vague recommendations

This process typically takes 4-6 hours per crawl and often misses issues because human attention is limited. A 10,000-page crawl has 10,000 rows. You're bound to miss patterns unless you have a systematic process.

Cowork systematizes the analysis. It checks every page against a comprehensive set of on-page signals simultaneously, identifies patterns across your entire site, and prioritizes recommendations by estimated ranking and traffic impact.

What Signals Does Cowork Check?

When you upload your Screaming Frog data, Cowork automatically analyzes:

Title Tags

• Length (optimal 50-60 characters; too short misses opportunities, too long gets truncated in SERPs)
• Keyword inclusion (primary keyword in title increases relevance signal)
• Duplicates (multiple pages sharing same title is usually a mistake; signals weak content differentiation)
• Specificity (generic titles like "Products" or "Home" provide no ranking signal)

Meta Descriptions

• Length (optimal 155-160 characters; too short wastes SERP real estate, too long gets truncated)
• Keyword inclusion (secondary keywords in meta increase relevance)
• Duplicates (same meta description on multiple pages suggests template issue)
• CTA language (meta with action-oriented language drives higher CTR)

Heading Structure

• H1 presence (every page should have exactly one H1; multiple H1s signal weak structure, missing H1 wastes opportunity)
• H1 keyword alignment (H1 should match or closely align with page target keyword)
• H2/H3 hierarchy (logical structure of subheadings aids crawlability and user understanding)
• Keyword distribution across headings (keywords in headings carry higher weight than body text)

Content Depth

• Word count (pages with 300+ words typically rank better than thin pages; Cowork flags pages under 300 words)
• List presence (bulleted/numbered lists improve readability and increase featured snippet chances)
• Internal linking (pages with 3+ internal links are more crawlable; pages with 0 internal links are often orphaned)

Technical Elements

• Alt text on images (missing alt text on images is both accessibility and SEO issue)
• Canonical tags (self-referential canonicals are correct; canonical chains or pointing to wrong URLs cause indexing issues)
• Meta robots (pages with meta robots noindex are intentionally deindexed; Cowork flags if this seems unintentional)

The 3-Step On-Page Audit Loop

This is the named workflow teams use to execute comprehensive on-page audits:

Step 1: Crawl Preparation (10 minutes)

Run a complete crawl of your site in Screaming Frog:

1. Set crawl scope to your root domain
2. Set crawl depth to unlimited (or high enough to get all pages)
3. Enable "Follow Internal Links" option
4. Run the crawl (may take 15 minutes to 2 hours depending on site size)
5. Export as CSV: Internal → All
6. Filter the CSV to status code 2xx (live pages only; exclude 3xx redirects and 4xx/5xx errors)

You now have a structured dataset of every live page on your site with technical metadata.

Step 2: Audit Analysis (15 minutes)

Create a Cowork skill called "On-Page Audit" and upload your crawl CSV. Configure with this prompt template:

Execute the skill. Cowork returns a prioritized list of optimization opportunities across your entire site.

Step 3: Prioritization and Implementation (varies)

Review Cowork's output and categorize by effort vs. impact:

• Quick wins: Low effort, high impact (fix 10-20 duplicate meta descriptions, add missing H1 tags). Do these immediately.
• Medium effort: 1-2 hours to fix (rewrite thin content pages, consolidate keyword-competing pages). Assign to your development team.
• Major rewrites: Full content updates (pages ranking 20+). Batch these into your content calendar.

Most sites have 50-100 quick wins per 1,000 pages. Fixing just the quick wins typically yields 5-15% organic traffic improvement within 30 days.

Example Audit Output

Here's what actual Cowork audit output looks like (example for an e-commerce site):

How to Prioritize Fixes by Impact

Not all issues carry equal weight. Prioritize by this framework:

Highest Impact (Do First)

• Fixing issues on high-traffic pages (pages generating 100+ organic sessions/month)
• Duplicate meta descriptions and title tags (low effort, affects multiple pages)
• Missing H1 tags on pages targeting keywords with search volume (easy fix, direct ranking signal)
• Broken internal links (3-4 hours to audit and fix, improves crawlability site-wide)

Medium Impact (Next Priority)

• Content expansion on pages ranking positions 11-20 (small ranking improvements compound to significant traffic gains)
• Adding internal links to high-value pillar pages (consolidates authority, improves crawlability)
• Improving meta descriptions on pages in top 20 (impacts CTR, drives more clicks from existing impressions)

Lower Impact (Nice-to-Have)

• Refining H2/H3 hierarchy on pages already ranking in top 5 (diminishing returns; these pages already have strong signals)
• Adding image alt text to non-featured images (accessibility benefit, minimal SEO impact)
• Consolidating tail pages with very low traffic (cleanup benefit, minimal direct impact)

Connecting to Competitor Analysis

On-page audits identify what you're NOT doing. But they don't tell you what your competitors ARE doing well. For that second layer of analysis:

1. Export the top 3 competitors' pages from Cowork results
2. Create a comparison template showing: Your page | Your title | Your meta | Competitor title | Competitor meta
3. Use a second Cowork skill to analyze the delta: "What's the competitor doing differently in their title/meta/content?"
4. Identify winning patterns and implement improvements based on competitor wins

This two-layer approach (audit what you have + analyze what competitors do better) generates comprehensive optimization roadmaps.

Scaling Audits to Multi-Site Operations

If you manage multiple sites or agency clients:

• Run individual Cowork audits for each site
• Generate a consolidated priority report: aggregate critical/high issues across all sites, identify patterns (e.g., "across all 5 client sites, duplicate meta descriptions affect 120 pages")
• Batch fixes by type: "This month we're fixing all duplicate metas across all clients" + "Next month we're addressing thin content"
• Track improvements: re-run audits quarterly to measure fix completion rate and organic traffic impact

For agencies managing 10+ clients, this audit standardization is significant—it ensures consistent quality across your portfolio and reduces manual analysis time by 80%.

Related Articles: 8 Claude Cowork Workflows • Content Briefs • Connecting to SEMrush, Ahrefs & GA4 • Scaling with Cowork at Agencies