The Best Prompt for a Technical SEO Audit

Why this prompt works

How to use it

1. Run a crawl with your usual crawler and export the internal URL report with status code, indexability, canonical, meta robots, title, inlinks and depth columns.
2. If the export is large, filter it to the section or template with the problem, and include a sample of healthy URLs for comparison.
3. Paste the robots.txt file and the sitemap index into the crawl data field along with the export rows.
4. Add Search Console details: the reasons and counts from the page indexing report, and URL Inspection output for two or three affected pages.
5. Open the example URLs for each P1 finding and confirm the issue in the page source or with URL Inspection before assigning work.
6. Use section 4 to decide what to export next, then rerun the prompt with the extra data.

AI model compatibility

“Expected to work” means the prompt uses plain, model-independent structure and has been reviewed for that — not that it has been run on that model. A model is marked Tested only when a test record is published below.

Worked example

Scenario: An illustrative case: a SaaS company whose new blog posts stopped appearing in search after a framework migration.

# Role
You are a technical SEO consultant auditing a site from exported data. You have not crawled the site and cannot fetch URLs in this conversation; everything you conclude must come from what I paste below.

# Site
- Site, platform and size: B2B SaaS marketing site rebuilt on Next.js, about 1,200 URLs
- Why this audit is happening: Blog posts published since the migration are not being indexed
- Templates or URLs that matter most: /blog/[slug] template, /blog/ listing pages
- Constraints on fixes: One front-end developer; fixes must ship within the current sprint

<crawl_data>
Crawler export filtered to /blog/ with columns URL, status, indexability, canonical, meta robots, inlinks, depth; robots.txt contents; sitemap index
</crawl_data>

<search_console_notes>
Page indexing report shows a growing count under 'Alternate page with proper canonical tag' for blog URLs; URL Inspection for two posts pasted
</search_console_notes>

# Method
First, state in two or three lines what the data contains (columns, row count, URL types) and what it lacks. Then check these areas in order, skipping any the data cannot speak to:
1. Crawl access: robots.txt rules, status codes, server errors, redirect chains and loops.
2. Indexing signals: noindex directives, canonical tags, and conflicts between them, redirects, sitemaps and internal links.
3. Duplication: parameter, trailing-slash, protocol and host variants; near-identical titles across indexable URLs.
4. Discovery: sitemap coverage, click depth, pages with few or no internal links.
5. Rendering: content or links that appear to depend on JavaScript.
6. Mobile and page experience: use Core Web Vitals only if I supplied field data.
7. Structured data and international tags, if present in the data.

# Rules
- Every finding cites evidence: the column values and up to three example URLs from my data. No evidence, no finding.
- Mark each finding CONFIRMED (the data shows it) or SUSPECTED (the pattern suggests it; say what would confirm it).
- Group by cause, not by URL. One template bug affecting many URLs is one finding with a count.
- Do not repeat tool warnings as problems unless you can explain the effect on crawling, indexing or users.
- Keep these distinct: a robots.txt disallow stops crawling, not indexing, and a noindex only works if the page can be crawled.
- Do not invent URLs, counts, scores or metrics, and do not predict ranking or traffic gains.

# Output format
1. Data summary
2. Findings table: Issue | Status | Evidence | URLs affected | Why it matters | Fix | How to verify | Priority (P1 to P3)
3. Fix order: the first five actions, respecting my constraints
4. Not checkable from this data: what to export or test next

What a good response should contain

• A short data summary noting which columns are present and that no Core Web Vitals field data was supplied
• Findings grouped by template, each quoting canonical or meta robots values from the export with example URLs and a count
• A clear CONFIRMED or SUSPECTED status on each row, with the test that would confirm suspected ones
• A fix order limited to what one developer can do in a sprint
• A list of unchecked areas, such as rendering and server logs, with the export needed for each

This is an illustrative scenario showing how to fill in the prompt and what to look for. It is not a recorded model output. Recorded outputs will be published here once this prompt has a test record.

Testing and editorial review

No test record has been published for this prompt yet. When it is tested, this section will show the model and version, the date, the inputs, the actual output and notes against our seven-point rubric. We do not publish numerical scores.

Limitations and things to check

• The audit is only as complete as the export. Issues outside the pasted data will not appear.
• The model cannot fetch pages, render JavaScript or read server logs, so rendering and crawl-rate findings stay suspected until you test them.
• Counts are read from pasted text and can be miscounted on large exports. Confirm totals in your crawler or spreadsheet.
• Fixing technical issues removes obstacles; it does not guarantee indexing or better rankings.
• Platform-specific fix instructions may not match your CMS or framework version. Check them against its documentation.

Alternative prompts

Related resources

All SEO prompts →

Frequently asked questions

Sources and references

1. Google Search Essentials — Google Search Central
2. SEO Starter Guide — Google Search Central
3. Web Vitals — web.dev

Was this prompt useful?