SEO audit
Find the blockers before you rewrite the copy.
A Siteory SEO audit is a crawl of the submitted host plus a Markdown report. It tells you what is hiding the page, what is duplicated, and what a developer should change first.
What a website SEO audit checks
The first question is whether a crawler can fetch and keep the URL. Siteory records status codes, robots.txt, noindex, and canonical tags before it spends time on title length. If those signals conflict, the report says so in the first section.
The on-page pass then looks at the HTML a bot actually downloaded:
- Unique title tags in the 30–60 character range, with the primary phrase near the front
- Meta descriptions that are 120–160 characters and not copied from the title
- A single H1 and an H2 outline that matches the page
- Internal links so public URLs are not orphans
- Image alt text that describes the image, or is empty when the image is decorative
- JSON-LD that parses, uses absolute URLs, and avoids retired types such as HowTo
That list is the same order as the website SEO audit checklist on the blog. The product is the live evidence. The article is the decoder ring.
Who this audit is for
Founders who launched a Next.js site last week. Agencies that inherited a theme with five H1s. Engineers who were handed a 40-page PDF and asked to “do SEO.” The report is written for the last group on purpose.
If you also need AI Overviews and ChatGPT visibility, run the AI search audit on the same URL. If the ticket is headers, use the security audit. One crawl feeds all three.
Findings that usually come first
New Next.js and Webflow launches fail in the same three places: a leftover noindex from staging, a homepage title that is only the brand name, and a sitemap that still lists the preview host. Those are one-file fixes. A 200-item “low contrast” list is not.
Siteory keeps crawl blockers above copy nits for that reason. Read crawlability, robots.txt, and sitemaps if the site is brand new, and JSON-LD types that still work if the only ticket is structured data.
What you can do with the Markdown
Paste the report into Cursor, Claude, or Codex and ask for a patch. Review the diff. Siteory does not change the website. The model does not change the website until you apply the patch. That boundary is the product.
See how we check for the live engine list, or how it works for the path from URL to pull request. Start a scan from pricing.