What it is.
You open Configuration, then Custom, then Custom Extraction, and add an extractor. You can build it visually by clicking the element on a rendered page, or write the XPath by hand if you know what you want.
Then you crawl, and the Custom Extraction tab holds one column per extractor for every URL. That is how a crawl becomes a dataset: authors, publish dates, prices, schema blocks, Open Graph tags, hreflang.
The published XPath examples cover headings, hreflang, structured data, and social meta tags, which is most of what an audit needs.
What you get.
- One column per extractor across every crawled URL.
- Visual selection, so you can point at an element instead of writing XPath.
- Published XPath examples for headings, hreflang, structured data, and social tags.
- An export of the extracted data alongside the rest of the crawl.
- The same capability against a competitor's public pages.
How to set it up.
Open Configuration, then Custom, then Custom Extraction, and add an extractor.
Use visual extraction first and click the element you want on a rendered page.
Switch to manual XPath when the visual pick grabs too much or too little.
Name each extractor for what it holds, because eight columns called Extractor 1 are useless later.
Test on a handful of URLs before running the full crawl, since a wrong expression wastes the whole run.
Crawl, then read the Custom Extraction tab and export.
Use cases
Pull one field from every page
Set the extractor and get a column across the whole crawl.
Point instead of writing a selector
Use visual selection when nobody wants to write XPath.
Watch a competitor's pricing page
Extract the values you care about and re-crawl later to compare.
Best for
Content audits at scale
Pulling author, date, and word count across a whole blog is a crawl, not a spreadsheet exercise.
Checking schema across a site
Extracting the structured data block per page finds the templates where it is missing.
Competitor research
Anything on a competitor's public pages can be extracted and counted.