Coming soon: Google Search is still in development. An Explorer License does not enable it in the released extension. The following is a development preview, not instructions for the current release.
Google Search is a pre-built workflow for desktop www.google.com Search with English headings. Start a new search from the worksheet or collect the active supported results page, then inspect retained output in the Dashboard.
Development preview
Open Tab Automation, choose Google Search in Library, and configure Search:
- New search is the default for a new workflow. Enter search terms or paste a supported Google results link. Run opens it in a new tab. Terms use English/US search defaults; a pasted link keeps its parameters.
- Current search uses the active supported Search results page and previews its query and URL. If the current page is unsupported, switch back to New search.
Save can keep an empty reusable setup. Run requires a valid search source and uses your worksheet draft. Restart repeats the saved source: a fresh tab for New search, or the bound tab for Current search. Resume continues a paused execution. Older configurations without a mode retain current-tab behavior.
When released, configuring, saving, and running this pre-built workflow will require Explorer access. It is currently unavailable even with a license. Custom automations are free. Explorer includes 30 days of grace after expiry, and an already-running workflow finishes its run. Saved results remain exportable without a license. How Explorer works.
The supported host is www.google.com with English headings. Current search requires a supported results page; other Google TLDs, the Google homepage, and Images are not accepted as current-search sources. Request a regional variant from Contact.
What it collects
Turn each block on or off:
- AI Overview — Google-generated text and citations, including newer top-level “AI Mode reply for …” blocks.
- Organic Search Results — primary blue links, ranked separately on each search page. Read more, sitelinks, ads, and navigation do not count. Collection defaults to all detected results; Limit results sets a per-page limit from 1 to 20.
- Videos — labeled and ranked separately from organic links. Video collection defaults on.
- People Also Ask Questions — questions. Open each question expands answers and sources. Collect more questions uses Show more (up to 8 extras).
- People Also Search For — related queries, collected only. Visiting those URLs would open another Google search.
- Nearby Places
Continue to the next page paginates, up to 5 pages. Delay between pages is 1,000–30,000 ms (default 2,000). A run stores at most 120 items.
Capture source pages
The type-specific Scrape options open every enabled collected source. A destination appearing in several results is scraped once, and its content is linked to each result. Each scrape stores the available title, resolved URL, content type, capture time, status, and on-page SEO metadata. SEO capture is automatic during enabled source visits, even when readable-text and raw-HTML saving are both off. Save readable page text saves the full readable content available in the loaded page, preserving headings, paragraphs, lists, and tables. Save raw page HTML retains up to 80,000 characters of unsanitized markup per source, with truncation labeled.
Organic scraping defaults on; other source scraping switches, including Scrape video pages, default off. Video-page scraping captures available page text and metadata, not video transcripts. An obvious PDF link is recorded as unsupported and the run continues.
A corroborated captcha pauses the run. It does not solve the captcha. Solve it in the tab, then Resume.
History and export
Results, history, and full source content stay on this computer. Open Dashboard from the side-panel header, then filter Saved results by Search Dataset and open a result. The worksheet’s View search results opens its retained dataset directly, including partial or failed runs. Results shows each organic or video rank with its search page, plus capture statuses and failures. Older runs without stored rankings show that the rank is unavailable. Collection limits and omitted results are labeled.
The selected dataset has JSON, CSV, MARKDOWN, and ZIP export buttons. Finish or pause a running run before exporting. Retained historical datasets are also exportable. Each format includes captured source SEO metadata alongside the query, run status, timestamps, settings, result types, rankings, source URLs, and capture status.
Markdown is a research document with a ranked result index, attributed AI Overview and PAA observations, and a source corpus containing each captured page once. Source citations connect results to full readable text. Missing or truncated content is explicit. It does not call a model or invent conclusions. Standalone Markdown omits raw HTML; JSON and CSV include it when captured. Complete ZIP includes referenced HTML files alongside results.json, results.csv, and results.md.
Rerunning or removing an automation preserves retained datasets and source content in Results. Explicit run deletion removes that run’s retained content. Uninstalling the extension or clearing its storage also removes local Results; downloaded copies remain independent.
For a practical research example, follow SERP analysis with the Google Search workflow to turn collected pages and questions into a content brief.
Review on-page SEO
Open Dashboard → Pages & SEO to review title tags, meta descriptions, meta keywords, headings, canonical links, robots declarations, language, social tags, and declared structured-data types. Pages appear first. Use View → By run or By domain to group captures; opening either group shows its matching pages directly. Each capture keeps its run and time; open a page for full metadata, retained HTML, and its originating run.
Choose Export metadata…, then a format and a scope: selected captures or all matching captures. JSON, CSV, and Markdown are available. Select runs, domains, or individual page captures to export their matching metadata. Filtered exports include all matching captures across pagination. Pause a running run before exporting. Reports preserve capture status and inspection limits; relevant ZIP exports also contain seo.json, seo.csv, and seo.md.
Google’s displayed titles and snippets remain separate from destination title tags and meta descriptions. Sources that were not visited have no on-page metadata; disabled, pending, failed and unsupported captures remain visible. Repeated source references within a run share one inventory record; captures in different runs remain separate. Metadata has a 256 KiB budget with omissions labeled. JSON-LD types are captured declarations, not validation results.
Limits and related controls
This workflow runs separately from the graph steps. It does not click consent buttons, follow related-search queries, or sync results to the cloud. Use the SERP analysis guide to turn the captured evidence into a content brief. Collection does not infer search demand, keyword difficulty, or ranking potential.