Collect organic results
Use Light mode in the MCP search tool when you need a list of ranking pages. Keep the original query and result status with the rows.
Google SERP API
Send a query and collect structured Google Search results for SEO research, competitor monitoring, or your application. Keep positions, page URLs, titles, and snippets together so your team can trace each result.
- Complete search · MCP Search · Light mode · 2026-11-08 UTC. Query
roof repair denver. 1 of 1 pages delivered, with 10 organic results on page one. The search location was not verified. Link to this complete search. - Partial snapshot · SERP Intelligence · REST · 2026-11-07 UTC. Query
public libraries seattle. 1 of 2 pages delivered, with 4 organic results on page one. Page two was unavailable, and the requested Seattle location was not verified. Link to this partial snapshot.
Different queries and capture modes. Use these examples to check coverage.
MCP Search · Light mode · 2026-11-08 UTC
roof repair denver
Complete · 1 of 1 pages delivered · 10 organic results
- Recomendations for roofing company in Denver - ASAP - Reddit
- Cenco Roofing: Denver Roofer for Roof Repair & Replacement
- Denver Roofing Services | Premier-Roofing.com
First three rows shown. Denver was part of the query. The response did not report a verified search location.
Partial Google SERP snapshot and coverage
This REST snapshot returned 1 of 2 requested pages and 4 organic rows. Its diagnostic reported failureCode: invalid_next. Keep those rows if your report can use page-one results. If it requires both pages, flag the capture as incomplete.
See the REST snapshot request and its coverage fields.
Test a Google SERP API importer with a partial result
Use the recorded partial JSON to test how your importer handles missing pages. This REST capture returned one of two requested pages and four organic rows. The fixture includes three of those rows and marks the fourth as omitted. Each included row has a position, title, and URL. The fixture omits snippets. The capture reported invalid_next.
Import the fixture and verify that your report marks the result as partial, the second page as missing, the fourth row as omitted, and the requested Seattle location as unverified.
Download the recorded partial fixture (JSON)
SERP Intelligence · REST · 2026-11-07 UTC
public libraries seattle
Partial · 1 of 2 pages delivered · 4 organic results
First three rows shown. Page two was unavailable. Seattle was requested, but location verification was unavailable in this capture.
Choose the search output your workflow needs
Start with a query your team already tracks. Check the result count, completion status, and fields before you use the output in a report.
Comparing providers? Compare SerpApi and DataForSEO for Google organic results. If you already use SerpApi, compare alternatives by what you need to change. If you run a DataForSEO workflow, check what a replacement must preserve.
Researching follow-up questions for a brief? Compare People Also Ask tools by their output.
Check additional search features
An unfiltered one-page Full search can include available search features. If you need a local pack, check that the response actually returned it.
Request a search snapshot
Use the SERP Intelligence API when your application needs a snapshot with request context and diagnostics. Check how many pages arrived before using the data.
Compare the partial example above to see how a two-page request can return usable rows while still missing requested coverage.
Collect related questions with the People Also Ask API
Use a People Also Ask harvest when you need to expand related questions for one Google Search query. Start a REST harvest with POST /harvest, then read GET /jobs/{id} using the returned job ID.
Keep the query and job ID with each question. Check which rows have an answer and a source link before using them in a brief. A completed job can return fewer questions than the limit you requested.
When you request two result pages, the extra page adds organic listings when available. Question expansion starts from the original first page.
In a November 7 MCP capture for “mcp server for web scraping,” 30 of 30 requested questions returned. The row “Which MCP server should I use?” included answer text and six source entries across five URLs. The row omitted the expansion flag, so full expansion remains unverified.
Inspect the recorded question, answer, and sources · Read the request options in Docs
From search results to sources you can cite
Use the returned URLs to shortlist sources. When your report needs facts from a ranking page, extract that URL and check the relevant passage. Use the page extraction demo to compare HTML, JSON, and Markdown. Follow the Google Search research guide for a repeatable capture and export workflow.
If you need those pages in another AI session, save the scraped research in Memory with the original source links.
Follow one roofing result to its source
In the recorded Denver search, Cenco Roofing appeared at position 2. A separate page extraction returned the site’s own description of its services. For the next source, decide when the search result is enough and when to extract the page.
Roof repair, roof replacement, storm damage restoration and gutters across Colorado's Front Range
Open the source page. This is the company’s description from the captured page.
How much does a search cost?
Search uses Credits. The charge depends on the capture and delivery. See the SERP API pricing example before scaling a workflow.
Can I run a search from my AI client?
Connect a compatible MCP client to use the Google Search tool. Use the API when you want the results in your own application.
Can I collect reviews for one business?
Use the Google Maps Scraper when you need business listings or an available review sample for a specific place.
Check what your query returned before you use it
Run a query from a real project and check that the response contains the fields your workflow needs.