01 / THE USE CASE
Read public Google results without operating a browser
Retrieval, monitoring, and research products often need the organic SERP as it appeared for a query and locale—not an ads mix, and not a screenshot. Clair fetches public google.com/search HTML, strips chrome, and returns position, title, url, displayed_url, and snippet.
This is a live scrape billed per page. It is not the News index and not a cache you can replay without spending credits again.
02 / HOW TO CALL IT
Query, locale, then page
Call GET /v1/search?engine=google_search with a required q. language (hl) and country (gl) select the Google edition. page is 1-based (default 1, max 10). Do not send limit or cursor. Parameters and examples are on the Google Search API page.
Pass page=2 for the next 10 hits. That call is billed. Pagination stops at page 10 because that is where Google remains reliable.
03 / BOUNDARIES
Public HTML, organic hits, documented caps
Clair does not log into Google, click through consent walls on your behalf in a way that would impersonate an end user session you supply, or return ads. Pagination stops at page 10 (start=90) because that is where Google remains reliable.
For editorial news as a prepared corpus, use the news monitoring use case. One Clair key covers both.
04 / FAQ
Questions from retrieval teams
Does Clair return ads or autocomplete?
No. The Google Search API returns organic hits only. Ads, knowledge panels, and Google chrome are stripped before the payload.
How many results per request?
Google's reliable page size is 10. Each page costs 5 credits. Pass page=2, 3, … up to 10 for the next SERP.
Can I search a specific site?
Yes. q supports operators such as site:. language and country map to Google's hl and gl.