Cost to scrape Google Search at 10K, 100K, and 1M

Access fact, checked 2026-07-26: Google’s Terms prohibit automated access that violates machine-readable instructions on its web pages (Source).

Cost to scrape Google Search at 10K, 100K, and 1M table

The comparison uses 10,000, 100,000, and 1,000,000 billable target calls or delivered results per month, calculated on 2026-07-26. It does not claim those calls all become useful rows. Credit, request, result, and search meters are passed through the dataset quote model, which selects the least expensive published self-service offer that covers each volume. A starting USD rate is shown as a lower bound because an unpublished minimum, tier, or contract can raise the invoice. Retries, pagination, duplicate outputs, taxes, and data licensing are outside these figures.

The published price answer

The record contains 7 published target-rate lines across 5 vendors. Those are the only rows allowed to produce a target-specific dollar result. Different light, normal, HTML, rendered, protected, or dedicated modes stay separate because collapsing them would hide the unit change that drives the bill. A plan name in a cell is the offer selected from that vendor record, not a recommendation.

Google SERP requests on all Google subdomains cost 25 credits. docs.scraperapi.com. Requests whose URL is a Google domain cost 20 credits in the HTML API. help.scrapingbee.com. Google API light mode costs 10 credits; normal mode costs 15 credits. help.scrapingbee.com. Google API normal mode costs 15 credits; light mode costs 10 credits. help.scrapingbee.com. Google domains cost 10 credits with residential routing applied by default. scrape.do. Oxylabs publishes a Google Web Scraper API rate starting at $0.001 per result. oxylabs.io. A Google Search engine call consumes one SerpApi search. serpapi.com.

The page is simple; the access path is not

This record classifies Google Search as hard and records rate limiting, CAPTCHA, bot detection and fingerprinting, IP reputation checks, geographic variation or restriction. JavaScript rendering is not required in the recorded profile, so the table does not add a rendering multiplier unless a published target mode already includes one. Residential routing is recorded as needed, but no generic proxy multiplier is silently added: the table applies only a vendor’s explicit target meter.

A search, map, shopping, and detail endpoint can return different records even when a vendor groups them under one Google product.

The official route is a separate product decision

Custom Search JSON API is recorded as available. Closed to new customers; existing customers may use it until January 1, 2027. Existing customers receive 100 queries per day free, then pay $5 per 1,000 queries up to 10,000 queries per day. Custom Search JSON API is closed to new customers and does not establish unrestricted parity with the public Google Search page. Source and access status checked 2026-07-26.

Endpoint evidence and the remaining unknown

The dataset confirms 5 dedicated endpoints or templates. This proves named target support in the cited product documentation. It does not prove that each endpoint has a public target-specific price.

• Google Search Engine Results API is paired with a published target meter in the cost table.
• Google Search target is paired with a published target meter in the cost table.
• Google Search API is paired with a published target meter in the cost table.
• Google Search API is paired with a published target meter in the cost table.
• Google target templates is documented, while its target-specific price remains unpublished in this record.

What volume means for Google Search

One useful unit is a results-page snapshot for one keyword, location, language, device profile, and page position. Scrape.do’s dedicated Search endpoint returns structured organic and other result blocks for that exact request context, so ten keywords across three locations and two pages are 60 search requests before any result-page URLs are fetched (Scrape.do Search endpoint, retrieved 2026-07-27).

The dedicated endpoint paginates with the start offset: start=0, start=10, and start=20 request successive groups. Each offset is another billed search request rather than free continuation inside the first response (Scrape.do Search endpoint, retrieved 2026-07-27).

Google’s Custom Search JSON API returns at most 10 results per response and supports at most 100 results for one query through its start index. It searches a configured Programmable Search Engine and is closed to new customers, so it does not establish parity with an unrestricted public Google Search results page (Custom Search cse.list, retrieved 2026-07-27; Custom Search overview, retrieved 2026-07-27).

Evidence to check before buying

• ScraperAPI prices Google SERP at 25 credits versus 1 for a normal request docs.scraperapi.com.
• ScrapingBee prices Google HTML requests at 20 credits and its structured Google API at 10 or 15 credits help.scrapingbee.com.
• Oxylabs publishes a Google rate starting at $0.001 per result oxylabs.io.
• Scrape.do automatically applies residential routing to Google domains at 10 credits scrape.do.
• Confirm the exact entity and output unit, whether a page, search, result, record, or API object. Then multiply by pagination and refresh frequency before selecting a plan.
• Fix the required country or market before pricing. The recorded geographic variation means one nominal URL can represent different jobs.