Google News Scraper: headlines with real publisher URLs, by keyword, topic, country and date range

Scrape Google News by search query or topic and get one clean row per article: headline, publisher name, publisher site, publication time, the Google News link and the decoded real publisher URL (reuters.com, nytimes.com... instead of the news.google.com redirect). Pick the country and language edition, set a start and end date, and the actor splits the range into day and week windows automatically to get past Google's limit of about 100 results per search. Export to JSON, CSV or Excel, or call it from the API.

Use it for news monitoring and media tracking, brand and competitor alerts, PR and reputation reports, market and financial research, and datasets for LLM, RAG and sentiment pipelines. Works with AI agents through the Apify MCP server. It reads Google News' public RSS feeds; no login, no cookies and no proxy are needed.

Why this one

• Real publisher URLs. Each Google News redirect is decoded to the article's own URL (about 98% success in our tests; the rest return url: null with a decodeStatus). Switch decoding off for a faster run.
• Gets past the 100-result cap. With a start date, ranges are split into 7-day windows and any window that comes back full is halved down to single days, newest first.
• Any country and language edition. US/en, ES/es, DE/de, GB/en, BR/pt and so on.
• Queries and topic feeds. Several queries per run, Google search operators ("exact phrase", OR, -word, intitle:, site:) and the eight topic sections (World, Business, Technology...).
• Duplicates removed. The same story found by two queries or two windows is returned once.
• Cheap and predictable. $0.002 per article; Max articles per query caps every run.
• Never fails the whole run for one input. An empty query, bad country, topic or date gets its own row with an error message and is not charged; the rest continue. A RUN_SUMMARY record lists articles, windows, decode results and HTTP statuses.
• Polite by design. Paced requests, retries with backoff, honours Retry-After and stops politely if Google keeps answering 429.

Input

Field	Meaning
Search queries	One or more queries, for example OpenAI, "electric cars" Germany, site:reuters.com chips
Topic feed	Optional: WORLD, NATION, BUSINESS, TECHNOLOGY, ENTERTAINMENT, SPORTS, SCIENCE, HEALTH (about 70 current items)
Country / Language	Edition codes, default US / en
Start date / End date	Optional YYYY-MM-DD range; enables automatic window splitting. End defaults to today
Decode publisher URLs	Default on
Max articles per query	Default 100, max 5,000 (above 100 needs a start date)
Remove duplicates	Default on

Output

Article row: type: "article", title, source, sourceUrl, publishedAt, googleNewsUrl, url, decodeStatus, snippet, query, country, language, window, scrapedAt.

Error row: type: "error", query, error, scrapedAt. See sample-output.json for real rows.

Notes on the data:

• url is the publisher's article URL, or null when decoding is off or failed (decodeStatus says why: ok, skipped, no-url, no-params, http-NNN).
• snippet is present only when Google's feed carries a description different from the headline (mostly topic feeds).
• window shows which date window found the article (for example 2026-10-22..2026-10-22).
• Google News returns about 100 items per request. A single day that still returns 100 hits for a very busy query may hold more stories than Google exposes; the RUN_SUMMARY marks these as cappedWindows.
• Without dates you get the most relevant recent items for the query, up to about 100.

Pricing

Pay per event: $0.002 per article ($2.00 per 1,000 articles) plus a $0.00005 start event per run. URL decoding is included. Error rows are not charged. A run returning 100 articles costs about 20 cents. See PRICING.md for the benchmark.

Public data only

The actor reads only public Google News RSS feeds and resolves Google's own redirect links to the publishers' URLs. It returns headlines, sources, dates and links only; it does not open or scrape the publishers' article pages or bodies, does not log in and does not bypass anything. Headlines and articles belong to their publishers: you are responsible for using the data in line with Google's terms, the publishers' terms and copyright law.

Notes and limits

• Not an official Google product or API; the feed format or the decoding call can change without notice.
• Google may rate-limit heavy use from one network. The actor slows down, retries and stops politely instead of hammering.
• Results are what Google News shows for the chosen edition; ranking differs by country and language.

FAQ

How much does it cost? $0.002 per article plus $0.00005 per run. 1,000 articles = $2.00. Error rows are free.

Do I need an API key, account or proxy? No.

How do I get more than 100 articles for one query? Set a start date (and optionally an end date) and a higher Max articles per query; the actor splits the range into windows.

Can I get the article text? No. This actor returns headlines and links only.

Which export formats are available? JSON, CSV, Excel, XML, HTML or RSS from the Console or API, or integrations such as Make, Zapier, n8n and Google Sheets.

Something broke or a field is missing? Open an issue on the Actor's Issues tab with the run link.

Use with AI agents / MCP

Agents can call this Actor through the Apify MCP server. Let the agent find it with the search-actors tool ("google news scraper") and call it by name:

• Actor: santis_hermes/google-news-scraper
• Example prompt: "Get the latest 50 Google News headlines about OpenAI with the real publisher links."
• Minimal input:

{
  "queries": ["OpenAI"],
  "country": "US",
  "language": "en",
  "maxItemsPerQuery": 50,
  "decodeUrls": true
}