Google News Feed: search, topics and locations with real article URLs

Pull Google News results for any search query, topic, location or language edition, with the option to decode every Google redirect link to the publisher's real article URL. Schedule it in new-articles mode and each run returns only the articles you have not seen yet.

• Real publisher URLs. Google News links point to news.google.com. This Actor turns each one into the article's own URL (reuters.com/..., nytimes.com/...).
• Searches, topics, locations, top stories. Any Google News search operator works: site:, intitle:, "exact phrase", -exclude, OR.
• Every edition. Run the same queries in US, UK, India, Germany, France, Mexico, Brazil, Japan and dozens of other country and language editions.
• New articles only, per query. No diffing on your side. Each feed remembers what it already returned.
• Past the 100-article cap. Split a date range by day to collect up to about 100 articles per query per day.
• Cheap. $1.00 per 1,000 articles, URL decoding included.

Who uses it

• PR and comms teams tracking brand, competitor and executive mentions
• Investors and analysts following tickers, sectors and deals
• Newsletter and media teams sourcing stories by topic and region
• Researchers building news datasets by query, country and date
• AI agents and RAG pipelines that need fresh headlines with fetchable URLs

Quick start

1. Enter Search queries, Topics or Locations (one per line). Leave Editions on US:en or add others.
2. Click Start. Articles appear in the Articles table and export as JSON, CSV or Excel.

Monitor news on a schedule (recommended)

1. Set Mode to New articles since last run.
2. Create a schedule (hourly or daily) under Schedules in Apify Console.
3. Send results to Slack, Google Sheets, a webhook, Make, Zapier or n8n with Apify integrations.

The first run returns everything currently in each feed as new and saves a baseline. After that each run returns only articles that were not in that feed before.

Input

Field	What it does
Search queries	Google News searches. Operators work, for example tesla recall, site:reuters.com inflation, intitle:"rate cut"
Topics	WORLD, NATION, BUSINESS, TECHNOLOGY, ENTERTAINMENT, SPORTS, SCIENCE, HEALTH, a CAAq... topic ID, or a news.google.com/topics/... URL
Locations	Local headlines for a city, region or country, for example Chicago or Bavaria
Include top stories	Adds the edition's front page
Editions	COUNTRY:language, for example US:en, GB:en, IN:en, DE:de, FR:fr, MX:es-419, BR:pt-419, JP:ja. Each feed runs once per edition
Mode	Full snapshot, or new articles since last run
Decode publisher URLs	Off by default, so runs read only the RSS feeds and finish in seconds. Turn on to fill url with the publisher's article link
Published within hours	Keeps recent articles. Search queries also get Google's when: operator
Search from / to date	YYYY-MM-DD. Adds after: and before: to search queries
Split search date range by day	One feed per day in the range, up to 92 days, to get past the 100-article cap
Exclude publishers	Drops articles whose publisher name or domain contains any of these
Max articles per feed	Snapshot mode only. Keeps Google's order
Remove duplicates across feeds	On by default. An article that shows up in several feeds is returned once

Example:

{
    "queries": ["openai", "site:reuters.com electric vehicles"],
    "topics": ["BUSINESS"],
    "locations": ["London"],
    "editions": ["US:en", "GB:en"],
    "mode": "new",
    "postedWithinHours": 24,
    "decodeUrls": true
}

Output

One row per article (this one with decoding on):

{
    "article_id": "CBMimAFBVV95cUxN...",
    "title": "A Once-in-a-Generation Opportunity: 1 Artificial Intelligence (AI) Stock to Buy ...",
    "publisher": "The Motley Fool",
    "publisher_domain": "fool.com",
    "publisher_url": "https://www.fool.com",
    "url": "https://www.fool.com/investing/2026/10/21/1-artificial-intelligence-ai-stock-to-buy-before-i/",
    "google_news_url": "https://news.google.com/rss/articles/CBMimAFBVV95cUxN...?oc=5",
    "published_at": "2026-10-21T19:15:00Z",
    "related_articles": [],
    "rank": 1,
    "feed_type": "search",
    "feed_value": "artificial intelligence",
    "edition": "US:en",
    "language": "en",
    "country": "US",
    "feed_url": "https://news.google.com/rss/search?q=artificial%20intelligence&hl=en-US&gl=US&ceid=US:en",
    "change_type": null,
    "fetched_at": "2026-10-22T00:59:39Z"
}

• related_articles lists other coverage of the same story when Google groups it (top stories and topic feeds), each with title, publisher and Google link.
• change_type is new in new-articles mode and null in snapshot mode.
• url is null when decoding is off (the default) or Google did not return a URL for that article. The run SUMMARY in the key-value store counts decoded and failed URLs, lists every feed read, and records any errors.

Pricing

You pay per event:

Event	Price
Article returned	$0.001 ($1.00 per 1,000), URL decoding included. With decoding on, an article whose URL could not be decoded is delivered free
Feed checked	$0.0005 per feed read (each query x edition, or each day when split by day)

A daily new-articles run over 10 queries that finds 150 new articles costs about $0.155. Set a maximum cost per run in Console and the Actor stops cleanly at that limit, without saving state for unfinished feeds so the next run picks them up.

Limits and good to know

• About 100 articles per feed. That is Google's cap per RSS feed. Use date splitting for more.
• Order is Google's. Search feeds are ordered by relevance, not strictly by date. Sort by published_at after export if you need time order.
• Requests are rate limited to stay polite to Google. Decoding runs at about 3 articles per second. If Google pushes back, the Actor pauses decoding (up to 7 minutes in total) and carries on more slowly. Articles still left without a URL are not charged and are counted in SUMMARY. For big decode runs, turning on Apify Proxy can help.
• Headlines and links only. Google News feeds carry title, publisher, time and link, not article text. Use a page scraper on url if you need full text, and respect each publisher's terms.
• This Actor reads Google News RSS feeds. It is not affiliated with or endorsed by Google.

For AI agents

Call the Actor with a JSON input like the example above and read the default dataset. Every row has a stable article_id, an ISO 8601 published_at, and a fetchable publisher url. In new-articles mode, repeated calls with the same input return only articles not returned before.

FAQ

Why are some url fields empty? Decoding is off by default. With it on, Google occasionally returns no URL for an article or rate-limits a large run; those articles are not charged, and their google_news_url still works in a browser.

Does new-articles mode share state between different inputs? Each feed (query, edition, date window) keeps its own history, keyed by your filters. Set State key to keep separate baselines for the same feeds, for example one per client.

Can I get articles older than a few weeks? Use Search from / to date. Google's coverage of older dates varies by query.