🌐 Global News Scraper — Google News + Publishers, One Schema

Search world news across Google News and top publisher feeds in a single run. Every other news scraper on the Store reads one source; this one merges search-engine discovery with curated publisher feeds and your own RSS, deduped across sources, into one clean schema — with the real article URLs, not Google redirect links.

Built for news monitoring — track a brand, topic or competitor across every outlet at once.

No API keys. No browser. No proxies. A default run finishes in a few seconds.

---

Why scrape Google News and publisher RSS together

The Store is full of Google News scrapers — and they all return the same thing: Google's redirect URLs, one source, no dedupe. This actor is built differently.

	Typical Google News scraper	This actor
Sources per run	1 (Google News)	Google News + 9 publishers + your feeds
Article URLs	news.google.com/rss/articles/... redirects	real publisher URLs (decoded)
Cross-source dedupe	none	✅ same story from two sources collapses, keeping both
Languages	usually English	en · vi · de · fr · es · ja editions
Schema	one shape per actor	one unified schema across every source

News article data fields you get

{
  "id": "a1b2c3d4e5f6",
  "title": "Tesla cuts EV prices across Europe",
  "url": "https://www.reuters.com/business/autos/tesla-...",
  "source": "Reuters",
  "source_type": "google_news",
  "url_decoded": true,
  "summary": "Tesla lowered prices on its Model 3 and Model Y across...",
  "language": "en",
  "published_at": "2026-08-14T09:12:00Z",
  "fetched_at": "2026-08-14T10:00:00Z",
  "also_seen_in": ["BBC"]
}

• url is the real publisher link, decoded from Google's redirect — competitors leave it as an unusable news.google.com URL.
• also_seen_in names every other source that carried the same story, so you keep the provenance instead of losing the duplicate.
• published_at is a proper timestamp — RFC-822 named-timezone dates (the standard RSS form) are parsed correctly, so nothing sorts to the bottom with a null date.

RSS feed scraper: Google News, BBC, NYT, Guardian and your own feeds

Source	What it covers
Google News	Any topic you search, from hundreds of outlets, in six language editions
Publisher feeds	BBC · Guardian · NYT · TechCrunch · The Verge · Ars Technica · Wired · Al Jazeera · CNBC — mapped to their topic sections when you search technology/business/science
Your feeds	Paste any RSS/Atom URLs to add outlets no pack anticipates

Google News scraper inputs

Input	What it does
search	Topic or keywords, e.g. "electric cars". Empty = top headlines.
sources	Which sources: google_news, publisher_rss, user_feeds.
locale	Google News edition: US · UK · Vietnam · Germany · France · Spain/LatAm · Japan.
feedUrls	Your own RSS/Atom feed URLs.
language	Keep only one language, or any.
excludeKeywords	Drop articles containing these words, e.g. "opinion, sponsored".
publishedWithinDays	Recency cutoff. Articles with no date are kept.
publishedAfter / publishedBefore	Explicit date range, e.g. 2026-01-01 to 2026-03-31.
maxItems	Hard cap, so a run can't produce an unexpected bill.

News archive: scrape a fixed date range

publishedWithinDays answers "the last 7 days". It cannot answer "January to March" — and a research or archive pull is almost always the second kind. Set publishedAfter and publishedBefore for an explicit window:

Input	Example
publishedAfter	2026-01-01
publishedBefore	2026-03-31 — the whole of that day is included

Either bound works alone: publishedAfter on its own means everything since, publishedBefore on its own means everything up to. Articles whose feed publishes no date are kept rather than dropped, so a source that omits pubDate doesn't vanish from a dated query.

Daily news brief — get only the new articles

Turn on onlyNew and schedule the Actor. The first run returns everything matching your search; every run after that returns only what was published since. A topic you track every morning becomes a short brief instead of the same headlines again.

Because you are billed per article returned, a daily brief costs a fraction of a full run.

Run	With onlyNew off	With onlyNew on
Monday	98 articles	98 articles
Tuesday	mostly the same 98	the 11 published overnight
Wednesday	mostly the same 98	the 9 published since

This also side-steps the ceiling described above: you are no longer trying to pull a large corpus in one run, you are collecting it a day at a time.

History is kept on your own account for 30 days. If storage is ever unavailable the run returns everything rather than silently returning nothing — an empty brief and a quiet news day look identical, and only one of them is a bug.

Send it to Slack every morning

1. Schedule → new schedule, 0 8 * * *, pointing at this Actor with onlyNew: true.
2. In n8n, Make or Zapier, add an Apify → Actor run finished trigger.
3. Map title, source, url and published_at into a Slack message.

url is already the real publisher link rather than a news.google.com redirect, so the message links straight to the article.

News Scraper pricing

Tiered by your Apify plan — bigger plans pay less per article:

Your Apify plan	Per article
Free	$0.004
Bronze	$0.003
Silver	$0.002
Gold / Platinum / Diamond	$0.002

Actor start: $0.00005 — effectively free.

Run	Articles	Free plan	Gold+
Default	50	$0.20	$0.10
Narrow search, everything available	~100	$0.40	$0.20
Top headlines, everything available	~310	$1.24	$0.62

You are charged per article returned, not per source read. A run pulls from every selected source and bills you only for what it delivers — so a query with little coverage costs little.

How many articles can one run actually return?

maxItems is a cap, not a target. Setting it to 1,000 does not produce 1,000 articles; it only stops the run from exceeding that. The real ceiling is set by the sources, and it is worth knowing before you plan a job around it. Measured:

Run	Articles returned
search: "artificial intelligence"	~100
search: "technology" (broad)	~170
no search — top headlines	~310

Two limits stack:

• Google News RSS returns at most 100 results per query. That is Google's own cap. No input changes it, and it applies per query — not per run.
• Publisher feeds are general news feeds. They carry ~280 articles between them, but when you search a narrow topic, only the articles that happen to mention it survive. "artificial intelligence" matched 1 of 281; "technology" matched 84.

To go beyond ~310, supply your own feeds. feedUrls accepts up to 25 feeds at 200 items each, so a run built on your own sources can reach several thousand articles. The built-in sources alone cannot.

News monitoring, headline tracking and research pipelines

• Media monitoring — track a company, product or topic across many outlets at once.
• Newsletter & digest automation — feed a daily brief from Google News + your chosen feeds.
• Research & sentiment datasets — one normalised, deduped, multilingual news corpus.
• LLM/RAG pipelines — clean article metadata with real URLs, ready to enrich downstream.

Google News scraper FAQ

Which news sources does this scraper cover?

Google News (any topic, six language editions), a curated pack of nine publishers — BBC, Guardian, NYT, TechCrunch, The Verge, Ars Technica, Wired, Al Jazeera and CNBC — and any RSS or Atom feed you supply yourself. All three are read in one run and merged into a single schema.

Does it return the real article URL or a Google redirect?

The real publisher URL. Google News hands out news.google.com/rss/articles/CBMi... redirect links that are useless for downstream fetching; this Actor decodes them. If a decode fails it returns the Google URL with url_decoded: false rather than guessing, so you always know which you have.

Can I scrape news in Vietnamese?

Yes — set locale to vn and language to vi. Language is detected from the script itself rather than assumed from the locale, so Vietnamese diacritics aren't confused with French accents and Japanese kana isn't misread as Chinese.

Can I use this for media monitoring?

That's the main use. Search a company, product or topic and get every outlet that covered it in one pass, with also_seen_in naming the other sources carrying the same story so you can measure pickup instead of counting the same article twice.

Does it return the full article text?

No — headlines, summaries, links and metadata only. Full article bodies are copyrighted, so redistributing them isn't something this Actor does. The real URL is there if you have your own licence to fetch the body.

Can I use this as an RSS feed scraper?

Yes — pass any RSS or Atom feed URLs in feedUrls. They're read alongside the built-in sources and deduped against them, so an outlet you add that also appears in Google News collapses into one row.

Can I use this as a headline scraper?

Yes — every row carries the headline, the outlet that published it, the publication timestamp and the real article URL. Headlines are the primary field; full article bodies are deliberately not returned (see above).

Can I scrape BBC News?

Yes. BBC is one of the nine built-in publisher feeds, read on every run alongside Guardian, NYT, TechCrunch, The Verge, Ars Technica, Wired, Al Jazeera and CNBC. A BBC story that Google News also carried collapses into a single row, with also_seen_in naming both.

Can I scrape news from a specific date range?

Yes — publishedAfter and publishedBefore take dates like 2026-01-01, and either can be used on its own. Note that this filters what the sources return; Google News RSS and publisher feeds are current-window feeds, so a range far in the past will match little. For deep archives, run the Actor on a schedule with onlyNew and build the corpus forward.

How much does it cost to scrape 1,000 news articles?

You're charged per article returned, not per source read. A run queries Google News plus every publisher feed and bills only for what it delivers. See the pricing table above.

Notes and limitations

• News articles are public data; this actor returns headlines, summaries and links, not full article bodies — no copyrighted text is redistributed.
• Google News redirect URLs are decoded to real publisher links where possible; if a decode fails, the row keeps the Google URL and url_decoded: false rather than guessing.
• One source failing degrades coverage, never the run — the log names what was skipped.