Google News Scraper

What does Google News Scraper do?

Google News Scraper turns Google News into structured data. Give it search queries or paste Google News links (topics, sections such as Business or Technology, publications, top stories), pick a country and language, and optionally a time range. You get one row per article with the headline, publisher, the real article link on the publisher's site (not a news.google.com redirect), the publish time in UTC and its position in the results. Switch on full article text to also get each article's text, summary, lead image and authors.

Try it with the prefilled query: open the Actor, click Start, and 20 articles arrive in about 5 seconds. Because it runs on the Apify platform, you also get API access, scheduled runs, integrations (Google Sheets, Make, Zapier, webhooks, Slack), run monitoring and automatic retries.

Why scrape Google News?

• Media monitoring and PR: track every story about your brand, products, competitors or executives, in any country and language Google News covers.
• Market and investment research: follow news on companies, sectors and commodities, and feed headlines into dashboards or sentiment models.
• News alerts: schedule a run every hour with Published within: Last hour and send new articles to Slack or email.
• Research and AI datasets: collect thousands of articles on a topic over a date range, with full text, for analysis or model training.
• SEO and content research: see which publishers rank in Google News for your keywords.

How to scrape Google News

1. Type what you want to follow into Search queries, one query per line, for example artificial intelligence or "interest rates" site:reuters.com.
2. Or open news.google.com, go to a topic, section or search, and paste the address into Google News links.
3. Set Country and Language for the edition you want (US + en, GB + en, DE + de, FR + fr, IN + hi, EG + ar…).
4. Optionally pick Published within (last hour, 24 hours, 7 days, 30 days, year) or a date range, and set Max articles per query or link.
5. Click Start, then download the results as JSON, CSV, Excel or HTML.

Input

Everything is set on the Input tab. In JSON, the input looks like this:

{
    "queries": ["artificial intelligence", "\"interest rates\" site:reuters.com"],
    "startUrls": ["https://news.google.com/topics/CAAqJggKIiBDQkFTRWdvSUwyMHZNRGx6TVdZU0FtVnVHZ0pWVXlnQVAB?hl=en-US&gl=US&ceid=US:en"],
    "country": "US",
    "language": "en",
    "timeRange": "7d",
    "maxResults": 500,
    "includeArticleText": false
}

Field	What it does
queries	Search words, one query per line. Google's operators work: "exact phrase", -word, OR, site:, intitle:.
startUrls	Google News links: a search, topic, section (/headlines/section/topic/TECHNOLOGY), publication, or the top stories page. Plain strings or {"url": "…"} objects; https:// is optional. A link with its own hl, gl and ceid keeps that edition.
country	Two-letter code of the edition, such as US, GB, DE, IN. Default: US.
language	Language of the edition, such as en, de, fr, ar, pt-BR. Default: en.
timeRange	any (default), 1h, 1d, 7d, 30d or 1y.
dateFrom, dateTo	A date range instead, such as 2025-01-01 to 2025-01-31. Days are Google's own calendar days (US Pacific time): an article is kept when its Pacific date is from dateFrom through dateTo. Google also sends articles of the day after the range; they are dropped, not saved.
maxResults	Articles per query or link. Default: 100, up to 10,000.
includeArticleText	Adds the full text, description, lead image and authors from the publisher's page. Default: off.

If Google has no edition for a country and language (for example Egypt in English), it serves its nearest edition; the run's status message says which, and each row's edition field shows the edition the article really came from.

Output

Each article is one row. A real row from the prefilled run:

{
    "title": "The AI doom scenario you've never heard of may be the hardest to stop",
    "publisher": "CBS News",
    "publisherUrl": "https://www.cbsnews.com",
    "url": "https://www.cbsnews.com/news/ai-doom-scenario-gradual-disempowerment/",
    "googleNewsUrl": "https://news.google.com/rss/articles/CBMieEFVX3lxTE5mMmdEVkNzd21X…?oc=5",
    "publishedAt": "2026-10-19T15:33:00Z",
    "publishedDateOnly": false,
    "position": 1,
    "query": "artificial intelligence",
    "feedUrl": "https://news.google.com/search?q=artificial%20intelligence&hl=en-US&gl=US&ceid=US%3Aen",
    "edition": "US:en",
    "articleId": "CBMieEFVX3lxTE5mMmdEVkNzd21X…",
    "scrapedAt": "2026-10-20T14:01:08.674347+00:00"
}

With Add full article text on, each row also carries the page's text and metadata (text shortened here):

{
    "title": "How electric school buses power 1,000 homes in California – and make money",
    "publisher": "Electrek",
    "url": "https://electrek.co/2026/10/20/how-electric-school-buses-power-1000-homes-in-california-and-make-money/",
    "publishedAt": "2026-10-20T14:00:00Z",
    "articleText": "A fleet of 74 school buses have spent the last two years powering about a thousand homes, taking stress off the grid, thanks to V2G chargers built in America.…",
    "wordCount": 893,
    "description": "These US-built V2G chargers power a thousand homes with school bus batteries, saving the grid and making money along the way.",
    "imageUrl": "https://electrek.co/wp-content/uploads/sites/3/2024/05/Zum_electric-school-buses-v2g.jpg?quality=82&strip=all&resize=1200,628",
    "authors": ["Jameson Dow", "Fred Lambert"]
}

You can download the dataset in various formats such as JSON, HTML, CSV, or Excel.

Data fields

Field	Description
title	Headline, without the " - Publisher" Google adds to it.
publisher	Publisher name, such as Reuters or BBC.
publisherUrl	Publisher's home page.
url	The article on the publisher's own site, decoded from Google's redirect link. In the rare case it can't be decoded, the Google News link (the run log counts these).
googleNewsUrl	The article's news.google.com link.
publishedAt	Publish time, ISO 8601 in UTC.
publishedDateOnly	true when Google gives only the publish day, not the time: publishedAt is then midnight US Pacific time of that day (07:00Z or 08:00Z). Common for articles older than a few days.
position	1-based order within the query or link.
query	The search words (empty for topic, section and top stories links).
feedUrl	The Google News page the article was listed on.
edition	The edition served, such as US:en or DE:de.
articleId	Google's id for the article.
articleText, wordCount	Full text and its length. Only with full article text; empty when the page is paywalled, blocked or has no readable body.
description, imageUrl, authors	The page's own summary, lead image and authors. Only with full article text.
scrapedAt	When the row was saved.

How much does it cost to scrape Google News?

You pay per article saved; see the Pricing tab for the exact price. Rows with full article text also count one article-text event each, and only when text was actually read. Nothing is charged for queries that find nothing or links that can't be read.

The Actor uses plain HTTP requests (no browser) and no paid proxy in normal runs, so it is fast: the prefilled run finishes in about 5 seconds, and 2,000 articles take about 2 minutes. Apify's monthly free credit is enough to try it on a few thousand articles.

Tips and advanced options

• More than 100 articles per query. A Google News feed holds about 100 articles. Above that, the Actor splits a search into date windows and reads them newest first, down to single days, so a popular query can return thousands of articles. Without a time range it goes back up to a year; set dateFrom/dateTo to go further back. For older dates Google files articles by day only, so one day yields at most about 100 articles.
• Order of results. Up to 100 articles come in Google's relevance order; above 100 (paged by date) they come newest first.
• Topic and section links list 30-100 current articles and can't be paged back in time. A time range is still applied to their articles.
• Hourly monitoring. Use timeRange: "1h" on a schedule; each run then reads one feed per query and finishes in seconds.
• Full text takes longer. Each article is then opened on the publisher's site and its page parsed, which is CPU-heavy: about 100 articles in 2 minutes at the default 512 MB, or in 45 seconds at 2048 MB (Run options → Memory), for a slightly higher platform cost. Leave it off when headlines and links are enough.
• Each article appears once per query or link; the same article found by two queries appears under each of them.

FAQ, disclaimers and support

Is it legal to scrape Google News? The Actor reads public pages and feeds that anyone can open without logging in. Whether you may use the results depends on your purpose and where you are: Google's terms and the publishers' copyright apply to article text, so check them before republishing. Articles can name people; personal data is protected by laws such as the GDPR, so only process it with a legitimate reason. If in doubt, ask a lawyer.

Why did a query or link return nothing? It is listed with the reason in the FAILED_INPUTS record of the run's key-value store, and the run's status message sums it up. Typical reasons: no articles match the query and time range, a link that isn't a Google News list page (story "Full coverage" pages and single article links have no article list), or a topic id Google doesn't know. These are not charged.

Why is articleText empty for some articles? Paywalled pages, pages that block automated visits and pages that load their text with scripts give no readable body. The row is kept with its headline and link, and the article-text event isn't charged for it.

Can I get a specific country's news? Yes: set country and language to an edition Google News offers, or paste a news.google.com link from that edition.

Found a bug or need another field? Open an issue on the Issues tab and we'll look at it quickly. Need a custom news pipeline? Get in touch through the Issues tab as well.