Google Condensed SERP Scraper

Google Condensed SERP Scraper

This Actor extracts structured public content from Google Search pages into one record per inspected results page. It is useful for research, SEO review, and sending a rich search snapshot to downstream tools or an LLM.

Verified target and scope

The verified target is the public Google Search /search page on supported Google web domains. The default query is machine learning. You can provide search phrases or approved Google Search URLs; a URL must use HTTPS, a supported Google hostname, and the /search path.

The Actor reads public search pages only. It does not log in, solve CAPTCHAs, bypass access controls, evade rate limits, or access private, paid, or authenticated content. Set renderJavaScript to false to use direct requests; with rendering enabled, an optional Apify proxy applies to browser requests.

What it extracts

When Google exposes these sections, a record can include:

• the searched phrase, result vertical, and approximate visible result-count label and estimate;
• AI Overview text, titled sections, and citation destinations;
• organic result titles, resolved destinations, displayed breadcrumbs, publisher/domain labels, snippets, visible dates, grouped links, and visible rating, review-count, or price labels;
• result-card images with their public URLs, alt text, positive HTML width/height attributes, and a derived aspect ratio when both dimensions are available; existing thumbnail, favicon, image URL, and image-count fields remain available;
• separately structured sponsored search cards, including advertiser, resolved destination, displayed URL, visible ad copy, grouped links, and card images;
• video titles, destinations, channel/date labels, duration, view-count text, descriptions, and thumbnails;
• structured People also ask questions with expanded answer text and citations when visible;
• related-search text and corresponding Google Search destinations;
• a bounded plain-text condensation.

Google can change its markup, localize sections, omit modules, or block access. Only values visibly exposed by the page are included; unavailable optional fields are omitted, not fabricated.

Quick start

1. Enter a search phrase or a public Google Search URL in the Input tab.
2. Choose which result sections to include and set page and result limits.
3. Start the Actor and review the search-page record in the Dataset tab.

Input

Use queries, query, or searchQuery for search phrases. Use startUrls, sourceUrls, urls, or url for approved public Google Search URLs. Explicit URLs take precedence over generated query URLs.

Important input fields and bounds:

Input	Default	Bound or behavior
queries	["machine learning"]	Up to 50 phrases, 1–300 characters each; query and searchQuery are single-phrase aliases
startUrls	[]	String array of approved HTTPS Google /search URLs; sourceUrls and urls are array aliases, url is a single-URL alias; at most 50 unique URLs total
startPage / page	1	One-based starting result page, 1–100; a positive start offset in a supplied URL takes precedence
maxItems	1	1–50 output records globally across all sources
maxPages	1	1–10 pages for each source
maxResults	20	1–50 organic web results per record
maxVideos	10	1–30 video cards per record
maxPeopleAlsoAsk	20	1–50 question cards per record
maxRelated	20	1–50 related searches per record
maxImages	10	0–50 result-card images; 0 omits image data
includeOrganic, includeSponsoredResults, includeVideos, includePeopleAlsoAsk, includeRelatedSearches, includeAiOverview, includeImages	true	Boolean controls for whether each section is retained
locale	en	Generated URL hl value, at most 20 characters
safeSearch	true	Adds Google's safe=active parameter to generated URLs
acceptLanguage	en-US,en;q=0.8	Request language header, at most 120 characters
renderJavaScript	true	Render the public search page; set false for direct requests
proxyConfiguration	omitted	Optional Apify proxy for browser requests; direct requests do not use the proxy

URL aliases are merged and deduplicated; explicit approved URLs take precedence over generated queries. A URL must be HTTPS, use a supported Google hostname, use the /search path, and contain a non-empty q parameter. Each output row has the same 21 top-level fields: absent result modules are empty arrays and unavailable optional values are null. Images retain structured element metadata in images[] alongside the URL and thumbnail fields.

Quick input:

{
  "queries": ["machine learning"],
  "maxItems": 1,
  "maxPages": 1
}

Custom URL with selected output sections:

{
  "startUrls": ["https://www.google.com/search?q=machine+learning"],
  "page": 1,
  "maxItems": 1,
  "maxPages": 2,
  "includeImages": false,
  "includeVideos": false
}

Available dataset fields

Record section	Source-backed fields
Search context	searchedPhrase, searchType, visible resultCountText, approximate resultCountEstimate
AI Overview	Full visible text, titled aiOverviewSections[], public aiOverviewSources[] citations
Organic results	Title, resolved URL, displayed URL, publisher/domain, snippet, visible date, breadcrumbs, grouped result links, rating/review/price labels, rich snippet, favicon, thumbnail, and structured image metadata
Sponsored results	Ad headline, advertiser, visible sponsored label and copy, destination, grouped links, source domain, and card media
Videos	Title, video URL, channel, publication label, duration, description, view-count label, source domain, and media
People also ask / related	Questions, visible expanded answer and citations, related query text and Google Search URL
Image results	Image title, destination, publisher/domain, alt text, source dimensions, and derived aspect ratio
Things to know	Visible topic/question pairs and any answer citations exposed in the page
People also search for	Visible suggestions with approved Google Search URLs
What people are saying	Public post/discussion link, visible copy, date, engagement label, and preview images
News / shopping / featured / local	Structured source links and visible card details, including price/rating when present
Condensed view	condensedText, composed only from the retained visible result sections

Optional modules are empty arrays when absent; individual source values that are not exposed remain unset inside their nested objects. Results have nested JSON objects and arrays; this keeps image, citation, answer, and result-link structure instead of flattening the page into one string. Although the row shape is stable and has 21 top-level fields, no individual Google result page is guaranteed to expose a non-empty value for every module.

Dataset output

Each dataset row represents one inspected search results page and its visible sections. If Google blocks a page or returns no usable content, the run summary and logs report the issue separately.

Example normal row:

{
  "searchedPhrase": "machine learning",
  "searchType": "web",
  "resultCountText": "About 1,230 results",
  "resultCountEstimate": 1230,
  "aiOverviewText": "Machine learning helps computers learn patterns from data.",
  "aiOverviewSections": [
    { "title": "How it works", "text": "Algorithms learn from examples." }
  ],
  "aiOverviewSources": [
    { "title": "Example Research", "resultUrl": "https://example.test/ml", "sourceName": "example.test" }
  ],
  "organicResults": [
    {
      "title": "Machine Learning Guide",
      "resultUrl": "https://example.test/ml-guide",
      "displayedUrl": "example.test › guides › ml",
      "sourceName": "Example Research",
      "sourceDomain": "example.test",
      "snippetText": "A practical guide to machine learning concepts.",
      "publicationDateText": "2 days ago",
      "breadcrumbTrail": ["example.test", "guides", "ml"],
      "faviconUrl": "https://example.test/favicon.ico",
      "thumbnailUrl": "https://example.test/ml-chart.jpg",
      "images": [
        { "imageUrl": "https://example.test/ml-chart.jpg", "altText": "Machine learning chart", "width": 320, "height": 180, "aspectRatio": 1.7778 }
      ],
      "imageUrls": ["https://example.test/ml-chart.jpg"],
      "imageCount": 1,
      "ratingText": "4.8 out of 5 stars",
      "ratingValue": 4.8,
      "reviewCountText": "235 reviews",
      "reviewCount": 235,
      "priceText": "$24.99",
      "resultLinks": [
        { "title": "Training and evaluation", "resultUrl": "https://example.test/ml-guide/training" }
      ]
    }
  ],
  "sponsoredResults": [
    {
      "title": "Train and deploy machine learning models",
      "resultUrl": "https://example.test/ml",
      "displayedUrl": "example.test › ml",
      "sourceName": "Example Cloud",
      "sourceDomain": "example.test",
      "sponsorLabel": "Sponsored result",
      "snippetText": "Managed training tools for machine learning.",
      "resultLinks": [
        { "title": "Pricing", "resultUrl": "https://example.test/pricing" }
      ]
    }
  ],
  "videos": [
    { "title": "Machine Learning Explained", "videoUrl": "https://www.youtube.com/watch?v=example", "channelName": "Example Channel", "durationText": "12:34", "viewCountText": "12K views" }
  ],
  "peopleAlsoAsk": [
    { "question": "What is machine learning?", "answerText": "Algorithms learn patterns from data.", "sourceTitle": "Example Research", "sourceResultUrl": "https://example.test/definition", "answerSources": [{ "title": "Example Research", "resultUrl": "https://example.test/definition", "sourceName": "example.test" }] }
  ],
  "relatedSearches": [
    {
      "text": "Machine learning course",
      "searchUrl": "https://www.google.com/search?q=machine+learning+course"
    }
  ],
  "imageResults": [
    {
      "title": "Machine learning diagram",
      "resultUrl": "https://example.test/ml-visuals",
      "sourceName": "Example Research",
      "sourceDomain": "example.test",
      "imageUrl": "https://example.test/ml-diagram.jpg",
      "altText": "Machine learning diagram",
      "width": 640,
      "height": 360,
      "aspectRatio": 1.7778
    }
  ],
  "thingsToKnow": [
    {
      "topic": "Types",
      "question": "What are the different types of machine learning?"
    }
  ],
  "peopleAlsoSearchFor": [
    {
      "text": "machine learning algorithms",
      "searchUrl": "https://www.google.com/search?q=machine+learning+algorithms"
    }
  ],
  "whatPeopleAreSaying": [
    {
      "title": "Machine learning discussion",
      "resultUrl": "https://example.test/discussion",
      "sourceName": "example.test",
      "sourceDomain": "example.test",
      "snippetText": "A public discussion about machine learning.",
      "publicationDateText": "2 days ago",
      "engagementText": "120 likes",
      "imageUrls": []
    }
  ],
  "newsResults": [],
  "shoppingResults": [],
  "featuredSnippets": [],
  "localResults": [],
  "condensedText": "Search: machine learning\nWeb results:\n- Machine Learning Guide (Example Research)"
}

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

The run summary and access information are available alongside the dataset in the Output tab.

Cost and performance

The number of query pages, result limits, and selected sections affect run time and dataset size. Google Search availability and result content can vary. Apify compute or proxy charges depend on your account settings.

FAQ

Can I pass any website URL?

No. This Actor accepts only public Google Search /search URLs.

Why is a section missing?

Google may omit AI Overview, videos, People also ask, related searches, or other modules depending on query, locale, device, consent state, and current markup. The Actor omits unavailable optional fields.

Are result ranks or request counters included?

The dataset contains the result sections exposed by Google. Page counts and collection status are shown in the separate run summary.

Does this bypass Google protections?

No. The Actor stops at access boundaries and reports them in the run summary and logs.

Disclaimer

Use this Actor only for lawful, public-page research and in accordance with Google’s terms, applicable law, and any applicable robots or access policies. Search results and markup are provided by Google and can change without notice. The Actor is not affiliated with Google and does not guarantee completeness, ranking accuracy, availability, or suitability for a particular decision. Do not use it to collect personal data without a lawful basis.