Similarweb Scraper

Both Similarweb datasets in one Actor. Pick a mode with scraperType: read the full monthly profile for a list of domains, or export the current global top-website ranking.

scraperType	What a row is	What you supply
websiteAnalytics	One complete monthly profile per domain	domains
topWebsites	One website from the global ranking	nothing

Website analytics

Paste hostnames or complete page links:

{
  "scraperType": "websiteAnalytics",
  "domains": [
    "openai.com",
    "https://github.com/features"
  ],
  "maxItems": 0
}

Links are reduced to their hostnames and duplicate domains are read once.

Field	What it records
domain, url, snapshot_date	Website identity and the month represented by the estimates
engagement	Visits, monthly change, bounce rate, pages per visit, duration, and history
rankings	Global, country, and category ranks with previous values and histories
top_countries, traffic_sources	Geographic and acquisition-channel shares
competitors	Related domains with category rank and affinity where published
demographics, audience_interests	Published age, gender, website, topic, and category interests
search, social, referrals	Search mix and keywords, social networks, and referral sites
technologies	Public technology-category counts
company	Name, founding year, headquarters, employee and revenue ranges, and parent domain

A measured July 2026 profile for openai.com included this engagement and ranking context:

{
  "domain": "openai.com",
  "snapshot_date": "2026-07-01T00:00:00+00:00",
  "engagement": {
    "visits": 197235347,
    "visits_change": 0.03595051048290677,
    "bounce_rate": 0.5786326520050872,
    "pages_per_visit": 2.966191555517269,
    "average_visit_duration": "00:02:31"
  },
  "rankings": {
    "global": { "rank": 212, "previous_rank": 200, "change": -12 },
    "country": { "rank": 261, "previous_rank": 250, "country_code": "US" },
    "category": { "rank": 5, "previous_rank": 5, "change": 0 }
  }
}

Global ranking

The ranking mode needs no websites at all:

{
  "scraperType": "topWebsites",
  "maxItems": 0
}

Rows carry rank, domain, rank_change, new_to_ranking, category_id, average_visit_duration, pages_per_visit, bounce_rate, icon_url, snapshot_date, and scope. The measured ranking page began with:

{
  "rank": 1,
  "domain": "google.com",
  "rank_change": 0,
  "category_id": "computers_electronics_and_technology/search_engines",
  "average_visit_duration": "00:09:58",
  "pages_per_visit": 8.676456201759066,
  "bounce_rate": 0.28492742035110014,
  "new_to_ranking": false,
  "scope": "global"
}

Things worth knowing before the first run

How many rows does the ranking return? The public surface currently contains the global top 50, so maxItems: 0 saves those 50 and a positive value takes a smaller slice from the top. The ranking mode ignores domains entirely.

Why does the ranking mode need a mode at all — can it not just run when domains is empty? Because the two datasets answer different questions and one of them costs nothing to ask. Choosing explicitly means a run never quietly produces the ranking when a domain list was meant to be filled in, or demands a domain list to read a table that takes none.

What does maxItems: 0 mean in the analytics mode? The whole supplied list, however long. A domain with no public profile is skipped without cancelling the rest of the batch.

Are these first-party analytics? No. They are third-party estimates tied to snapshot_date. Some values on an anonymous profile are unavailable; missing values stay null, and locked referral entries remain explicitly locked rather than being converted to zero.

Can one run measure rank movement over time? No — both modes return a snapshot. Schedule the Actor and retain each run, then join on snapshot_date.

Prefer a narrower form? Both modes also ship on their own: Similarweb Website Analytics Scraper and Similarweb Top Websites Scraper.