Similarweb Traffic and Rank Scraper

Collect the public, unauthenticated Similarweb traffic snapshot for one or many domains. Each successful row can include global, country, and category ranks; estimated visits; bounce rate; pages per visit; visit duration; traffic sources; country shares; and the category label exposed by Similarweb's public data endpoint.

Important endpoint limitation

This actor does not use a Similarweb API key. It first calls the legacy public endpoint at data.similarweb.com/api/v1/data?domain=.... Similarweb may return a CloudFront/WAF challenge, rate limit, empty response, or require credentials. Those cases become an uncharged diagnostic row with errorCode, HTTP status, and proxy-cost notes. The actor never invents traffic metrics from a blocked response.

Input

Field	Default	Description
domains	[]	Up to 100 public domains or HTTP(S) URLs.
domain	empty	Single-domain shortcut.
maxConcurrency	3	Parallel requests, from 1 to 10.
requestTimeoutSecs	15	Per-request timeout, from 5 to 30 seconds.
retryWithProxy	true	When direct access is blocked, retry through the supplied proxy configuration.
proxyConfiguration	off	Optional Apify proxy. Residential is a reliability fallback, not the default.

Example:

{
  "domains": ["similarweb.com", "apify.com", "example.com"],
  "maxConcurrency": 3,
  "retryWithProxy": true,
  "proxyConfiguration": {
    "useApifyProxy": true,
    "apifyProxyGroups": ["RESIDENTIAL"],
    "countryCode": "CA"
  }
}

Output and billing

One row is written per accepted domain. Empty input returns one labeled _sample row and is not charged. Invalid domains, unsafe DNS answers, WAF blocks, rate limits, and no-metric responses are diagnostics and are not charged. A successful metric row emits the PPE event domain.

proxyUsed and proxyCostRisk are included in every row. Residential proxy traffic can cost more than the actor compute, especially when the endpoint challenges and retries requests. For profitable usage, test the user's real proxy group and country with a batch before setting a low per-domain price; do not assume a direct-only margin applies to residential traffic.

Safety

Only DNS domain names and HTTP(S) URLs are accepted. IP literals, internal/special-use suffixes, credentials, malformed hostnames, and domains resolving to non-unicast addresses are rejected before the Similarweb request. The actor requests only the fixed Similarweb public endpoint; user input is encoded as a domain query value.

Local verification

npm install
npm test
npm run test:live -- example.com

The live smoke reports endpoint status and returns exit code 2 when Similarweb is blocked or exposes no usable public metrics. That is an endpoint diagnostic, not a fabricated success.