Similarweb Scraper — AI Traffic, $0.80/1K

Get Similarweb traffic estimates, rankings, channels, keywords, and available AI referral data for $0.80 per 1,000 domains with data. Enter domains and export a structured table; domains without data are not charged.

Tool preview: Explore Similarweb Scraper for recorded examples and the current preview status. Browse the tools directory for other sources.

• 19 core fields validated 100% against a leading Similarweb Actor on 105 domains
• Additional AI referral totals, assistant shares, trends, and prompt insights when available
• No charge for invalid, failed, duplicate, or no-data domains
• No Similarweb API key required

The table can include monthly visits, global and country rankings, engagement, traffic sources, top countries, top keywords, and AI assistants sending visitors. Export to CSV, Excel, JSON, or a warehouse.

Quick start: get website traffic estimates

1. Put your domains in the Domains field.
2. Click Start.
3. Export the Dataset, or read it with the API calls below.

{
  "domains": ["stripe.com", "nytimes.com", "openai.com"],
  "maxDomains": 1000
}

Similarweb traffic scraper input

Field	Type	Default	Notes
domains	array of strings	required	Hosts, bare or as full URLs. Scheme, path, query, port, credentials, trailing dot and case are all cleaned
maxDomains	integer	1000	Upper bound on unique domains in one run, 1 to 5000
maxConcurrency	integer	2	How many batches of 16 run in parallel, 1 to 10: up to 16–160 domain calls in flight. This is not a requests-per-second limit

Valid, unique domains preserve their first-input order. Duplicate domains are processed once, and invalid inputs are skipped without charge.

Internationalized domains are used exactly as you type them. The source keeps separate records for a Unicode spelling and its ASCII form, and in a check on 2026-09-19 only the Unicode spelling carried a rank and real traffic. Submitting both spellings gives you two rows, because upstream they are two different records.

Every row carries inputDomain (what you typed) and normalizedDomain (what was queried).

Memory and concurrency

Set run memory in Apify's run options and maxConcurrency in the input separately. Increasing memory gives the run more memory and CPU; it does not automatically raise concurrency. Each concurrency unit starts a batch of up to 16 domain calls, including calls waiting on the network or retrying. Calls in flight are not sustained QPS.

Use	Run memory	maxConcurrency	Maximum domain calls in flight	Evidence and limits
General starting recommendation	256 MB	2	32	Leaves more resource headroom than 128 MB. Verified with one 32-domain short run; not a long-run stability guarantee
Lower-resource option	128 MB	1	16	One 1,000-domain run completed with all rows, in input order, and matching charges
Current defaults	128 MB	2	32	Actual published defaults; the recommendation above has not been applied

The published input accepts maxConcurrency 1–10. Higher experimental values are not available in the published build. Short-run experiments reached 128 / 256 / 512 logical calls at 1 / 2 / 4 GB, but they do not establish sustained capacity or linear throughput scaling. Apify also provides a temporary startup CPU boost, which can affect short-run timing.

See the resource and no-data guide for configuration boundaries, evidence, and limitations.

Similarweb domain traffic data output

One row per domain. Several numeric fields in the core set are strings; that is the shape existing pipelines already parse, so it is preserved rather than cleaned up.

Core fields

Field	Type	Notes
domain	string	The queried host
title	string	"" when the source has none
description	string	"" when the source has none
category	string	Two-level, e.g. finance/banking_credit_and_lending. "" when absent
globalRank	number or null	null for sites without one, including many subdomains and link shorteners
countryRank	object	{ Country, CountryCode, Rank }, each null when absent
categoryRank	string or null
visits	string	Visits reported for the snapshot month. "0" does not by itself mean no data or prove zero real visitors
bounceRate	string	0 to 1
pagesPerVisit	string
timeOnSite	string	Seconds
estimatedMonthlyVisits	object	{ "YYYY-MM-01": visits }, three months
trafficSources	object	SocialOrganic, SocialPaid, Mail, Referrals, SearchOrganic, SearchPaid, Direct, GenAi, Affiliate, DisplayAds, each 0 to 1. A source the upstream did not report is null, not 0
topCountryShares	array	{ Country, CountryCode, Value }
topKeywords	array	{ name, esitmatedValue, cpc }. The misspelled key comes from the source and is preserved so existing consumers keep working
competitors	array	Empty, see Limits
isDataFromGA	boolean	The site owner publishes its own analytics
screenshot	string	Source-hosted preview image
snapshotDate	string	The month the data describes

Additional fields

Field	Type	Notes
inputDomain	string	What you typed
normalizedDomain	string	What was queried after cleanup
dataAvailable	boolean	Whether the source has measurable traffic. false rows are written so your list stays aligned and are never charged
keywordVolumes	array	{ name, volume }, aligned with topKeywords
aiTraffic	object	See below

aiTraffic carries available, totalVisits, share (0 to 1), distributionBoundary, sources and prompts. Each entry in sources is one AI assistant with source, rank, share and monthlyHistory (three months of { date, share }). Unknown numbers are null, never 0; a confirmed absence of entries is an empty array.

aiTraffic.available and dataAvailable are different states. A site can have measurable traffic while the snapshot has no available AI referral data.

Example row, trimmed:

{
  "domain": "stripe.com",
  "title": "Link by Stripe: One-click payments",
  "globalRank": 312,
  "countryRank": { "Country": 840, "CountryCode": "US", "Rank": 226 },
  "visits": "133172721",
  "bounceRate": "0.5545453726745924",
  "estimatedMonthlyVisits": { "2026-06-01": 120792625, "2026-07-01": 129949921, "2026-08-01": 133172721 },
  "trafficSources": { "Direct": 0.4318, "SearchOrganic": 0.0596, "Referrals": 0.3898 },
  "topKeywords": [{ "name": "stripe", "esitmatedValue": 3087760, "cpc": 2.36 }],
  "competitors": [],
  "snapshotDate": "2026-08-01T00:00:00+00:00",

  "inputDomain": "stripe.com",
  "normalizedDomain": "stripe.com",
  "dataAvailable": true,
  "keywordVolumes": [{ "name": "stripe", "volume": 6120000 }],
  "aiTraffic": {
    "available": true,
    "totalVisits": 1607549,
    "share": 0.0136,
    "distributionBoundary": "<100M",
    "sources": [
      { "source": "grok.com", "rank": 1, "share": 0.4607,
        "monthlyHistory": [{ "date": "2026-08-01", "share": 0.4607 }] }
    ],
    "prompts": ["What are the steps to integrate a modern online payments platform…"]
  }
}

Pricing

$0.80 per 1,000 delivered domains that carry data.

• A domain with no measurable traffic is written to the Dataset so your list stays aligned with your input, and it is not charged.
• Invalid entries, failed lookups and duplicate inputs are not charged.
• A resumed run does not refetch or recharge a domain already in the Dataset.

The run summary records charged alongside delivered, so you can reconcile the bill against the Dataset.

Traffic estimate coverage and limits

Stated here rather than discovered later.

This Actor returns the traffic and AI insights available in the current public website snapshot. It does not provide every dataset available in Similarweb's paid platform.

Included: global, country and category rankings; monthly visits and three months of history; bounce rate, pages per visit and time on site; traffic share by country; the channel split; top keywords with estimated value and volume; AI referral totals, per-assistant shares with monthly history, and prompt text where the source exposes it; title, description and category.

Not included:

• Similar sites and competitors. competitors is an empty array; the public source does not return it.
• Full keyword database, incoming and outgoing referral lists, named social sources, ad networks and publishers.
• Demographics, audience interests and audience overlap.
• Subdomain, folder and page-level traffic.
• Arbitrary country, device, channel or date-range queries. Every row is the current monthly snapshot; snapshotDate says which month.
• Long-term history. estimatedMonthlyVisits holds three points.
• App intelligence and company or industry databases.

Two behaviours worth knowing: small sites, most subdomains and link shorteners often have null ranks even with hundreds of millions of visits; and a subdomain returns its own numbers, not the parent domain's.

Compatibility

Designed as a compatible superset of the leading Similarweb traffic Actor's core dataset. The 19 core fields keep their original names, types and null rules, verified field by field against a live run of that Actor on 105 domains. Everything else is additive and sits alongside them, so an existing pipeline keeps working without changes.

API and automation

Read the Dataset directly:

curl "https://api.apify.com/v2/datasets/DATASET_ID/items?format=csv&clean=true" \
  -H "Authorization: Bearer APIFY_TOKEN"

Start a run and wait for its Dataset in one call:

curl -X POST "https://api.apify.com/v2/acts/ACTOR_ID/run-sync-get-dataset-items" \
  -H "Authorization: Bearer APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"domains":["stripe.com","nytimes.com"]}'

Each run also writes an OUTPUT record to the key-value store with how many domains were requested, unique, skipped, duplicated, delivered, charged, delivered without data, and failed.

Schedules and webhooks work as with any Actor. Runs are independent, so a monthly schedule matches how often the underlying numbers change.

FAQ

Why is visits a string but globalRank a number?

Because that is the shape the reference Dataset publishes, and changing it would break consumers that already parse these columns. The types are pinned by tests.

A domain returned a row with zeroes. Was I charged?

Check dataAvailable, rather than an individual zero. A delivered row with dataAvailable: false is not charged: the public snapshot reports zero visits, no global rank, and marks the site as small. This means no measurable traffic in that snapshot, not proof that the site has no real visitors. Other rows with a zero can still have dataAvailable: true and are charged.

HTTP, rate-limit, or parsing failures are separate from no-data rows. Failed lookups are counted in OUTPUT.failed and are not charged.

Why did I get fewer rows than domains I submitted?

Duplicates collapse to one row, and entries that cannot be a domain are skipped. The run summary reports uniqueDomains, duplicateInputs, skippedInvalid and failed.

A site has traffic but aiTraffic.available is false.

The snapshot has no available AI referral data for that site. This does not prove that no AI assistant ever sent visitors, and it is separate from dataAvailable.

Why is the data a month old?

The source publishes monthly. snapshotDate tells you exactly which month each row describes.

Can I get competitors or similar sites?

Not from this Actor. See Limits.

My run stopped early.

If stoppedOnBudget is true in the run summary, the run hit the maximum charge you allowed. Raise the limit and start it again — already delivered domains are not refetched or recharged.

Legal

This Actor is independently developed and is not affiliated with, endorsed by, or sponsored by Similarweb.

It reads a public endpoint. It does not sign in, does not use cookies or credentials, and does not reach anything behind a login. You are responsible for how you use the data, including any applicable terms and data-protection rules.