Human-Like Website Pageview Traffic Simulator

Send real Chromium pageviews to a website you own or are authorized to test. Choose how many pages to load, how deep each visit goes, how long it stays, whether it scrolls or follows internal links, which device and country it appears from, and which referrer it reports. Every attempt is written as a dataset row with its measured status, timing, device, referrer, analytics requests and error. A run-level summary is checkpointed while the run progresses, and a graceful stop writes an aborted row for an in-flight page.

This is a testing and monitoring tool. It does not solve CAPTCHAs, click ads, inflate YouTube views, or promise that the visits will appear in Google Analytics or improve rankings. Use it on your own sites or sites you have permission to test.

What this actor does

• Real browser visits: each page loads in Chromium with JavaScript, cookies, redirects and lazy content working normally.
• Pageview volume: request 1–1,000 pageviews; each visit gets its own browser context, cookies and device profile.
• Session depth: attempt to follow one visible same-site link per page, up to 10 pages per visit. A visit can stay on one page when no link is followed or the global pageview budget is used.
• Human behaviour: a jittered dwell target, mouse movement, measured scroll depth and optional internal clicks. A failed click attempt can extend the measured time on a page.
• Device, referrer and geo controls: desktop, Android mobile, Android tablet or a mixed run; direct, search, social or custom referrer; optional residential country for geo QA.
• Evidence, not claims: per-page HTTP status, final URL, load time, actual dwell, actual scroll, links followed, analytics requests observed and a clear error code. A SUMMARY record is written to the key-value store and checkpointed during the run.

Output per pageview

• url — the target URL you asked for (before redirects or UTM tagging)
• finalUrl — the final URL after redirects
• status — success, failed, blocked, timeout or aborted
• statusCode — HTTP response status
• pageTitle — HTML title, omitted for non-HTML pages
• loadTimeMs — time until the DOM was ready
• timeOnPageSec — actual time spent on the page after loading, including click attempts
• scrollDepthPercent — measured scroll depth, omitted when scrolling is off or no range is measurable; a very short page with a tiny overflow can report 100
• linksFound — same-site links present in the page, not only visible ones
• linksFollowed — internal links followed from this page
• referrer — the actual document.referrer seen by the page
• userAgent, viewport, device — the browser profile used
• visitId, pageviewIndex, pageIndex — traceability for sessions and loops
• analyticsTags — analytics requests observed, e.g. ga4, gtm, meta_pixel, hotjar (not proof that the platform accepted a hit)
• errorCode, error — a short code and message for failed, blocked or timed-out pages
• timestamp — ISO-8601 UTC time the page attempt started

Empty or unmeasured fields are omitted, not filled with empty strings or fake zeros. A successful page has no errorCode. A graceful stop writes aborted for pages that have not yet emitted a row; a hard kill can still omit an in-flight row, while pages already written stay success.

A SUMMARY record contains requested/attempted pageviews, success/failed/blocked/timeout/aborted counts, error rate, median and p95 load time, median dwell, analytics-request rates and per-target counts. It is checkpointed during the run, so a stopped run still has the latest version.

Input

Field	Type	Default	Description
targetUrls	array	–	Public pages you own or are authorized to test (the prefill uses https://example.com); legacy startUrls is accepted in code
pageviews	integer	10	Total page loads to attempt (1–1,000)
pagesPerVisit	integer	1	Maximum pages in one visit (1–10); above 1 attempts same-site link clicks when clicks are enabled and the global pageview budget allows
visitStyle	string	human	human for dwell/scroll/click, quick for a fast load check
dwellSeconds	integer	20	Target seconds per page for human visits (1–300)
referrerMode	string	direct	direct, search, social or custom
customReferrer	string	–	Full http(s) URL used when referrerMode is custom
scroll	boolean	true	Scroll during human visits; measured depth is omitted if not measurable
clickInternalLinks	boolean	true	Attempt one visible same-site link per page; never ads, forms or external links
deviceMix	string	mixed	desktop, mobile, tablet or mixed (Android Chromium emulation for mobile/tablet)
visitCountry	string	none	Optional ISO country for residential-proxy geo QA; none keeps visits direct
blockAds	boolean	true	Block ad domains; analytics requests still load so tag firing can be measured
blockImages	boolean	false	Skip images, fonts and media for faster visits
utmSource	string	–	Optional utm_source label appended to target URLs, e.g. traffic-test

Example: load check on your own site

{
  "targetUrls": [
    { "url": "https://www.example.com/" },
    { "url": "https://www.example.com/pricing" }
  ],
  "pageviews": 20,
  "pagesPerVisit": 1,
  "visitStyle": "human",
  "dwellSeconds": 15,
  "deviceMix": "mixed",
  "referrerMode": "direct"
}

Example: analytics-tag QA with a test-source label

{
  "targetUrls": [{ "url": "https://www.example.com/" }],
  "pageviews": 10,
  "pagesPerVisit": 2,
  "visitStyle": "human",
  "dwellSeconds": 25,
  "scroll": true,
  "clickInternalLinks": true,
  "referrerMode": "search",
  "utmSource": "traffic-test",
  "blockAds": true
}

Use the analyticsTags field and the SUMMARY rates to confirm which analytics requests the page made. Test traffic may still be filtered by an analytics platform; this actor does not guarantee attribution.

Example: mobile and geo QA

{
  "targetUrls": [{ "url": "https://www.example.com/" }],
  "pageviews": 12,
  "pagesPerVisit": 1,
  "visitStyle": "human",
  "dwellSeconds": 10,
  "deviceMix": "mobile",
  "visitCountry": "DE",
  "referrerMode": "social"
}

Country visits use a residential proxy in that country. If the proxy is unavailable, the page is reported as failed instead of silently falling back to a direct request.

Example: quick smoke test

{
  "targetUrls": [
    { "url": "https://www.example.com/" },
    { "url": "https://www.example.com/health" }
  ],
  "pageviews": 2,
  "visitStyle": "quick",
  "blockAds": true,
  "blockImages": true
}

Use cases

• Load and cache warm-up — load key pages after a deploy with a realistic browser and short dwell.
• Analytics and tag QA — check whether GA4, GTM, Meta Pixel, Hotjar or other tags actually send data on each page.
• Mobile checks — load pages with Android mobile or tablet profiles and compare status, title and timing.
• Geo checks — request visits through a residential country to test redirects, consent banners, prices or language.
• Uptime and broken-link checks — schedule a small run and watch for failed, blocked, timeout or non-2xx statuses.

FAQ

Do I need a proxy?

No, for new inputs. Visits are direct by default. Set visitCountry only when you need a residential exit in a specific country for geo testing; that option adds proxy usage. Old saved inputs that still contain proxyConfiguration can also create a proxy; if no group is set, the account's residential group is used and billed. Remove that legacy key for direct visits.

Will these visits show up in Google Analytics or other analytics tools?

Not necessarily. The actor loads the page in a real browser and records which analytics requests fired, but analytics platforms filter automated traffic and may exclude the visits. Use this actor to test whether tags work, not to inflate audience numbers.

Why did a page report blocked?

The site answered with 401, 403 or 429, or served a recognizable bot-challenge page. A 403 is reported as blocked, a 404 as failed with errorCode=not_found, and 5xx as failed with errorCode=server_error; the HTTP status is retained even when the body is empty. The actor does not solve CAPTCHAs and does not try to bypass the block. Change the target, test a staging URL, or use a country proxy where appropriate.

Why is scrollDepthPercent missing on some successful pages?

The page had no measurable scrollable range, scrolling was turned off, or the depth could not be measured. A page with no range is never forced to 100%, but a very short page with a few pixels of overflow can still report 100% because a wheel step reaches its bottom.

Why is the actual dwell or page count different from what I set?

dwellSeconds is a target, not a hard wall. Waiting for an internal click to navigate can make the measured timeOnPageSec longer, and pagesPerVisit is a maximum: the run's total pageview budget can end a visit before it reaches that maximum. Both the actual dwell and the actual pages are recorded per visit.

Does it click ads or inflate YouTube views?

No. It only attempts visible same-site links; ad elements, forms and external links are skipped. YouTube URLs are treated as ordinary pages, and the actor does not generate views.

What happens if the run is stopped early?

Rows already written stay in the dataset. A graceful stop writes aborted rows for in-flight visits, writes the latest SUMMARY, and sets a partial status such as Stopped 3/20. A hard kill can still omit the in-flight row, but the summary is checkpointed after each page.

Are my saved inputs from the old version still supported?

Yes. An input with only startUrls is accepted and mapped to targetUrls. Legacy waitOnPage, mode=STRESS, enableCrawling/maxPagesPerUrl, enableAdvancedFingerprinting, crawlingLinkSelector and blockResources are applied in code, including when the platform supplies the new defaults. endAfterSeconds and YouTube mode are ignored with a warning; the run stops after the requested pageviews.

Is this affiliated with the websites I test?

No. It is an independent browser automation tool. Use it only on sites you own or have permission to test, and follow their terms and your local laws.

Limitations

• No login, basic-auth, form submission or file upload.
• No CAPTCHA solving or guarantee of bypassing bot protection.
• No ad clicks, ad impressions or YouTube view generation.
• Mobile and tablet profiles use Android Chromium emulation, not iOS Safari.
• Direct visits come from the actor's network; some sites will block them. Country visits depend on residential proxy availability.
• Analytics platforms may filter automated traffic; analyticsTags records requests the page made, not hits the platform accepted.
• Target URLs and redirect destinations are checked against the private/loopback guard before navigation.
• A hard kill can omit the in-flight page row; graceful stops write aborted and a checkpointed summary.
• The actor does not claim to improve SEO rankings or audience metrics.