Website Screenshot & Visual Change Monitor — Full Page, Mobile ($2 per 1,000)

Paste a list of web addresses and get one screenshot per page: full page or a single element, on desktop, laptop, wide, tablet or mobile, as PNG or JPEG, with cookie banners hidden (nothing is clicked or accepted). Each dataset row links to its image and also tells you the page's title, HTTP status, final address after redirects and load time.

Run it again — by hand or on a schedule — and every page is compared with its capture from the last run: the row says new, changed or unchanged, how much of the page changed (%), and links a diff image with the changed pixels in red. Turn on Return only new and changed pages and a daily run returns (and charges) only the pages that actually changed.

Try it now — the input is already filled in. Press Start: the two example pages come back as full-page PNGs in about 25 seconds, for $0.014 (measured on 28 Oct 2026). Press Start again and each row says whether the page changed. Three rows from our test runs that day (Wikipedia added):

url	changeStatus (1st run → 2nd run)	changedPercent	imageWidth × imageHeight	bytes	loadSeconds
https://example.com/	new → unchanged	0	1280 × 800	72,616	0.2
https://apify.com/	new → unchanged	0	1280 × 10,036	1,026,699	1.1
https://www.wikipedia.org/	new → changed	0.12	1280 × 1,103	246,438	0.5

screenshotUrl opens the image directly (a signed link, no login needed); a changed page also gets diffUrl, the same page with the changed pixels in red.

What you get

Field	What it is
url, finalUrl, httpStatus	the address you gave, where it ended up, and the status it answered
screenshotUrl	link to the image (PNG or JPEG) in the run's key-value store
title	the page's title
changeStatus	new (first capture), changed or unchanged since the last run; null with comparison off
changedPercent	share of pixels that differ from the last capture (0–100)
diffUrl	link to the diff image (changed pixels in red), only when the page changed
sizeChanged	true when the page got taller, shorter or wider
previousScreenshotAt	when the capture it was compared with was taken
imageWidth, imageHeight, bytes, format	the image itself
device, viewport, fullPage, selector	how it was captured
truncated	true when the page was taller than 16,000 pixels and was cut there
loadSeconds, loadTimedOut	how long the page took; true when the timeout ran out and what was drawn was captured
contentHash	a fingerprint of the image (same hash = identical image)

Pages that could not be captured come back as free error rows with an errorKind (not-found, server-error, bot-check, blocked-by-robots, selector-not-found, network …) and a plain-English message. A free summary row at the end counts screenshots, new, changed, unchanged and failed.

Use it for

• Competitor and pricing page watch — a daily capture of rivals' pricing and landing pages; changedOnly returns just the ones that changed, with the diff image showing where
• Your own site after a deploy — catch a broken layout on desktop and mobile before customers do
• Defacement and compliance checks — proof of what a page looked like on a given day, with a fingerprint
• Thumbnails and archives — bulk captures of a list of sites for a directory, a report or a deck

How the comparison works

• The capture is kept in a key-value store named website-screenshot-monitor-state in your own Apify account, keyed by the address, device, size and selector. Change any of these and that page counts as new once.
• Identical images are unchanged at once (same fingerprint). Otherwise the two images are compared pixel by pixel; a taller or shorter page counts the extra part as changed.
• Pages with a clock, a rotating banner or ads change a little every time: raise Change threshold (for example to 2%) or hide those parts with Also hide these elements.
• One capture per page is stored, replaced on each run. Delete the store to start over.

What it does not do

• It opens public pages only: no login, no cookies, no headers. Local and private network addresses are refused.
• It reads each site's robots.txt and skips pages it does not allow (a free error row says so).
• It does not get around bot checks (Cloudflare challenges, CAPTCHAs): such pages come back as a free bot-check error row instead of a picture of the check.

Price

$2.00 per 1,000 screenshots, plus $0.01 per run. Error rows, the summary row and — with Return only new and changed pages — unchanged pages are free. The $0.01 run-start fee is the one charge every run pays, even one that returns nothing.

Input example

{
  "urls": ["https://apify.com", "https://example.com"],
  "device": "mobile",
  "fullPage": true,
  "changedOnly": true
}

Returns one row per page with screenshotUrl, changeStatus and changedPercent. Rows have type: screenshot, error or summary.