Website Change Monitor & Diff Alerts

Watch any list of public pages — competitor pricing, terms and conditions, public price lists, documentation, job pages, regulatory notices — and get a precise answer to one question every run: what changed since last time? The Actor stores a normalized text snapshot of every page in a named key-value store, compares the current page with it and reports NEW, UNCHANGED, MINOR, CHANGED or ERROR together with the added and removed lines.

What it does

• Fetches each URL with timeouts, retries and polite per-host delays; no browser needed for server-rendered pages.
• Extracts the main content: scripts, styles, navigation, header, footer, sidebars and cookie banners are removed automatically, or you pass a CSS selector such as main or #pricing.
• Normalizes whitespace and removes volatile fragments you define as regular expressions (dates, counters, timestamps) so they never trigger false alerts.
• Compares against the previous snapshot with a line diff and classifies the result; MINOR versus CHANGED is controlled by a configurable threshold.
• Saves the new snapshot, pushes one dataset item per page and writes a Markdown REPORT with changed pages first, including the added and removed lines.
• Works with Apify schedules and integrations: run it hourly or daily and connect the dataset to Slack, e-mail, Make, Zapier or a webhook; filter on status.

Input

Field	Type	Default	Description
urls	array of strings	required	Pages to watch (max 2000 per run).
selector	string	""	CSS selector to scope the comparison.
ignorePatterns	array of strings	[]	Regular expressions removed before comparing (dates, counters).
minorThreshold	number	0.02	Changes below this share of lines are MINOR.
includeDiff	boolean	true	Include up to 20 added / 20 removed lines per changed page.
emitUnchanged	boolean	true	When false, unchanged pages are not pushed (and not billed).
stateStoreName	string	website-change-monitor-snapshots	Named key-value store holding snapshots (created by the Actor on the first run); one per schedule. If the store is not accessible under the Actor's limited permissions, the run uses its default store and reports state.mode = default-fallback in OUTPUT.
maxItems	integer	200	Hard cap on checked pages per run.

Output

One dataset item per page: status, changedRatio, added, removed, addedSample, removedSample, title, titleChanged, wordCount, textHash, previousHash, previousCheckedAt, httpStatus and error for failed fetches. The overview view lists changes first. The key-value store holds REPORT (Markdown) and OUTPUT with counts per status.

Example item (shortened):

{
  "finalUrl": "https://competitor.example/pricing",
  "status": "CHANGED",
  "changedRatio": 0.18,
  "added": 2,
  "removed": 1,
  "addedSample": ["Pro: 35 EUR / month", "Annual plans available"],
  "removedSample": ["Pro: 30 EUR / month"],
  "previousCheckedAt": "2026-10-20T06:00:03.120Z"
}

Pricing and cost control

Pay per event: Page checked (page-checked) — one event per page fetched in a run, including unchanged pages (set emitUnchanged to false to pay only for pages with news). maxItems caps the number of pages per run. A daily check of 100 pages costs 100 events per day and runs in about a minute at 256 MB.

Limitations and fair use

• Server-rendered HTML only; content injected by JavaScript after load is not seen. Pages that are entirely client-rendered produce a low wordCount so you can spot them.
• Snapshots are capped at 200 KB of text per page; larger pages are compared on the first 200 KB.
• Only publicly reachable pages: no logins, paywalls or CAPTCHA solving, and robots directives are respected by not crawling beyond the URLs you list.
• The diff is textual; image or layout changes are not detected.

Example

Input: three public pages with date patterns ignored. First run: three NEW items (snapshots created). Next run: UNCHANGED for untouched pages and CHANGED with the exact lines for a page whose price list moved. See examples/input.json and examples/output.json.

Use cases

• Competitive intelligence: pricing, plans, feature pages, campaign landing pages.
• Compliance: terms of service, privacy policies, supplier conditions, insurance terms.
• Operations: public documentation and API changelogs that break integrations.
• Sales: career pages and "news" pages of target accounts as buying signals.