Terms of service and privacy policy change monitoring by market

Document URL and type

Terms, privacy policy, cookie notice, acceptable use, pricing, subprocessor list. Changes mean different things in each.

Market and language

Which regional version you fetched, and from where. The same service can change its EU terms and leave its US terms alone.

Extracted main text

The policy content without navigation or banners. This is what you diff.

Content hash

A hash of the normalised text. Equal hashes mean no change and no need to store a new version.

Stated effective date

The 'last updated' or 'effective from' line, stored separately from your own fetch date, because the two diverge.

Raw HTML and screenshot

The evidence copy. If a change is disputed, this is what shows how the page looked.

Section-level diff

Which clauses were added, removed or reworded. Reviewers read the diff, not the whole document.

Prices, currency and tax display

For pricing pages: plan, price, billing period and whether tax is included, per market.

1. 01

   Inventory the documents per service and market

   For each service you track, list every legal and pricing page and every regional variant. Follow links from the main policy to annexes and include them.

2. 02

   Fetch on a schedule, from the right country

   Check each document once or twice a day through an exit in its market, with conditional requests. Pricing pages that render by script go to a headless browser on a sticky exit.

3. 03

   Normalise, hash and store only real changes

   Extract the main text, strip dates, tokens and banners, collapse whitespace and hash. Store a new version only when the hash changes.

   Fetch, normalise and hash a policy page24 linescopy

   import hashlibimport reimport httpxfrom selectolax.parser import HTMLParserGATEWAY, PASSWORD = "proxy.arcproxies.io:8000", "PASSWORD"NOISE = re.compile(r"\b(csrf|nonce|session)[^\s]*|\d{1,2}:\d{2}(:\d{2})?", re.I)def snapshot(url: str, country: str, etag: str | None = None):    user = f"USERNAME_pool-arc_country-{country}"    headers = {"If-None-Match": etag} if etag else {}    with httpx.Client(proxy=f"http://{user}:{PASSWORD}@{GATEWAY}", timeout=30,                      follow_redirects=True) as c:        r = c.get(url, headers=headers)    if r.status_code == 304:        return None  # unchanged since the last check    tree = HTMLParser(r.text)    for sel in ("nav", "header", "footer", "script", "style", "[id*=cookie]"):        for node in tree.css(sel):            node.decompose()    main = tree.css_first("main") or tree.body    text = NOISE.sub("", " ".join(main.text(separator=" ").split()))    return {"final_url": str(r.url), "etag": r.headers.get("etag"),            "sha256": hashlib.sha256(text.encode()).hexdigest(), "text": text}

4. 04

   Diff by section and route the alert

   Split the text on headings and diff section by section. Send the changed sections, with a link to both versions, to the reviewer who owns that service or market.

5. 05

   Keep the archive

   Retain every version with raw HTML, screenshot, fetch time, exit location and hash. The archive is what turns an alert into evidence.

Prices are list prices in EUR, excl. VAT. Volume discounts apply automatically at checkout — full ladder on the pricing page.