Wayback Tracking ID History

Find out which analytics, ad-pixel and tag-manager IDs a website used in the past, and when it added or dropped each one. The actor reads archived copies of a site's home page from the Internet Archive (Wayback Machine), takes the IDs out of each copy and gives you one row per website with a timeline for every ID. It looks for 53 kinds of tool; the test runs returned an ID for 25 of them: Google Tag Manager (GTM-), Google Analytics 4 (G-), Universal Analytics (UA-), Google tag (GT-), Google Ads (AW-), Meta Pixel, LinkedIn Insight, Pinterest, X and Reddit pixels, Microsoft Advertising UET, Hotjar, Mixpanel, Heap, PostHog, HubSpot, Klaviyo, Intercom, LiveChat, OneTrust, CookieYes, Optimizely, Sentry, Stripe and Tealium.

Who it is for

• OSINT and fraud researchers who link sites through a shared old analytics or ad account: a former owner may have left a UA- or GTM- ID in the page.
• Agencies and sales teams: when did this company switch tag manager, drop Universal Analytics, or start a new ad pixel?
• Buyers of expired or acquired domains who want to see which accounts the site belonged to before.

What you get

One dataset row per website:

Field	Meaning
history	one entry per ID (a tool found without an ID, such as OneTrust or Stripe.js, has id: null; 4 of 90 entries in our 20-site test): tool, id, firstSeen, lastSeen, status (current or removed), snapshotsSeen, lastSnapshotWithoutIt (the added-after date: the last copy that did not have it) and firstSnapshotWithoutIt (the removed-by date: the first copy without it), plus firstSeenArchive and lastSeenArchive, the archive links you can open to check
changes	per archived copy: IDs added and removed compared with the previous copy
snapshots	the IDs found in each copy
firstSnapshot, lastSnapshot, snapshotsRead	the span and number of copies read

Example (shortened, allbirds.com, 8 years, 2 copies a year):

{"domain": "allbirds.com", "snapshotsRead": 16, "firstSnapshot": "2019-01-01", "lastSnapshot": "2026-07-01",
 "history": [
  {"tool": "Universal Analytics (legacy)", "id": "UA-73307718-1", "firstSeen": "2019-01-01", "lastSeen": "2022-01-01",
   "status": "removed", "snapshotsSeen": 7, "firstSnapshotWithoutIt": "2022-07-01",
   "firstSeenArchive": "https://web.archive.org/web/20190101.../http://allbirds.com/"},
  {"tool": "Google Tag Manager", "id": "GTM-NJVZ6WQ", "firstSeen": "2019-07-02", "lastSeen": "2024-01-02",
   "status": "removed", "snapshotsSeen": 10, "lastSnapshotWithoutIt": "2019-01-01", "firstSnapshotWithoutIt": "2024-07-02"}],
 "changes": [{"snapshot": "2019-07-02", "added": ["gtm:GTM-NJVZ6WQ"], "removed": ["gtm:GTM-M55BQF2", "gtm:GTM-W4HDTKM"]}]}

Input

{"domains": ["allbirds.com", "nytimes.com"], "years": 8, "snapshotsPerYear": 2}

• years (1-25, default 8) and snapshotsPerYear (1, 2, 3, 4, 6 or 12, default 2; any other number is read as 2): how many archived copies are read. More copies narrow the date range of every change and make the run longer.
• concurrency (default 2) and maxDomains (default 200).

Price

$0.02 per website with at least one readable archived copy (event domain-analyzed). A website with no readable copy, or an invalid domain, is not charged. A website with copies but no IDs in them is charged: the copies were read.

What it can and cannot see (read this)

• Home page HTML only. An ID is found when it is written in the HTML of the archived home page. IDs that a site loads from a JavaScript file or from inside a Google Tag Manager container are not in the page and are not returned: the Internet Archive does not keep those containers reliably. So status: removed means "no longer in the page HTML", and the ID may have moved into a tag manager.
• Sampled, not complete. The actor reads evenly spaced copies, not every copy. A change is dated to the range between two copies (lastSnapshotWithoutIt to firstSeen), not to the day.
• Not every site is archived. Sites with no copy in the chosen years get no row data and no charge.
• Speed. The Internet Archive refuses clients that send many requests, so the actor sends about one request every 1.5 seconds and waits and retries when it is throttled (429, 5xx, connection errors). One website with 2 copies a year over 8 years took 35 seconds; a run of 20 websites took 593 seconds (about 10 minutes) in our test, one of them excluded from the archive. A site the archive refuses (excluded, blocked: HTTP 400, 401, 403, 410, 451) is not retried: that date is skipped in seconds, and a site where every date is refused is returned as "no readable copy" without a charge.
• Only data the Internet Archive publishes is used. No login, no proxy, no browser.

Not the same as

Actors that list Wayback snapshots (dates, status codes, digests) leave you to download and read every page yourself. Actors that read BuiltWith give tool names and first/last detected dates from BuiltWith's own data, not the account IDs found in the archived page. This actor returns the IDs, with the archive link of the copy where each was first and last seen so you can check it.