GTM Spyby Trackable

Security

Draft, under review. Last updated 15 November 2026.

Agencies connect their clients' Google, Meta and Shopify accounts to GTM Spy. This page says how we look after that access and the data that comes with it. Who does what with personal data is in the data processing agreement, and who helps us is on the sub-processors page.

Read-only by default

  • Google: read access to Google Analytics (analytics.readonly) and Tag Manager (tagmanager.readonly), and Google Ads (adwords: Google offers no read-only Ads permission, and we only read).
  • Meta: read access to ad accounts (ads_read).
  • Shopify (not switched on yet): read access to orders (read_orders). The audit fetches each order's time, test flag, source, channel and totals, keeps the online orders, and compares them with what the ad platforms counted. It never asks for the customer.

Connecting is optional, and needs the account holder's sign-in and consent on the platform's own screen.

Changes in Tag Manager, only when you allow them

GTM Spy can make changes in a client's Google Tag Manager, for example from an AI assistant through your API key. It is off until you turn it on.

  • Per client, off by default. An owner or admin of your team turns on “Allow changes in Tag Manager” for one client. Members can see whether it is on, but can't switch it.
  • Google asks first. The first time it is turned on for a client, Google asks the same login that connected the client for three more permissions: edit containers, edit versions and publish. We never ask to delete containers or to manage Tag Manager accounts or users. Turning it off leaves Google's permission in place, so turning it on again doesn't ask Google again; to withdraw it, disconnect the client or remove GTM Spy from the Google account.
  • Owners and admins only. Only a key belonging to an owner or admin of the team can make a change.
  • Two-factor sign-in. Turning it on needs two-factor sign-in on for the person doing it, and a key makes changes only while its owner has two-factor sign-in on.
  • A dry run first. Every change is first answered as a dry run: the exact call (and, for an update, a delete or a revert, the item as it is now), with nothing changed. The change is made only when the same call is repeated with the dry run's token, within 15 minutes and once. Deleting and publishing also need the item's ID repeated, as a confirmation.
  • On the record. Each use is logged against the connection: who asked, for which tool, and when. Each change, and turning it on or off, is also in the team's activity log (Settings › Team), for its owners and admins.
  • Off at once. Each change asks gtmspy.io for a fresh token, so turning it off stops any further changes straight away.

How we keep access tokens

  • The long-term access Google, Meta or Shopify gives us is encrypted (AES-256-GCM) before it is stored. Each token is sealed to its own record, so it can't be read in place of another. The key is kept as a secret of our hosting, never in the database or the code.
  • Tokens are never shown or sent back, to you or anyone else. Only our scanner receives one, for the job in hand, and it never stores it: for a Google audit, a short-lived token (an hour at most) made from the stored one; for a Meta audit, the stored token itself (Meta's last 60 days). For Tag Manager reads through the API, the scanner keeps the short-lived Google token in memory for at most 5 minutes.
  • Disconnecting a client's account removes the connection at once. If no other client of your team, or of another team, uses the same Google login, Meta user or shop, we also withdraw the access with the platform and delete our copy of the token. If one does, the token stays, sealed, for those connections only.

Signing in

  • Accounts are for work email addresses only.
  • We keep a hash of your password, never the password. Passwords seen in data breaches are refused: only the first 5 characters of the password's SHA-1 hash are sent to Have I Been Pwned to check.
  • Sign-in codes sent by email are stored as a hash, work for 10 minutes and end after 3 wrong tries.
  • Sign-in attempts are rate-limited. Resetting your password signs out every other session.
  • A session lasts 30 days and is renewed once a day while you use it. Signing out ends it.
  • Two-factor sign-in is available: a code from an authenticator app at every sign-in (by password, emailed code or Google), with 10 backup codes that each work once. The app's secret and the codes are stored encrypted.
  • Passkeys are available: sign in with your device's fingerprint, face or PIN. We keep only the passkey's public key.
  • New accounts confirm their email address with a code. You can delete your account yourself, in Settings › Profile.

The scanner's signed identity

Our scanner visits websites as a real Chrome browser and says who it is. Its requests are signed with Web Bot Auth, our public key is at bot.gtmspy.io, and its user agent ends with gtmspy/1.0 (+https://gtmspy.io/scanner). It never fakes a browser, solves CAPTCHAs or rotates IP addresses. More on the scanner page.

Backups and retention

  • Accounts, teams, clients and connections are in a Cloudflare D1 database, which can be restored to any point in the last 30 days (Time Travel). Older states can't be restored, so data you delete is gone from it after 30 days.
  • The scanner's database (audits, results, shares and the rest) is copied every night into Cloudflare R2 storage. Each copy is consistent, taken while the scanner runs, and is deleted after 30 days.
  • The IP address and request counter kept for sign-in rate limits are deleted at the next sign-in by anyone once they are a day old. Audits and their results are kept while your account exists. How long we keep everything else is in the privacy notice.

What we never store or hand out

  • Your password (only its hash).
  • The GTM export files you upload: they are read once and never stored.
  • Platform access tokens, in anything we show or send you: on the website, through the API or by email.
  • The visitors of the websites we audit: an audit records a website's tracking, never its visitors' data.
  • Data downloads: results are read on the website, printed as a PDF, or read by your own tools through your API key. There is no JSON or CSV export.

Reporting a vulnerability

If you think you have found a security problem in GTM Spy, email hello@gtmspy.io with what you found and the steps to reproduce it. Please don't access or change other people's data, and don't run automated or bulk scans against the service while testing. We'll reply, and tell you when it is fixed.