Third party data access
One setting per project decides how much of your audience data anything outside the dashboard actually receives — your own integrations, a connected assistant, or a third-party app. This page is where that setting is explained.
Everything that reaches your project from outside — your own integration with an API token, an assistant you connected, a third-party app — is limited twice. The first limit is what you approved when the connection was created. The second is this page: a setting per project that decides how much of that project's data actually leaves it, whatever anyone was granted.
You'll find it in Settings → Project → Third party data access, and it is called the access type.
The three access types
Full accessdefaultContact data is read the same way you see it in the dashboard: emails, phone numbers, names, your custom attributes and the contents of events. The right choice for your own tools, and for providers you trust like your own team.
AnonymizedIt can still work with your audience, but it cannot tell who anyone is. Counting, segmenting, measuring and sending keep working; emails, phone numbers, names, your custom attributes and event contents are hidden, and it cannot look someone up by email or phone. In short: it keeps working with groups, and stops being able to identify people.
BlockedNo data from this project is served at all, and requests made for it are rejected — reads and writes alike. It does not disconnect anything: an app that is already connected simply stops receiving data from here, and starts receiving it again if you change the setting back.
What it governs, and what it doesn't
This setting governs what is read through the Product API, MCP connections included. It does not affect the Ingest API, the A2P Messaging API, outbound webhooks, or exports and reports, which follow their own permissions.
Two things are never anonymized, because hiding them would break the work without protecting anybody: the filters of your campaigns and segments, and your company configuration.
The permissions underneath
Before the access type applies, a connection only ever holds the permissions it was granted, over the projects it was granted them on. Those permissions cover your audience (contacts, attributes, events and segments), your campaigns and automations and how they performed, your data sources, and your messages — each one readable, and some of them writable, depending on what was approved.
How the permissions are approved differs by what is connecting: an assistant connected over MCP takes them as a package, while a third-party app shows them as a list you can trim before approving. Both are covered where they belong — Connect your AI and Third-party apps.
Verified apps
Instasent reviews and verifies applications, and every connection shows whether its app is verified. When you authorize one that isn't, you are warned and asked to confirm explicitly before continuing.
A project can also be set to only allow verified apps, and this is worth reading carefully:
Where this sits in the bigger picture
The access type is about what leaves your project. What a contact has agreed to receive is a different question, answered by your subscription and consent model — see Consent. And the record of which connections exist at all, with what permissions, is Connected apps.
Building your own app on Instasent? The technical contract for OAuth, tokens and scopes is in Connected apps (OAuth).