
Compliance and security in Braze campaigns come from controls you set before anyone builds a message: sign-in rules, role design, approvals, personal data visibility and API key scope. Braze documents each control, and each one can be configured badly. This guide sets out what the documentation says and the order in which CustomerIK would apply it.
Key Takeaways
- Settle sign-in before roles: enforcing two-factor authentication lifts the maximum session length from 1,440 minutes to 10,080 minutes.
- The approval workflow is off by default, and the same permission can let a person self-approve, so give build and approve permissions to different people.
- Any change after approval resets every approval status to pending, so finish every edit, save, and only then approve.
- Hiding personal data fields changes what users can view on the dashboard and does not change how the end user data is handled.
- API key scope and allowlisted addresses cannot be edited after creation, so issue one key per integration.
- The manual security report holds the latest 10,000 events, so set up the cloud storage export before you need it.
What does compliance and security in Braze campaigns actually cover?
Braze gives you controls in five places. They are sign-in rules, permissions, the campaign approval workflow, personal data visibility, and API keys. Each lives in a different part of the dashboard, so no single screen shows whether you are covered.
This post does not settle any legal question. Which rules apply to your sends depends on your business, your audience and where they live. What follows is how the product controls work, and how a Braze implementation partner would sequence them.
How should you lock down dashboard sign-in first?
Start here, because every other control assumes the right people are the ones logged in. The Security Settings page covers password rules, session length, single sign-on, address allowlisting and two-factor authentication. The default minimum password length is eight characters, and you can raise it.
We recommend settling sign-in before touching permissions. A strong role design is worthless if a shared login or a stolen password bypasses it.
Session length has a documented ceiling that depends on a choice you make. The maximum is 10,080 minutes if two-factor authentication is enforced for your company, otherwise it is 1,440 minutes. Enforcing two-factor authentication therefore buys longer sessions as well as a second check.
| Control | What the documentation says | Where it bites |
|---|---|---|
| Dashboard IP allowlisting | You can restrict log-in to listed addresses and subnets. If left empty, users can log in from any IP address. | An empty field is the default, so nothing is restricted until you act. |
| Two-factor authentication | If enforced company-wide, users who skip setup are locked out of the dashboard. | Run the rollout when an administrator is free to reset people. |
| SSO | 2FA is required for all company users only if SSO isn't enabled. | Choose one route on purpose, not by accident. |
| Remember this account | A checkbox that appears once 2FA is on, valid for 30 days. | Fine on a managed laptop, risky on a shared one. |
| Elevated Access | Users re-verify before exporting a segment or viewing an API key. | Re-verification lasts about an hour. |
Elevated Access suits the two actions that move data or credentials out of the dashboard. After a user re-verifies, the documentation says they won't need to do so again for the next hour. We suggest turning it on early, because it costs your marketers one extra prompt.
Who should be able to build, approve and launch a campaign?
Braze separates what people can do in three layers. Permission sets control what company users can do across all workspaces. Roles control what they can do in specific workspaces, and teams control the audiences they can reach with their messages.
The approval workflow is off by default for campaigns and Canvases, and you turn it on under Settings, then Approval Workflow. Once it is on, a user with the right permission must approve each section of the summary before the message can launch. Approval does not apply to API campaigns or transactional email campaigns, so those need a different safeguard.
The same permission can let a person self-approve their own campaign, so switching the workflow on does not give you a second pair of eyes. We recommend giving the build permission and the approve permission to different people, and writing down who they are.
Any change after approval resets every section to pending. Per the docs, subsequent changes to a campaign reset all approval statuses when saved. The sequencing rule is simple: finish every edit, save, and only then approve.
How should you handle personal data in the dashboard?
Braze lets you designate standard and custom attributes as personal data and hide them from users without the View PII permission. By default, all admins have that permission turned on. So the first audit question is how many people hold it, and whether the answer matches your intent.
Read the scope carefully. Defining fields as personal data only changes what users can view on the dashboard, and does not change how the end user data in those fields is handled. Treat it as a visibility control, and decide separately how the data itself is stored and retained.
There is also a preview exception. When previewing a message, the View PII permission isn't applied, so users can see redacted attributes if the message references them through Liquid. A person who can build campaigns can therefore see values that a profile search hides.
If the harder question is which attributes should exist at all, CustomerIK's customer data management service covers identity resolution, CDP integration and data governance.
What happens when you delete a user?
When a user is deleted, Braze removes all personal data from event data but keeps the anonymised data for analytics. You then choose which user-defined fields, such as custom event properties, count as personal data and should be removed too. The documentation says you are responsible for determining the correct preferences for your workspace.
Take that sentence seriously, because the action is final. After Braze applies the settings to historical event data for deleted users, you cannot restore the data. Agree the field list with whoever owns each data source before you save it.
Permission to delete users is its own permission, described as permanently deleting users from the dashboard individually or in bulk. We suggest holding it back from anyone who does not need it.
How do you protect API keys?
A server-side API key can send messages, export user data and modify account settings if it is exposed. Braze says to revoke it immediately and create a replacement. The SDK key is different: it can only ingest data and retrieve client-side messaging, and Braze describes it as designed to be public.
The control where you have the greatest say is scope. Keys can be individually scoped to specific endpoints, and after you create one you cannot edit the permissions or the allowlisted IPs. That shapes how you should work.
- List each integration that calls Braze, and the endpoints each one needs.
- Create one key per integration with only those endpoints and a named address range.
- Store each key in the secrets manager your engineers already use, never in a ticket or chat.
- To change scope, create a new key, switch the integration over, then delete the old key.
One key per integration means a leak costs you one integration instead of all of them. A 403 response is the documented signal that a key lacks permission for an endpoint, so a scoped key shows up as an error at test time instead of passing with too much access.
Where do you look when something goes wrong?
Braze offers a Security Event report covering invitations, removals, failed and successful logins and other account activity. The manual download contains only the latest 10,000 security events. If your export has exactly 10,001 rows including the header, you have hit the cap and older events may be missing.
For longer history, Braze documents exporting security events to cloud storage without that row limit. Set that up before you need it. An audit that starts after an incident is the wrong moment to learn the cap exists.
What does this look like for a hypothetical retailer?
Imagine a retailer on one production workspace, with a small in-house team and an outside agency that builds Canvases. This is an illustration, not client work, and it carries no figures.
In the first week, the administrator enforces two-factor authentication and restricts dashboard log-in to the office and the agency's fixed addresses. In the second, they build two roles, a builder and an approver, and give no one both.
In the third, they turn on the approval workflow and list which journeys are API-triggered, since those sit outside it. In the fourth, they issue one scoped key per integration and retire the shared key everyone had been using.
The order matters because each step makes the next one safer. Locking sign-in first means the roles describe real people. Roles come before approvals so the workflow has someone to route to.
Keys come last because they need the closest coordination with engineering.
How does this connect to the rest of the programme?
Governance rarely lives in one post. If you are adding AI features, see Governing Braze AI: Permissions, Limits and Approvals and Agentic Standards: Automated Campaign QA in Braze. Both extend the permission thinking here into newer features.
Personalisation brings its own exposure. Read Using Braze Connected Content for Personalized Campaigns before you let a message call an outside service, and Integrating CRM Data into Braze Marketing Automation before you decide which attributes arrive at all.
For build discipline, see Braze Onboarding Best Practices: Entry, Exit and QA and Using Templates for Faster Marketing Automation in Braze. For approval workflows alongside data unification in a hospitality programme, CustomerIK's guest profile unification case study covers how approvals were embedded in the Braze interface.
What does CustomerIK recommend?
As a Braze implementation partner, CustomerIK recommends treating this as a four-week sequence: sign-in, roles, approvals, then keys. Keep a short written record of who holds each sensitive permission, and review it whenever a person joins or leaves. Re-check the documentation before you act, because limits and defaults can change.
Frequently Asked Questions
1. Does turning on personal data hiding protect the data itself?
It does not. Braze says it affects only what users can view on the dashboard and does not change how the end user data is handled.
2. Can I edit an API key's permissions after creating it?
You cannot. You cannot edit the scope of permissions or the allowlisted IPs, so create a new key with the scope you want and retire the old one.
3. Is the SDK key a secret?
Braze describes it as designed to be public. It can only ingest data and retrieve client-side messaging, so the exposure risk sits with the server-side key.
4. How far back does the manual security report go?
It contains only the latest 10,000 security events. Use the cloud storage export for longer history.
5. Does this post tell me which regulations apply to my campaigns?
It does not. It describes product controls and does not state any legal conclusion.
Sources
- Braze, Security settings. Retrieved 11 Nov 2026.
- Braze, Braze permissions. Retrieved 11 Nov 2026.
- Braze, Approvals for campaigns and Canvases. Retrieved 11 Nov 2026.
- Braze, API overview. Retrieved 11 Nov 2026.
- Braze, APIs and identifiers. Retrieved 11 Nov 2026.
- Controls, limits and defaults described here can change after the retrieval date, so check the Braze documentation before you act.
- Written by CustomerIK, a Braze implementation partner.
CustomerIK is a Braze implementation partner working across onboarding, technical integration, marketing operations and customer data management.
If you want help sequencing sign-in, roles, approvals and API keys for your Braze workspace, let's talk.




