Checklist

Security baseline for a brand new cloud account

Learn the first settings to lock down in a new cloud account, from identity and logging to network defaults and billing alerts.

By the Cloudverg engineering team · · 5 min read

Who this is for

This checklist, Security baseline for a brand new cloud account, is for the person who just created an AWS account, Azure subscription, Google Cloud project or Oracle Cloud tenancy and wants the first settings to be right. That is often a founder, a CTO or an IT lead who also owns the billing login. Finance and operations leads should read it too, because a few of the choices are about money and contacts, not servers.

The advice assumes you have not put production workloads in yet. If the account is already live, use the same list as an audit and expect to find gaps. The order still works: identity first, then guardrails, then logging, then workload defaults.

The goal is not a perfect security program. It is a small set of defaults that stops the easy failures: a public bucket, a root login with no MFA, logs nobody can read, and a bill nobody owns.

Before you begin

Pick one named owner for the account and one backup. Use a company recovery email on your own domain, not a personal address, and make sure the phone number reaches a monitored desk. Shared mailboxes are fine for alerts, but a person must be accountable for the login.

Have your identity provider ready before you invite anyone. You want single sign on and groups from day one, because retrofitting them after ten admins have personal roles is slow. Decide the naming pattern for accounts, subscriptions or projects and the tags you will require: owner, environment, data class, cost center.

Choose the regions you will allow. Pick a short list, write it down, and treat every new region as a change request. Ask finance for the budget alert threshold and the person who should receive it.

Do not create virtual machines, buckets or databases until the organization or tenant structure is in place. It is much easier to deny public storage at the top than to chase it later.

The checklist

Work from the top down. Identity and organization settings first, then logging, then network and data defaults. If you start with the workload, you will spend the next month fixing exceptions.

Record who completed each item and the date. A baseline that lives only in someone's memory is a wish. Where the provider supports it, put the settings in policy as code so a new account inherits them.

Check inherited settings, not only the local console. Organization policies, management group policies and folder policies can override what you see on the account page.

  • Create the account inside an organization or tenant you control, not as a standalone login, so policies and billing can be managed from one place.
  • Turn on multi-factor authentication for the root user or global administrator and store the hardware key and recovery codes in a sealed company safe.
  • Remove or disable root access keys and long lived console passwords, and use federated roles for daily work.
  • Connect your identity provider and require single sign on for every human user, with group based roles instead of personal admin grants.
  • Set a default deny policy for new resources, then allow only the services and regions the business has approved.
  • Block public access at the account or organization level for object storage and databases, and log any exception with a named owner.
  • Enable control plane audit logging in every region you use and send a copy to a separate log archive account with delete protection.
  • Turn on default encryption for storage, disks and databases, and enforce TLS for data in transit at the load balancer or service endpoint.
  • Create a minimal network baseline with private subnets, no default internet route, and flow logs turned on before the first workload.
  • Set billing alerts and a monthly quota review so a forgotten resource or a leaked key cannot run without someone noticing.
  • Tag every resource with owner, environment, data class and cost center at creation time, and enforce the tags with policy.
  • Record the account owner, security contact and billing contact in the provider console, and test that the addresses reach a monitored inbox.
  • Enable versioning and delete protection on backup and log buckets, and run one restore test before go live.
  • Review the provider hardening guide for the account and note every deviation with a reason.

What good looks like

A new account is in good shape when a developer can create an approved resource without asking, and cannot make a storage bucket public or launch in an unapproved region. The denial should come with a clear message and a path to an exception. Root or global admin sign in should raise an alert that someone reads.

Logs should land in an account that workload admins cannot delete, and the retention period should match the business and customer contracts. Backups should have versioning and delete protection, and one restore should have been tested. Billing alerts should reach a person, not a dead inbox.

Access should be boring. Every human uses single sign on. Admin rights sit in one group with a short list. Service accounts have a named owner and narrow permissions. If you still have long lived keys, they are exceptions with an expiry date.

What to do next

Run the list in order over the first week. Fix anything that exposes data or grants full control before you tune the rest. If you run more than a few accounts, move the repeated items into a landing zone template or an infrastructure as code module so every new account starts the same way.

Put a review on the calendar. Check new regions, new admin grants, public access exceptions and log destinations on a regular schedule. Do the same review before a customer security questionnaire or an audit, because the evidence will be fresh.

If you want an outside read, the Cloud security review from Cloudverg is $6,500/month for a fixed scope of up to five cloud accounts, delivered in two to three weeks. It is read only, covers identity, public exposure, encryption, logging and backups, and ends with a prioritized fix list and a readout. Compare that with doing the checklist yourself or hiring a local consultant.

Keep this checklist next to your account creation runbook. The next account should take less time than this one.

Frequently asked questions

What should I do first after creating a new cloud account?

Secure identity. Turn on MFA for the root or global admin, remove root access keys, connect your identity provider, and create a small admin group. Then enable audit logging and guardrails before you add workloads. This order keeps the first admin actions accountable.

Is a separate account needed for logging?

Yes, for most companies. A log archive account that workload admins cannot delete gives you a copy of audit events even if a production account is compromised. Forward control plane logs there with delete protection and a retention rule. Test that the destination still receives events after a policy change.

How do I check a new cloud account before go live?

Try to break your own baseline. Attempt to make a bucket public, launch in a blocked region, create a user outside SSO, and delete a log object. Each attempt should fail or raise an alert. Record the results with the date and owner.

Monthly plan

Want help with this? Cloud security review

$6,500/month

2 to 3 weeks · Starts within 10 business days

A read-only review of identity, network exposure, encryption and logging across your cloud accounts, with a prioritized fix list.

Renews monthly. Cancel any time.
What is included

Related guides

More cloud security guides   Security and compliance

Smallest way to start

Get a senior architect on your problem this week

Ninety minutes a month on the cloud, AI or security question that is blocking you, a pre-read of your material and a written summary within two business days.

$450/month

Renews monthly. Cancel any time.