Skip to content
Velloreth
Cybersecurity 11 min read

Zero Trust Explained for Small Teams

Zero trust is not a product you buy. It is a set of access decisions you make per request. Here is how small teams can implement it without a security department.

By

Zero trust is often sold as a product category. It is not. It is a set of access decisions: authenticate the user, check the device, authorize the specific resource, and re-evaluate as conditions change. The term comes from a 2010 Forrester research report, and the reference architecture most people cite is NIST SP 800-207, published in 2020.

For a five-person team, that definition sounds expensive. It is not. Most of the work is configuration on tools you already pay for: your identity provider, your SaaS apps, and your cloud accounts. This post walks through what zero trust actually changes, what a small team should implement first, and what you can safely ignore.

The problem zero trust solves

The traditional security model is perimeter-based. You put a firewall between the office network and the internet, run a VPN for remote workers, and assume anything inside the perimeter is trustworthy. That model had one fatal assumption: trust based on network location.

Three trends broke it:

  1. Applications moved out of the office. Your code is on GitHub, your files in Google Drive, your CRM in a browser tab. None of that traffic goes through your office firewall.
  2. People stopped working from the office. A laptop on a cafe WiFi network is now a normal endpoint.
  3. Attacks target identity, not the firewall. Phishing, credential stuffing, and session hijacking all get past a network perimeter because the attacker logs in like a legitimate user.

Google learned this the hard way after Operation Aurora in 2009, a sophisticated attack that reached internal systems. Their response was BeyondCorp: an internal initiative started in 2011 to let every employee work from untrusted networks without a VPN, with access decisions based on the user and the device rather than the network. Google has run this model internally for over a decade.

The same logic applies to a team of five. If an attacker steals a password for your AWS account, your firewall does nothing. If your contractor keeps access to your production database three months after their contract ends, no perimeter notice will catch it.

The NIST tenets, translated

NIST SP 800-207 defines seven tenets of zero trust. Here they are, compressed and translated for a small team:

NIST tenetWhat it means for you
All data sources and services are resourcesTreat your SaaS apps, cloud accounts, and code repos as resources — not just servers
Trust is evaluated per-requestRe-check identity and device posture on each session, not once at login
Access is granted with least privilegeGive people the minimum role that lets them do their job
All traffic is logged and inspectedKeep audit logs on admin actions in every system
The enterprise network is not trustworthyAssume an attacker is already inside; verify everything anyway
Device posture is part of the decisionOnly managed, patched devices get access to sensitive resources
Authentication and authorization are dynamicA risk signal (new country, unmanaged device) should force re-authentication

Notice what is absent: firewalls, network segmentation, and appliance purchases. The tenets are about decisions, not hardware.

What to implement first, in order

A small team cannot do everything at once. Here is the order that gives you the most risk reduction per hour spent.

1. MFA on everything, enforced

This is the single highest-value step. Multi-factor authentication blocks the vast majority of automated credential attacks. The key word is enforced — an optional MFA prompt that users can dismiss is decorative. In your identity provider and every critical SaaS app, make MFA mandatory, and prefer app-based authenticators or hardware security keys over SMS.

Check your SaaS apps individually. Many teams enable MFA in Google Workspace or Microsoft 365 and forget that their payment processor, code host, and CRM each have their own setting.

2. Centralize identity with SSO

If every service has its own username and password, you have no way to revoke access quickly and no single place to enforce policy. Move everything possible behind your identity provider — Google Workspace, Microsoft Entra ID, Okta, or a cheaper option like JumpCloud or Cloudflare Zero Trust Access.

The practical win is offboarding. When someone leaves, disabling one account should disable access to everything. If you have to hunt through thirty SaaS admin panels, you will miss something, and that something will be the thing that matters.

3. Apply least privilege to every role

Most small teams over-permission by default: everyone is an admin everywhere because it was convenient during setup. Spend an afternoon demoting.

SystemCommon defaultBetter
AWS / GCPAdministratorAccess for all usersScoped roles per function (read-only, deploy-only, billing)
GitHubAdmin on every repoWrite access only on owned repos; read on the rest
Google WorkspaceSuper admin for the whole teamAdmin for one or two people; standard users otherwise
Payment tools (Stripe, etc.)Full access for everyoneView-only for most; write for the owner
Production databaseShared read-write loginIndividual accounts; read-only where possible

Rotate shared credentials as you find them. A shared root password for anything is a zero trust violation on its own.

4. Check device posture before granting access

Zero trust treats the device as part of the identity. You do not need a full MDM deployment to get most of this benefit. For a small team:

  • Require a screen lock and disk encryption on every laptop (both are default on modern macOS and Windows).
  • Turn on automatic updates for the OS and browser.
  • For cloud admin access, use your identity provider’s device-trust features if available — for example, requiring a managed or compliant device for admin roles.

If you are on Google Workspace or Microsoft 365, both have conditional access policies that can require a compliant device or block access from risky sign-ins. Cloudflare Zero Trust and Tailscale offer similar checks for self-hosted apps at low or no cost.

5. Replace the broad VPN with scoped access

A traditional VPN grants a device access to an entire network. That is the opposite of least privilege. Zero trust says: expose the application, not the network.

For web apps, use an identity-aware proxy. Cloudflare Access (free for up to 50 users), Google’s IAP, and Teleport all sit in front of an app and check identity on every request. Your internal Grafana or admin panel becomes reachable by authenticated users only — no subnet access, no VPN client.

For SSH or database access that is not HTTP, tools like Tailscale, Teleport, or a bastion host with short-lived certificates give you the same scoping without a full-tunnel VPN.

6. Keep audit logs

Zero trust without logs is just trust with extra steps. You need to be able to answer: who accessed what, from where, and when. Most SaaS apps and cloud providers have audit logs; the work is turning them on and occasionally reading them. At minimum, keep sign-in and admin-change logs for your identity provider, cloud accounts, and code host, and set alerts for admin-role changes.

What you can skip

Small teams get paralyzed by zero trust marketing. Here is what you can safely ignore for now:

  • Network segmentation. If your team has no on-premises network, there is nothing to segment.
  • Microsegmentation products. Same reason.
  • Full EDR/XDR suites. For a team of five on modern laptops with auto-update and disk encryption, MFA and scoped access give you more protection per dollar. Revisit when you hit twenty people.
  • A formal zero trust maturity model. Those are for enterprises justifying budgets. You do not need a scorecard; you need MFA enforced and access revoked on time.

A realistic one-week plan

If you want to move this week, here is a concrete sequence:

Day 1: Turn on enforced MFA in your identity provider and every SaaS app with an admin role. Use an authenticator app or security key, not SMS where avoidable.

Day 2: Inventory every account each team member has. A spreadsheet is fine. Mark which ones have MFA and which role they hold.

Day 3: Demote unnecessary admin roles. Rotate any shared passwords you find.

Day 4: Move your internal web apps behind an identity-aware proxy (Cloudflare Access is the fastest path).

Day 5: Enable device encryption and auto-update checks; write one paragraph in your handbook saying what is required.

Day 6: Turn on audit logging in your cloud provider and identity provider. Set an alert for admin-role changes.

Day 7: Test the offboarding flow. Disable a test account and confirm it loses access to everything. Fix whatever still works.

That is zero trust. No product purchase, no six-month project — a week of focused configuration that removes the most common ways small teams actually get breached.

Zero trust on the stack you already have

The theory is abstract; the implementation is concrete. Here is what the model maps to on the three stacks most small teams actually run.

Google Workspace

  • Enforce 2-Step Verification at the org level, with security keys or passkeys for admin accounts.
  • Turn on context-aware access in your Cloud Identity or Google Cloud setup: require a managed device or a specific country for admin consoles.
  • Use Advanced Protection Program for the accounts that matter most — it restricts app-specific passwords and enforces hardware keys.
  • Review connected third-party apps with access to Workspace data; revoke anything unused. OAuth grants outlive employee tenure unless you check.

Microsoft 365

  • Security defaults or Conditional Access: require MFA for all users, block legacy authentication protocols entirely.
  • CISA’s SCuBA project publishes free secure configuration baselines for Microsoft 365 and Google Workspace, and ships assessment tools (ScubaGear for M365, ScubaGoggles for Google Workspace) that check your tenant against the baselines and generate a report. For a small team without a security specialist, running ScubaGear once and fixing the red items is the highest-leverage hour you will spend.
  • Disable Basic Authentication remnants and per-mailbox POP/IMAP access unless someone genuinely needs it.

AWS / GCP / Azure

  • No root or owner logins for daily work. Create individual users or federate through your identity provider.
  • SCPs (AWS) or organization policies (GCP) that prevent disabling CloudTrail or leaving S3 public, even for an admin.
  • Short-lived credentials: IAM Identity Center, workload identity federation, or OIDC from your CI provider instead of long-lived access keys.
  • Budget alerts and cost anomaly detection on from day one — spend anomalies are often the first visible sign of a compromised key.

Domain and DNS

  • Registrar lock on your domain.
  • DNS hosted where it can be changed only by the people who need it, with MFA on that account.
  • If you use Cloudflare or similar, zone-level audit logs turned on.

How to buy a product, when you eventually do

Small teams often ask which zero trust product to buy. The honest answer is that you should not start there. When you outgrow pure configuration, the categories worth evaluating are:

CategoryWhat it doesExamples
Identity-aware proxyReplaces VPN for web apps; checks identity per requestCloudflare Access, Google IAP, Teleport
Device trustVerifies device health before granting accessBuilt into Entra ID and Google; standalone from CrowdStrike, etc.
Secret managementShort-lived credentials instead of static keysVault, Infisical, cloud-native KMS
Access management for infrastructureBrokered SSH/Kubernetes access with auditTeleport, Pomerium, Teleport
Email authenticationDMARC/DKIM/SPF enforcement and reportingEasyDMARC, dmarcian, or native tools

Notice what is not on the list: a “zero trust platform” that claims to do all of it. No small team needs that. Buy the one category that solves the access path you actually have, and revisit in six months.

Where teams get stuck

Three patterns show up repeatedly:

“Our developer needs admin access.” They probably need deploy access, not account-admin access. Split the roles.

“We tried conditional access and locked people out.” Roll it out per-app, not tenant-wide, and keep a break-glass admin account with a hardware key that bypasses the policy.

“Zero trust slowed us down.” If it did, the policies were too coarse. The goal is to make secure access the path of least resistance — SSO should be easier than a shared password, and scoped access should be easier than a VPN.

Zero trust will not stop a determined nation-state attacker. What it does is remove the cheap wins: stolen passwords, forgotten contractor accounts, and lateral movement through a flat network. For a small team, that is most of the realistic risk, and it is fixable without a security department.

Frequently asked questions

Does a small team need a zero trust product?
Usually no. Most of the zero trust gains for small teams come from configuring the identity provider, SaaS apps, and cloud accounts you already have — SSO, enforced MFA, least-privilege roles, and conditional access. Products add value later, once those basics are in place.
Do we still need a VPN if we adopt zero trust?
For web apps, an identity-aware proxy or SaaS SSO usually replaces the VPN. You may still need a VPN for non-HTTP protocols like SSH to a private database, but you can scope it narrowly instead of exposing a whole subnet.
What is the first thing to implement?
Turn on MFA everywhere, then move every account into a single identity provider with SSO. That one step cuts off the largest class of attacks — credential theft — and gives you one place to revoke access when someone leaves.
Is zero trust only for remote work?
No, but remote work is what made it urgent. Once your apps live in SaaS and your team works from cafes and homes, the office network perimeter stopped existing. Zero trust is simply admitting that and designing access control around it.