Skip to content

Blog

User provisioning: what it is and how to do it

User provisioning is creating accounts and setting access when someone joins or changes role. Manual, SCIM and other automated approaches for small teams.

· 10 min read

User provisioning is creating someone's accounts and giving them the right access when they join, and changing that access when their role changes. Deprovisioning is the other end: removing it when they leave. In a small company, user provisioning usually means one person working through a list of tools on a new hire's first morning, sending invites one at a time and hoping the list is complete.

This post covers what gets provisioned, the three ways to do it (by hand, through SCIM, and with other automation), how it differs from single sign-on, and how to make it repeatable without paying for an Enterprise plan in every tool.

The definition

Provisioning has three parts, and it is worth naming them because people tend to stop after the first.

Create the account. An invite to Slack, a seat in Figma, a member in the GitHub organization. This is the part everyone remembers, because the new hire notices within an hour if it is missing.

Give the right access. The right role (member, admin, viewer), and the right teams, channels, projects or repositories inside each tool. An account at the default role is half a job. The designer who can only view the Figma files they were hired to edit will tell you so. The salesperson who was made an admin "to save time" will not.

Keep it right. When someone moves from sales to customer success, their access changes too. This is the part most often forgotten, and it is how people end up with access to everything they have ever touched. A three-year employee often carries their onboarding tools, every project they joined since, and a few admin roles nobody remembers granting.

"Access provisioning" and "user provisioning" mean the same thing: the account plus the permissions. "Account provisioning" usually means only the first part. You will also see "identity provisioning", which is the same job described from the identity provider's side.

Provisioning vs SSO

These get confused because they are often sold together, and because a lot of tools put both on the same expensive plan. They do different jobs.

Single sign-on handles logging in. One account, usually in an identity provider or Google Workspace, signs a person in to every connected tool. How SSO works walks through the login itself. SSO does not create the account in each tool, set its role, or remove it when the person leaves.

Provisioning handles the accounts themselves. A tool can have provisioning without SSO (you invite people by email and they set a password) or SSO without provisioning (people sign in through Google, but someone still has to invite them and remove them). The one place they overlap is just-in-time provisioning, where an account is created on the first SSO login. SCIM vs SAML explains the split in standards terms: SAML is the login, SCIM is the accounts.

The practical point: if what you actually want is "add people to the right tools, and get them out when they leave", that is a provisioning problem. Buying SSO to solve it means paying for login features to get account features.

What gets provisioned

For a typical 20-person company on Google Workspace, a new hire needs some mix of the following.

The core account comes first. Google Workspace or Microsoft 365: email, calendar, drive. Everything else hangs off this, because most invites go to that address and many tools let people sign in with it.

Then communication. Slack, Zoom, maybe Loom. Everyone gets these, which makes them the easiest to standardize and the most annoying to forget.

Then the work tools for the role. GitHub, Linear and Vercel for engineers; Figma for designers; HubSpot and Apollo for sales; Intercom or Zendesk for support. This is where roles inside tools matter most, because the same tool can mean read-only access for one person and full control for another.

Then shared resources. Google Groups, shared drives, a password manager vault, shared inboxes. These are easy to miss because they are not separate tools; they are places inside tools you already provisioned.

And finally the money and people tools. Expense cards, payroll and HR portals. These often belong to someone outside IT, which is exactly why they get skipped.

The list is usually longer than anyone thinks. Ask whoever does it now to write it down for one role and count the items. Then compare it with what the last person in that role actually has. The gap between the two lists is your provisioning problem in one number.

Manual provisioning, and where it breaks

Most small companies provision by hand: someone opens each tool's admin page, invites the person, sets the role. It works, and for the first ten people it is fine. It breaks in predictable ways.

It depends on memory. The engineer gets GitHub but not Sentry, because Sentry is not on anyone's list. Roles drift, too: people get admin "for now", and nobody takes it back. It is slow, so a new hire spends their first day waiting for invites and asking in Slack who can add them to things. And nobody records it. Without a record of what was given, there is nothing to check against when the person leaves, which is how offboarding turns into guesswork and how orphaned accounts pile up.

The size of the company changes which of these hurts. At five people, the founder knows every tool and every account, and manual provisioning is the right answer; writing anything more elaborate would cost more than it saves. At fifty, nobody knows every tool, three different people send invites, and hires arrive every few weeks. That is where "by hand" stops being a method and becomes a series of favors.

SCIM provisioning

SCIM (System for Cross-domain Identity Management) is a standard API for exactly this job, defined in RFC 7643 and RFC 7644. Your identity provider, such as Okta, Microsoft Entra ID or JumpCloud, holds the list of people and their groups, and calls each tool's SCIM API to create, update and deactivate accounts. Add someone to the "Engineering" group, and they appear in GitHub; remove them, and they are deactivated.

When it is available, it is the cleanest option. The catch is availability: SCIM needs an identity provider that supports it, and a tool plan that includes it. Tools that offer SCIM usually keep it on their Business or Enterprise plans, often the same plan that has SSO, sometimes a higher one. What SCIM is explains the standard, and SCIM provisioning covers setup and cost.

Google Workspace can do some of this itself. Its help page on automated user provisioning, as we read it on 27 October 2026, says Business Starter and Frontline Starter customers can set it up for up to 3 apps, and higher editions for up to 100. The setup starts with configuring Google as the SSO identity provider for the app, so the tool's own plan still has to allow SSO.

Other automation

Between "by hand" and "SCIM everywhere" there are other routes, each with a clear weakness.

Just-in-time provisioning is the simplest. Some tools create an account the first time someone logs in through SSO. It never removes anyone, and it still needs SSO on the tool's plan. Just-in-time provisioning covers when it is enough.

Domain auto-join is the cheapest. Many tools let anyone with an @yourcompany.com email join the workspace. It is convenient for joiners, but it gives everyone the default access whether they need it or not, and it does nothing for leavers.

Tool APIs and scripts are the developer's route. Most tools have a member API, and some offer it on every plan. GitHub's REST API, for example, has an endpoint to create an organization invitation that takes a role and a list of team IDs, so one call can invite a person as a member of the right teams. Others draw the line higher: Slack's admin.users.invite method says it is only available to workspaces on an Enterprise plan, as we read it on 27 October 2026. A script per tool works until the API changes or the person who wrote it leaves.

Workflow tools are the no-code version. Zapier-style automations can send invites when a row is added to a sheet. They can only do what each app's integration exposes, and they rarely cover removal.

The last route is a tool that drives each tool's admin page for you. This is the SmashSSO approach, covered below. Automated user provisioning without SCIM compares all of these in more depth.

Provisioning by role

The step that makes provisioning repeatable is deciding, once, what each role gets. Write it as a bundle:

RoleTools and access
EveryoneGoogle Workspace, Slack (member), Zoom, password manager, HR portal
EngineerEveryone, plus GitHub (member, engineering team), Linear, Vercel, Sentry
DesignerEveryone, plus Figma (editor), Linear, Notion
SalesEveryone, plus HubSpot (sales seat), Apollo, Gong
SupportEveryone, plus Intercom (agent), Notion (help center space)

Even if you keep doing it by hand, a written bundle turns "what does this person need?" into a checklist. When someone changes role, you compare the old bundle and the new one and change only the difference. That is the whole idea behind role-based access control, scaled down to a table.

Keep bundles few and coarse. Five bundles that are mostly right beat twenty that nobody maintains. Anything outside the bundle is an exception, and an exception should have a name next to it (who asked) and an end (when it stops being needed).

Common mistakes

Putting admin rights in a bundle. It is tempting to give every engineer admin in the cloud console so nobody waits. Bundles copy whatever is in them to everyone who joins, mistakes included.

Inviting personal email addresses. An invite sent to someone's Gmail creates an account your directory cannot see. Match every account to the company address, and record the exceptions, usually contractors.

Forgetting the paid seat. Some tools count a pending invitation as a seat. Know which before you send ten.

Provisioning without a record. If the only record of an invite is the invite email, the leaver's checklist will be built from memory.

What to do this week

Pick your most common role and write its bundle, tool by tool, with the role in each tool. Then pull up the last person hired into that role and compare. Fix the gaps in both directions: what they are missing, and what they have that the role does not need. Finally, decide who sends invites, and make that one person or one system. That alone makes the next hire faster and the next leaver cleaner.

Where SmashSSO fits

SmashSSO does provisioning by role without SCIM or Enterprise plans. You define access bundles like "Engineer", pick a person from your Google Workspace directory, and it invites them to every tool in the bundle with the right role. Role changes and removals run the same way, and when someone is suspended in Google Workspace it alerts you and shows where they still have access.

Tools with a member API are handled through the API when you add a token; the rest are handled by an agent working in your own Chrome, in tabs where you are already signed in as an admin. That browser has to be open for those changes to run, agent runs are slower than an API, and they can fail when a tool changes its screens. Any step that would add a paid seat stops and asks you first, and every change goes to an audit log. It is not SSO, and if an auditor or a customer contract requires enforced login through an identity provider, you need real SSO. It costs a flat fee from $199 a month billed annually, however many people and tools you have (pricing). How it works has the details.

More on onboarding and access

If you are setting up automation, start with automated user provisioning without SCIM, which compares member APIs, domain joining, scripts and browser agents, and how to automate onboarding across your tools, which turns that into a first-day routine. Both assume you have written bundles, so read them after the role table above.

If you want a list to work from on the day someone starts, the IT onboarding checklist for new hires covers accounts, devices and security settings in order, and the employee onboarding checklist focuses on the accounts part a manager or operations lead owns.

If your problem is less the first day and more everything after it, read provisioning and deprovisioning for why the record you keep at grant time becomes the leaver's checklist, the JML process for small teams for handling joiners, movers and leavers as one routine, and user lifecycle management for the full arc from hire to exit. Role-based access control for small teams explains how to design the bundles themselves without enterprise jargon.

FAQ

What is user provisioning?
Creating a person's accounts in the tools they need, with the right role and access, when they join, and changing that access when their role changes. Deprovisioning is removing it when they leave.
What is access provisioning?
Another name for user provisioning: creating the account and giving it the right role, teams and projects. Account provisioning usually means only creating the account, without the permissions.
What is the difference between provisioning and SSO?
SSO handles logging in: one account signs a person in to every connected tool. Provisioning handles the accounts themselves: creating them, setting roles and removing them. A tool can have one without the other.
How do you automate user provisioning?
With SCIM, if your identity provider and each tool's plan support it; with just-in-time accounts created at first SSO login; with domain-based joining; with scripts against each tool's member API; or with a tool that does the admin work across your tools for you.
What is a provisioning tool?
Software that creates, changes and removes user accounts across your other tools from one place, usually driven by your directory and by role. Identity providers with SCIM are the best-known kind.

Keep reading

Stop paying for a login button.

Sign in with Google Workspace, connect your tools, and see who has access to what in minutes.

Sign up with Google