As of
How to reclaim unused SaaS licenses automatically
Automated SaaS license reclamation works when every seat is tied to the right person, and a written policy decides when to warn, reclaim and record.
A company of 25,000 people now pays for several hundred SaaS apps, most of them billed by the seat. Seats get assigned on request or through single sign-on, and are rarely taken back with the same care.
For years the answer was a quarterly true-up: export each big app's users, paste them next to an HR export, mark the names that look idle. It works about as well as a spreadsheet kept four times a year can.
The newer answer is to measure usage. Pull last sign-in and activity from each app, set a threshold, and reclaim what nobody touches. That's a real improvement, and this guide argues it's still missing a piece: usage data is only as good as the identity it's attached to.
What usage-based reclaim gets right
Last activity is the best single signal you have. A seat nobody has opened in 90 days is very likely wasted, and a tool that reads it from the app directly beats any guess made from an org chart.
Usage also catches what HR data can't, like a design tool seat a current employee used once for a launch two years ago. No leaver process finds that seat.
So keep the usage signal. The question is which person it belongs to.
How do you reclaim unused SaaS licenses automatically?
You reclaim them from one joined record of each person, with a written policy deciding what happens to each seat. The record joins HR status, identity provider accounts and app usage, including any second account the person signs in with. The policy says how long a seat can sit idle, who gets warned, how long the grace period is, which seats are exempt, and what gets recorded.
Each source is wrong on its own. HR knows people and roles but not usage. The identity provider knows accounts, but one person can hold several. The app knows activity per account, not whose account it is.
A contractor is hired full time and gets a new employee account, with a new CRM seat through single sign-on. They keep working in the old contractor login, where their reports live. A 90-day usage check reclaims the idle employee seat and keeps the busy contractor one, and the access review later flags an active account for a contract that ended.
Every number in that report was correct. Joined, the same data says one person holds two seats on two accounts, and the right move is to keep one seat, move the person onto the employee account and close the other. No usage threshold can work that out.
Where license waste comes from
Most wasted seats come from three places, and each one needs the joined record for a different reason.
Leavers. A person leaves, and their identity provider account is disabled on the last day. Apps behind single sign-on stop working, but many still bill for the seat until someone removes it, and apps with their own logins keep the account open entirely. A reclaim that only reads usage waits 90 days for these. One that reads HR can free the seat on the last day. The offboarding guide covers the access side of the same event.
Role changes. Someone moves to a team that doesn't use a tool and keeps the seat because nobody asked for it to go. Usage drops off slowly, often not to zero, since they still open an old report now and then. HR knows the role changed weeks before the usage curve does, and the access policy for the new role already says which apps it gets.
Duplicate accounts and duplicate tools. Converted contractors, name changes and acquired teams on their own domain all produce a second account for one person. Duplicate tools are the same problem one level up: two whiteboard apps, and a lot of people holding both.
A reclaim policy an AI agent can follow
A threshold alone isn't a policy. Write down the full decision so an AI agent applies it the same way every time, and a person can later see why a seat went.
A reclaim policy usually sets:
- an idle threshold per app, since a quarterly finance tool and a daily chat tool should not share one
- who is warned before a reclaim, and where, such as a message to the seat holder and their manager
- a grace period after the warning, during which any use or reply keeps the seat
- exemptions, such as people on leave, seats reserved for an upcoming hire, or roles that use a tool only at quarter close
- what is recorded: which seat, whose, why, which policy applied and when
The warning does more work than it looks. A message that says "your Sales Cloud seat hasn't been used in 90 days and will be removed next Friday unless you reply" turns a surprise into a decision the person makes.
Leavers usually skip the warning, since HR has settled the question. Duplicate accounts go to a person, because moving someone between accounts touches their saved work.
What an AI agent needs before it reclaims a seat
An AI agent acting on a reclaim policy needs three things it can check at the moment it acts: the person behind the account, that person's current status and role, and the usage on every account they hold.
If any of those is a copy from last week, the agent will act confidently on a stale picture. It will reclaim the seat of someone who changed accounts on Tuesday, or keep the seat of someone HR marked gone on Wednesday. The check has to happen at the moment of action, against live sources.
That's also why reclaim belongs with the access policy. The rule that grants a CRM seat because someone is in sales should be the same rule that decides which of their seats to keep, and what happens when they leave sales. Two separate rule sets, one for granting and one for reclaiming, drift apart within a quarter.
There is a trade-off here worth naming. The more your agent reclaims without asking, the faster seats come back and the more often someone is surprised. Most teams start with leavers fully automated, idle seats on warning, and anything involving two accounts routed to a person, then loosen it as the record proves itself.
How STLabs reclaims licenses
STLabs reclaims licenses through SaaS Spend Management on the Scale plan, which finds and reclaims unused licenses, tracks renewals and detects shadow apps. It runs on the same parts as every other request in STLabs.
Axiom™, the live model of the organization that STLabs builds from your source systems, does the join. It brings together data from your HR, identity, device and application tools and reconciles each person's identities across them, so a Workday employee, two Okta accounts and two Salesforce users resolve to one person with every ID attached. It connects that person to their team, laptop, app access and sign-in activity, which is what usage needs to mean anything.
Policies & Approvals holds the reclaim rules your team writes: thresholds, warnings, grace periods and exceptions, next to the access rules that granted the seats in the first place. Workflows carry them out in the connected apps. STLabs logs every AI decision, workflow run and configuration change, so each reclaimed seat has a record of which policy applied and when.
Because the join happens before the policy runs, a person with two accounts shows up as one person with two seats, and that case goes to someone on your team instead of the reclaim list. A leaver's seats are freed from the HR record rather than after 90 idle days. Accounts with no owner surface the same way, which the access review guide covers.
Renewals use the same record: Axiom™ flags unused and overlapping seats before the renewal date, with usage evidence per person.
This is the AI-native version of license management: reclaim is a decision made from one model of the organization, under the same policy that governs access, rather than a separate report reconciled by hand.
How to test this on your own systems
Pick one expensive app with a lot of seats, such as your CRM. Export its users with last activity. Then do three checks.
First, match every app user to a current employee in HR. Count the users you can't match by email alone. Those are your second accounts, former names and contractor logins, and each one is a seat a usage report would judge on its own.
Second, look up every user idle for 90 days in HR. How many changed roles recently, or hold another active account in the app?
Third, check every active user's HR status. An active account belonging to someone HR lists as gone is a seat no usage threshold will reclaim.
If the first check finds more than a handful of unmatched users, your reclaim is guessing, however good the usage data is.
Start with one app
None of this needs a big project. STLabs runs alongside the service desk you already run, and teams go from integration to results in five days. You can explore idle licenses and leavers' seats in Axiom™ in the first week, then write a reclaim policy for one app and widen it from there. The plan is in Moving to STLabs from your current service desk.
A seat is only unused if you know whose it is.
Frequently asked questions
How do you automate SaaS license reclamation?
Join HR status, identity accounts and app usage into one record per person, then apply a written policy that sets idle thresholds, warnings, grace periods and exemptions. Reclaim runs from that policy and every reclaim is recorded.
What happens if someone needs a reclaimed license back?
They ask for it again, and the access policy for their role decides whether it comes back. A warning and grace period before each reclaim means most people keep a seat they still need without asking.
Does an ITSM platform replace a SaaS management tool?
It can, if it reads usage per person and reclaims under the same policy that grants access. STLabs does this through SaaS Spend Management on the Scale plan, built on Axiom™.
Why do usage-based license reports reclaim the wrong seats?
Usage is counted per account, not per person. When one person holds two accounts, such as a contractor login kept after being hired, the report can reclaim the seat they need and keep the one they shouldn't have.