Excess AI spending usually comes from licenses that were paid for and nobody uses, duplicate tools across teams and subscriptions that have no owner. Each line looks small, and the total adds up at the end of the month. To reduce it, you need to measure usage.
Finding AI licenses that are not in use
These are the places where we find the unnecessary charges, mostly at companies that started buying AI tools before they had a purchasing procedure.
- Unused licenses. Licenses were bought for the whole team, and in practice three people log in.
- Duplicate tools. Marketing, development and support bought three separate tools that do the same thing.
- Shadow subscriptions. An employee pays with a personal card and files an expense claim, and IT doesn't know the subscription exists.
- Trials that rolled over. A free trial that renewed as an annual subscription without anyone approving it.
- Expired credits. Prepaid credit that expired before it was used.
- Employees who left. The license stays active and the billing continues.
These lines have no owner. When nobody is responsible for a line item, nobody cancels it either. So the fix is a standing procedure: who is responsible for each tool and when usage gets reviewed.
Dormant licenses inflate spending
A manager sees a high price on one tool and tries to bring it down, while paying for twenty licenses nobody uses.
An expensive tool that ten people use every day pays for itself. An idle license inflates spending, lowers average utilization and makes it harder to justify more licenses when you really need them.
There's an indirect cost too. When management sees a large amount spent with no result, it stops buying new tools, and even teams that would benefit from the right tool end up going without.
The metric to check is cost per active employee. Two tools with the same license price can end up costing very different amounts, depending on how many people actually work with them.
Duplicate tools, personal cards and shadow subscriptions
At companies that grew fast this almost always happens. Each team bought a tool for its own problem, and there is no single list of tools and no one who sees the full total.
- Build one list: every AI tool, who owns it, what it costs and which payment method it is billed to.
- Go through six months of expense reports. That is where the shadow subscriptions show up.
- Ask accounting for every recurring charge, especially the ones billed in dollars.
- Consolidate tools that do the same thing and pick one per need.
- Move every subscription to a single company card or a managed vendor account.
Once you have a list, you can manage the spending. Also decide who approves a new tool, or the list will be out of date within two months.
Per-license or usage-based pricing
Before you approve a new tool, check how it bills. There are a few common models, and each one suits a different kind of use. The price on the vendor's site is almost always in dollars and before VAT, and in Israel 18% is added.
| Pricing model | When it pays off | What to look out for |
|---|---|---|
| Per user | Daily, steady use by the same people | An unused license is a total loss |
| Per usage or tokens | Uneven use, projects and peak periods | Easy to miss a sudden spike in the bill |
| Base subscription plus usage | A small team that runs a lot | Needs a monthly cap and an alert |
| Prepaid credits | A discount in exchange for commitment | Credit not used by the deadline is lost |
In practice, tools the team opens every morning are better bought per license, and tools used for projects are better run on usage with a monthly cap and an alert. If a vendor offers a discount for an annual commitment, first check how many licenses were active over the last three months.
How to measure actual usage per employee
A team manager's estimate isn't enough. Almost every vendor shows usage data in the admin portal or through an API. The first time it takes about an hour, and after that the check can run automatically.
- Pull the list of paid licenses from every vendor, and who is assigned to each one.
- We pull login and usage data directly from the API.
- Flag as dormant any license untouched for thirty days.
- Split the active users into heavy, medium and light usage.
- For each tool we calculate the cost per active employee and bring that number to management for discussion.
AtIT management dashboard we build for every client, there is a module that shows, for each tool, paid licenses against active users, and the split between heavy and light use. For example, one row might show 40 paid licenses, 28 users who logged in during the last month, 12 dormant and 70% utilization. The numbers are for illustration only and are not client data.
A 30-minute monthly routine
A one-time cancellation is not enough, because new subscriptions keep coming in. Set a short meeting on a fixed date, with the same table and the same questions.
- Open the subscription table and compare spend against last month.
- Go through the dormant licenses and cancel them or reassign them to another employee.
- Check whether new tools came in, and whether they duplicate a tool you already have.
- Approve next month's budget based on actual usage, not on what was requested.
- Flag renewals and credits that expire within the next 30 days.
A short meeting every month saves more than a big review once a year, because canceling a dormant license only saves money if it happens before the next charge.
What to ask before approving a new tool
Cost aside, an AI tool receives business data and sometimes client data too. So before approving a new tool, ask these questions:
- Where the data is stored, and who on the vendor's team can access it.
- Whether what you enter is used to train models, and how to turn that off.
- What deletion actually means on their side, and how long logs are retained.
- Whether it connects to Entra ID or Google Workspace, so that access can be cut off the day an employee leaves.
- Who in the company approves a new tool, and which payment method it is billed to.
With our clients managed IT services we go over these questions before a new tool goes into use, and connect it to the organizational identity. That way spending and data access are managed in one place.
FAQ
What is a dormant license?
A paid license that nobody has signed into during the period you defined, usually thirty days. Some vendors show this directly in the admin console, and with others you have to pull the figure from the API. Once a license is flagged as dormant, the decision is either to cancel it or to reassign it to an employee who will actually use it.
How long does it take to see savings?
The first month is mostly mapping, and that is where shadow subscriptions and the licenses of employees who left come to light. Cancellations can be made as early as the next billing cycle, and from the second month a short monthly check is enough.
Should we just ban employees from using AI tools?
A blanket ban pushes usage onto personal credit cards and accounts you cannot see, which is riskier from a security standpoint. It is better to approve a small number of tools, define what may be entered into them and measure usage.
How do we know a tool is safe for company data?
Ask where the data is stored, whether it is used to train models, how to turn that off, and what actually happens when you delete. It should also connect to the organization's identity provider, so that offboarding an employee cuts off access immediately. We go through these questions with the client before a new tool is approved.
Who should own the spend?
Every tool needs one owner, usually the manager of the team that uses it, and above them someone who sees all of the spending. In a managed service, we bring all the spending together and present it for a decision once a month.
More on this topic: IT management dashboard · Managed IT services · What managed IT services include · IT and security assessment