You are paying for seats nobody sits in

If you manage a company in the GCC, you already know the feeling: a tool someone bought two years ago, still on the credit card, still in the budget pack, used by four people out of the forty you pay for.

That is not a rounding error. It is the normal state of software ownership.

Zylo’s 2026 SaaS Management Index, covering 40 million licences and $75 billion in spend, puts unused licences at 36% against recommended utilisation. Vertice, looking at a similarly large spend book in Q2 2026, puts the wider waste number higher: 65% of licences unused or underused (less than half the purchased seats actually working), of which 14% are fully abandoned shelfware.

• Licences unused vs recommended utilisation36%

  Zylo 2026. The cleanest single ‘empty seat’ number.

• Unused or underused (<50% of seats active)65%

  Vertice Q2 2026. Includes tools that are ‘live’ but mostly idle.

• Fully abandoned shelfware14%

  Vertice Q2 2026. Paid for. Nobody logs in.

Unused rate: Zylo 2026 SaaS Management Index (40M licences, $75B spend). Unused + underused / fully abandoned: Vertice Q2 2026 ($75B+ processed spend). These are not GCC-only samples, so treat them as a floor rather than a local census.

You do not need to be a 10,000-person enterprise for this to hurt. At 40 people, 36% unused is not “$20 million of waste.” It is a CRM you oversized, a project tool the last manager bought, a design suite on a founder’s card, and a second helpdesk that lost the argument and never got cancelled. That is still real money, every year, in a region where software is often bought once and lived with for years.

Why it keeps happening

Zylo’s 2026 index is blunt about ownership: business units now control 81% of SaaS spend. IT directly manages 15%. Expense-based SaaS spend (the card, the personal ChatGPT, the “just this once” subscription) was up 267% year on year, and ChatGPT is now the most-expensed app in that dataset.

• Spend controlled by business units81%
• Spend IT directly manages15%
• IT leaders hit by surprise consumption charges78%

Zylo 2026 SaaS Management Index. 78% of surveyed IT leaders reported unexpected consumption / AI charges in the last 12 months; 61% cut other projects to cover them.

This is the GCC pattern too, even without a local census: a department head buys what the last company used, a partner “includes” extra seats in a bundle, a renewal auto-rolls because nobody owned the calendar. Nobody is stealing. Nobody is looking.

Seats vs usage is not a report. It is a conversation.

A usage export from the vendor is a starting point, not the answer. Vendors count logins the way that makes the renewal look healthy. You want:

• Licences purchased vs licences assigned vs licences that did a real job last month. A login is not a job.
• Who requested the tool, and whether they still work there.
• Whether a cheaper tier, a shared licence, or a different product already in the stack does 80% of it.
• The renewal date and the notice window. Unused seats you discover 11 days before auto-renewal are a complaint, not a negotiation.

Do this per product, not as a grand “SaaS transformation.” Start with the three largest contracts. CRM, Microsoft 365 / Google, and whatever the company uses to get paid.

What “optimise” actually means

There are only four honest outcomes, and they are not equally comfortable:

1. Reclaim seats. Turn them off, or stop paying for the idle ones at renewal.
2. Downgrade the tier. You bought Enterprise because the demo had a dashboard. Three people use it.
3. Consolidate. Two tools doing one job. Pick one. The other is a migration, not a “maybe later.”
4. Leave. If usage is low and the job is real, the product is wrong. That is a switch, not a discount.

A vendor will offer a discount to keep the seat count. That is not optimisation. That is a cheaper version of the same waste.

What to do this week

• Export seat lists for your three biggest contracts. Add a column: last meaningful action, not last login.
• Send that list to the department head who “owns” it. Ask them to mark keep / drop / don’t know. Don’t know becomes drop in 30 days unless someone claims it.
• Put renewal + notice-window dates on one calendar. 180 / 90 / 60 / 30. Not in someone’s head.
• Anything purchased on a personal card in the last year: write it down. That is your shadow stack. It is already in the Zylo numbers. It is in yours too.

If you want a structured pass at this (seats, contract value, whether to renegotiate or migrate) that is the “optimise what you pay for” branch of the tech check. The spreadsheet is still yours. The sequence is the part companies skip.

Where this goes next

Unused seats are the easy half. The harder half is timing: the price you pay for the seats you do use is set months before the invoice arrives, which is the argument in why a renewal date is not when you have leverage. And if the tool in question is a CRM, the seat line was never the real number anyway. See why the licence is the smaller invoice.

Reconciling licences against real usage across a whole stack is what a SaaS spend audit does end to end.

Sources

• Zylo, 2026 SaaS Management Index (Jan 2026): 36% unused licences; IT manages 15% of spend; business units 81%; expense-based SaaS +267% YoY
• Vertice unused-SaaS insights, Q2 2026: 65% unused or underused; 14% fully abandoned
• Zylo, “What Is Shelfware?” (2026): 53% unused or not used enough to warrant the spend, on their blended definition