Microsoft 365 Licensing: Finding the Money You're Already Spending

Microsoft 365 spend grows by accretion. Someone joins and gets whatever the last person got. Someone leaves and the licence sits assigned. A pilot adds a premium SKU for twenty people and never gets reviewed. None of it is negligent; all of it is money.

Four categories account for most of the waste, and all four are visible from the admin centre without a consultant.

1. Assigned licences for people who have gone

The simplest and most common. Licences remain assigned to accounts of people who left, because offboarding handled access — disabling the account — and never handled cost.

# Accounts that are blocked from signing in but still hold licences
Get-MgUser -Filter "accountEnabled eq false" -All `
  -Property DisplayName,UserPrincipalName,AssignedLicenses |
  Where-Object { $_.AssignedLicenses.Count -gt 0 } |
  Select-Object DisplayName, UserPrincipalName

Mailboxes often need to be retained after someone leaves. The right mechanism is usually to convert to a shared mailbox (free below the size limit) or apply a retention hold and remove the licence — after confirming your retention obligations with whoever owns them. Do not delete first and check later.

Then fix the cause. If licence removal is not a step in your offboarding checklist, this grows back within a year.

2. Licences assigned to people who never use the feature

The expensive one. Premium SKUs are bought for capabilities that a fraction of holders use.

The admin centre reports per-service usage. Look for people holding a premium licence whose activity shows only email and basic Office use. Common patterns:

• frontline and shift workers on full E3/E5 when a frontline SKU covers their actual use
• premium security features licensed organisation-wide but deployed to nobody
• Power BI Pro assigned to people who have never opened a report
• Project or Visio licences from a project that finished
• Audio Conferencing on users who never dial in

The honest caveat: some licences are bought for rights rather than usage. A security feature you have not deployed yet is a plan, not waste. The question is whether there is a dated plan to use it, or whether it was bought in a bundle and forgotten.

3. The wrong SKU mix

Most organisations have a single standard licence for everyone, which is administratively simple and almost always wrong on cost. Real organisations have distinct user populations:

• Information workers — full productivity, collaboration, meetings
• Frontline workers — shift-based, shared devices, often no personal desk. Purpose-built SKUs cost a fraction of E3
• External or occasional users — may need guest access rather than a licence at all
• Service and shared accounts — shared mailboxes, resource mailboxes for rooms and equipment, and most service identities do not require a user licence

That last point is worth checking carefully: shared mailboxes under the size limit, and room and equipment mailboxes, do not need a licence. Licensing them is pure waste and it is common.

4. Overlap with other spend

Microsoft 365 bundles capability you may be buying separately:

• endpoint protection where a third-party EDR is also licensed
• a mobile device management product duplicated by Intune
• a meeting platform alongside Teams
• a document signing product where a bundled capability would do
• backup where retention policies and a separate backup product overlap in scope

Overlap is not automatically waste — "best of breed" is a legitimate choice, and the bundled option is sometimes genuinely weaker. The waste is when nobody has decided, and both are being paid for because each was bought by a different person at a different time.

Running the review

Quarterly, an hour:

• Licences assigned to disabled accounts → reclaim
• Licences assigned, zero activity for 90 days → investigate
• Licence count versus active user count → the gap is your buffer, and it should be small
• New SKUs added since last review → who approved, for what

Annually, before renewal:

• Full user population segmented by actual need
• Every premium feature mapped to whether it is deployed
• Overlap audit against your other security and productivity spend
• Growth forecast, so you commit to a number you will use

Two practical cautions

Reserved commitments cut both ways. Annual commitments are cheaper per seat and lock the count. Over-committing to get the better rate is the most expensive optimisation available. Commit to your floor, add monthly seats above it.

Changing someone's SKU can remove capability they quietly depend on. Downgrade in small batches, tell the people affected, and watch for a week before the next batch. The saving is not worth an outage in someone's workflow discovered at month-end.

What good looks like

A named owner for licensing. A documented standard for each user type. Licence assignment driven by group membership rather than by hand, so joining a role grants the right licence automatically. Removal as a step in offboarding. And a quarterly report that someone actually reads.

The savings from doing this properly are usually in the range that funds a project. The reason it goes undone is that it belongs to nobody in particular — which is also why assigning an owner is the single highest-value step on the list.