The licence you keep paying for after the seat goes dark
License waste in Microsoft 365 has two honest layers: seats you can reclaim from hard data today, and a term premium you can model but not read from Graph. Here is how I separate them so the number I quote is defensible.
License waste is easy to overstate and easy to miss. I keep it honest by splitting it into two layers that carry different confidence, and only the first is a number I will assert. This memo sets out the split I use in the ITSailor CSP benchmark.
Last verified: 2026-08-03.
The two layers
Layer one, recoverable now. Dead seats and duplicate or overlapping SKUs, derived from the scan. A disabled account still holding an assigned licence is the clearest case, read straight from Graph. This is measured, not estimated.
Layer two, commitment opportunity. Microsoft New Commerce prices a month-to-month term about 20 percent above the annual term, so moving the right-sized estate to an annual commitment removes roughly 16.7 percent of that bill. The scan cannot see the customer term from Graph, so this is presented as a clearly labelled potential applied to the right-sized estate, never asserted as certain.
Recommendation
Quote layer one as the hard saving and layer two as a modelled potential, in that order, with the arithmetic shown. Right-size first (remove dead and duplicate seats), then price the remaining estate on annual terms. For the market context, LicenseIQ's 2026 playbook puts wasted Microsoft 365 spend at 8 to 22 percent of the estate with up to 35 percent recoverable, which is a useful outside anchor but not a promise about any one tenant.
# The benchmark arithmetic, in the open
currentLicensedMonthly = sum(monthly cost of every account holding a priced licence)
rightSized = currentLicensedMonthly - recoverableNow # dead + duplicate
commitmentSaving = rightSized * (0.20 / 1.20) # ~16.7%
optimizedMonthly = rightSized - commitmentSaving
annualOpportunity = (currentLicensedMonthly - optimizedMonthly) * 12
Trade-offs
An annual commitment trades flexibility for the lower rate, so it is wrong for a seasonal or shrinking headcount. Right-sizing removes seats a manager may want back next month, so pair it with the offboarding evidence rather than a blind sweep. The licensing service is at /services/cloud-licensing-and-procurement.
Where this does not apply
The commitment layer does not apply to a tenant already on annual terms: there is no premium left to remove, so quoting it would double-count. The recoverable layer does not apply where the "dead" seat is a deliberately parked account with a business reason. This is evidence for a procurement decision, not a compliance statement, and the LicenseIQ market figures were not independently re-verified by ITSailor.
Sources and further reading
Turn the trade-off into a scoped brief.
Share the constraints that differ in your environment. Michal will identify the next check needed before a delivery decision.
Start a scoped briefMore from Ops Log
Shadow AI is a governance-readiness problem, not a deadline
The first artifact three separate AI frameworks all demand is the same: a register of the AI systems in use, their data posture, and an owner. Build that register now for the governance value, and let the shifting AI Act dates be a secondary driver.
Annual or month-to-month: the New Commerce term is a 20 percent decision
Under Microsoft New Commerce, the same Microsoft 365 seat costs about 20 percent more on a month-to-month term than on the annual term. The term you pick is a pricing decision on its own, before any tier change.
DMARC at p=none is not protection: what a deliverability check reads from public DNS
A DMARC record at p=none observes spoofing without blocking it. A free deliverability check reads SPF, DKIM, and DMARC from public DNS and tells you exactly which of those three is only watching. Here is how to run it and read it.