Budgeting Annual Software Renewals Without the January Surprise

Fix it now

The January surprise is rarely a price rise. It is four renewals landing in the same six weeks because nobody aligned them, plus seats bought for a project in July that nobody reclaimed. One register, one owner and one anniversary fixes most of it, and none of it requires negotiating with anybody.

1. Build one register with a row per purchased thing, and give it a named owner rather than a department. The fields that drive money are the term, the end date, the auto-renew status and the counting unit.
2. Split the commitment deliberately: an annual term only for headcount you would still employ in a bad quarter, and monthly terms for everything above that floor.
3. Co-terminate what you intend to keep. Microsoft calls it coterminosity, charges are pro-rated where a term is shortened, and it turns four budget conversations a year into one.
4. Do not try to align annual or three-year terms with monthly ones – Microsoft does not permit that combination. A monthly term can co-terminate with anything else.
5. Run a reclaim pass before you forecast. Count assigned seats against purchased seats in the Microsoft 365 admin center at Billing then Licenses, and move users who only need mail and web apps to a lighter plan.
6. Forecast on the standing figure your supplier confirms in writing for the coming term, not on last year’s invoice, and add the lines that scale on something other than headcount.

The only point at which quantities can be reduced is the cancellation window: seven calendar days from purchase, with a pro-rated refund, and it does not reopen until the subscription renews into a new term.

If you have a register, one anniversary and a reclaim pass in the diary, the budget will not surprise you. Below are the fields that earn their place, how alignment actually works, and the lines that ambush a headcount-based forecast.

Why it happens

Most of the surprise in a software budget is structural rather than commercial. Renewal dates scattered through the year mean nobody sees the total until it has already been committed. Terms that cannot be reduced mid-year mean a headcount drop in March does not reach the bill until the following January. Seats added for a project and never reclaimed compound quietly. All three are fixable with a spreadsheet and one named owner, and none of them involves asking anybody for a discount.

The commitment decision is the one with real money in it. New commerce licence-based subscriptions support monthly, annual and three-year terms, with monthly, annual or up-front billing. An annual term is cheaper per seat and, in practice, cannot be reduced after the first week: the cancellation window is seven calendar days from purchase, with a pro-rated refund, and it does not reopen until the subscription renews into a new term. Monthly costs more per seat and buys the ability to shrink. Most organisations should hold both rather than choosing one.

Alignment is the other structural lever and it has a name and rules. Coterminosity lets a new or renewed subscription be aligned to an existing end date, with charges pro-rated where the original term is reduced by the coterminous setting. It is available for licence-based new commerce subscriptions, add-ons can align to an existing subscription’s end date, and there is one restriction worth knowing before you plan around it: you cannot co-terminate an annual or three-year term with a monthly term, though a monthly term can co-terminate with anything else. Trials are excluded.

Full reference

The register, and why each field is there

Aligning the year so it has one shape

Coterminosity is the mechanism, and it works in both directions: you can specify an end-date alignment value when buying a new subscription, view existing subscription end dates and pick one, or set alignment on an existing subscription through its renewal settings. Charges are pro-rated where the original term is reduced by the coterminous setting, and where the alignment is to a calendar month the charges are calculated for the billable days in that month.

The restriction to plan around is the term mismatch. An annual or three-year term subscription cannot be co-terminated with a monthly term subscription, though a monthly term can co-terminate with any other. Trials are excluded. So the practical shape is one anniversary for the annual core, with the monthly headroom sitting alongside it rather than folded into it.

Forecasting growth without guessing

Run the arithmetic in the open rather than adding a percentage. Start with seats actually assigned today, not seats purchased. Subtract seats still assigned to people who have left. Add the hires already in the plan for the coming term. Add whatever contingency you choose, but choose it deliberately and write down why. Then multiply by the standing figure your supplier confirms in writing for the coming term, not the figure on last year’s invoice.

Finally, add the items that scale on a different axis to headcount, because these are the ones that break a headcount-based forecast.

• CALs when a server version changes. Windows Server 2025 CALs cover 2025 and prior versions, so a new server version means new CALs for every user or device.
• Core licences when a host is replaced. The 8-per-processor and 16-per-server minimums apply regardless of load, so a bigger replacement costs more before it runs a workload.
• Backup, which is usually billed by consumption and grows without anybody deciding it should.
• Per-device antivirus during a hardware refresh, where the seat count moves with the machines rather than the people.
• Anything priced in a currency you do not bill in.
• Software Assurance renewals on products where lapsing is not an option, such as Exchange Server SE and SharePoint Server Subscription Edition.

The lines that ambush a budget

• The standing figure after an introductory term, particularly on security products sold with automatic renewal on by default.
• Seats never reclaimed from leavers, which compound year on year and are invisible unless somebody compares assigned against purchased.
• Auto-renew still running on a product you replaced and forgot to cancel.
• A mid-year project that added seats nobody entered in the register.
• A notice period missed by a fortnight, locking in another full term.
• Support or maintenance bundled in year one and billed separately from year two.

Which cadence to run, by size

• Under ten people: one spreadsheet reviewed twice a year, with calendar reminders 60 days before each end date. That is genuinely enough.
• Ten to fifty: the register plus a quarterly reclaim pass on assigned seats. The reclaim usually pays for the time spent doing it.
• Fifty to three hundred: co-terminate everything you intend to keep, split annual and monthly commitment deliberately, and review the plan mix annually rather than renewing like for like. Watch the 300-seat Business family cap as a budget threshold as well as a licensing one.
• Multiple entities or countries: one register per entity, one consolidated view, and an early check on whether the agreements can be combined at all.

Questions people ask about this