How to stop paying for software seats nobody uses

Almost every line on a software invoice was decided by somebody. The seat count is the exception. It gets set once, at purchase, by a guess about headcount, and then it moves on its own: up when a team grows, up again when a project needs three logins for a fortnight, and never back down, because nothing in the process ever asks it to.

The result is the most reliable saving in a software estate. It needs no negotiation, no vendor conversation and no change to how anybody works. It only needs somebody to count.

What you are actually buying when you buy a seat

A seat is one named user's right to use the software. Per-seat pricing charges for the seats you have bought, not the seats anybody logged into, which is the whole reason unused seats are invisible: nothing about them looks different on the invoice, and the product works perfectly well while you pay for them.

Two details in the contract decide how much that costs you. The first is whether the number of seats is a minimum commitment or a ceiling you draw down against. The second is whether you may reduce it mid-term at all, or only at renewal. Most agreements answer minimum and renewal, in that order, and those two answers together are what turn a stale number into a bill.

The four ways a seat count drifts

• Leavers. Somebody leaves and their accounts are closed for security reasons, which is not the same as the licence being given up. Deactivating a user inside the product very often leaves the seat purchased and billed.
• Bulk buying. Seats are cheaper in blocks, so a team buying for twelve people buys twenty. That is a reasonable decision at the time. It stops being reasonable at the third renewal, when nobody remembers it was a decision at all.
• Short-term additions. A contractor, a migration, an audit. The seat is added in a minute by whoever runs the tool and is never mentioned again, because removing it saves money on somebody else's budget line.
• Tier thresholds. Where pricing steps at a band boundary, one extra user can move the whole account into a higher bracket and raise the rate on every seat, not just the new one. That is the most expensive single seat you will ever add, and it is added by exactly the same two clicks as any other.

None of the four is a mistake. They are all sensible acts by people doing their jobs, which is precisely why no individual is going to catch the drift. Only a periodic count will.

What the unused seats actually cost

Take a tool at £14 per user per month, bought for 60 seats. That is £840 a month and £10,080 a year. Suppose 11 of those seats belong to people who have left or who have not opened the product in six months. Those 11 seats cost £154 a month and £1,848 a year, for nothing at all.

Now add the part that makes seats different from every other kind of waste: the timing. If the contract only permits a reduction at renewal, and the renewal is 11 months away, those seats will cost a further £1,694 before you are allowed to do anything about them. Finding the waste does not stop it. Only the renewal date does, and it arrives once a year.

On a three-year term the same 11 seats are worth £5,544. That is one tool, at a modest per-user price, in an estate that probably has thirty of them.

Why seats are hard to cut at renewal, and easy to cut before it

The seat count is one of the few commercial terms you can change without asking permission, but only in one direction and only in one window. Miss the window and the number rolls over untouched for another term, which is how a count set by a headcount guess three years ago survives into its fourth year.

So the review has to happen before the renewal, not at it, and how far before is set by the notice period rather than by the renewal date. A tool with 90 days' notice needs its seat count settled more than three months out; by the renewal itself the decision has already been made for you. Our guide to notice periods covers reading that clause properly.

There is a second reason to do it early. A seat reduction is the one request a vendor cannot really argue with, but it is also the one they will offer to soften: a lower rate to keep the count, a longer term to keep the rate, credits against a bigger commitment. Those are all worth hearing, and none of them can be heard in the last week before a renewal. Arriving with a verified number and time to spend is what turns the count into leverage rather than a form to submit.

A seat review that takes about an hour

1. 1List every per-seat tool with its seat count, its price per seat, its renewal date and its notice period. If that list does not exist yet, building it is most of the work and you only do it once.
2. 2Sort by seats multiplied by price. The top five tools will hold nearly all of the recoverable money, and the long tail is rarely worth the hour it costs.
3. 3For each of those, open the vendor's own admin console and export the user list. The vendor's list is the authority on what you are billed for; your intranet directory is not, and the two disagree more often than anybody expects.
4. 4Cross off leavers first. This is the fastest, least arguable category, and in most estates it is the largest.
5. 5Then look at last-active dates where the console shows them. Six months without a login is a conversation, not a decision: ask the owner before removing anything, because the quiet seat is sometimes the one person doing the annual filing.
6. 6Write the new number down with the date you checked it, and put the next check in the calendar for a month before the notice deadline rather than a month before the renewal.

Step six is the one that decides whether you ever do this again. A count with no date attached is worth nothing in six months, because nobody can tell whether it predates the last three joiners.

What to record so the next review is ten minutes

The reason a seat review feels like a project is that the four facts it needs live in four different places: the seat count in the vendor's console, the price in a purchase order, the renewal date in somebody's inbox, and the notice period in a PDF. Put them on one record and the review stops being an exercise and becomes a look.

• The seat count, and the date you last verified it against the vendor's own list.
• The price per seat and the billing cycle, so the annual cost of a stale count is arithmetic rather than a guess.
• The renewal date and the notice period, because those two together, not the renewal alone, set the day the decision has to be made.
• An owner: the person who is asked before a seat is removed, and who is asked why the count grew.

StackTrackr keeps all of that on one record: the seat count sits beside the cost, the billing cycle, the renewal date and the notice period that decides when you can change it. Its reminders fire on the cancellation deadline as well as the renewal, so the date you actually have to act on is the one that reaches you. Have a look at what it tracks, or start a free trial and put your five biggest per-seat tools in it.