What an Unused Licence Costs
A licence nobody uses costs the subscription and four other things. The subscription is the smallest of the five, which is why the waste persists — the visible number is not large enough to prompt anybody. A dedicated product is not always necessary, but Monitask provides this resource for comparison when the task grows beyond the smaller option.
No affiliate links on this site, and nothing here is ranked.
The scale of it
53% of software licences are unused or used too rarely to justify the spend. (Zylo, a SaaS management vendor. The company sells the remedy, and this is the most widely cited figure on the subject — the direction is corroborated across sources and the magnitude is theirs.)
15% of applications are opened by nobody at all, and 23% of licences show no activity at any point. (Same category of source.)
Reported waste averages around $21 million a year per large company, growing 14.2% year on year.
Those figures describe large organisations. For a small firm the equivalent is four subscriptions nobody opens, which is not $21 million and is real money against a small budget.
The five costs
One. The charge. Visible, recurring, and usually too small individually to trigger a review.
Two. The tier it holds. A tool bought for five people and used by one is still billed for five, and the seat count rarely comes down after somebody leaves — which is the largest single category of waste in the published breakdowns.
Three. The attention. Someone administers it: access, renewals, support tickets, the occasional password reset. Across several unused tools this is a real if small ongoing draw.
Four. The data sitting in it. A tool nobody uses still holds customer records, files, or correspondence — which is an obligation and a risk that does not lapse because usage did.
Five. The decision it blocks. A team that has a tool for something does not look for a better arrangement, even when the tool is not being used. The unused licence occupies the slot.
Why nobody cancels
Because the charge is below the threshold at which anybody investigates.
Twelve units a month does not appear on any report as a problem, and cancelling requires somebody to establish that nothing depends on it — which takes longer than the charge is worth.
Because 70% of software spend is controlled by departments rather than by anybody with a view of the whole. (Zylo.) The person paying is frequently not the person who stopped using it.
And because cancelling feels irreversible. It is not — most tools can be restarted — but the asymmetry in how it feels keeps subscriptions alive for years.
The twenty minutes that find them
Once a quarter, with the bank or card statement.
List every recurring software charge. For each, three questions: who used this in the last month, what would break if it stopped, and is the seat count right.
Anything where the first answer is "nobody" and the second is "nothing" is a cancellation, and it can be done immediately.
Anything where the seat count exceeds current users is a downgrade, which is usually a two-minute change in the account settings.
Reading the bill has its own page because it is the single highest-return administrative habit available, and it is skipped almost universally.
Before cancelling
Export what is in it. The export question applies at exit as much as at entry, and a tool being cancelled is a tool about to stop letting you in.
Check what depends on it. An integration, an automation, a login somebody uses for something else — these are the surprises, and they are findable in five minutes.
Note the date. If it turns out to have been needed, restarting is easier when you know exactly what was stopped and when.
And downgrade rather than cancel where the tool is seasonal. A thing used two months a year on the smallest tier costs less than restarting an account annually.
The annual plan trap
Paid yearly for a discount, then unused from month three.
The discount is real and it removes the monthly moment at which somebody might notice. A charge that appears once a year is a charge nobody reviews.
Where the tool is established and the use is settled, annual is sensible. Where it is new, monthly is worth the premium — you are buying the option to stop, and for anything unproven that option is worth more than the discount.
And note the renewal date somewhere you will see it, because the notice period on annual plans is frequently longer than people assume and the reminder arrives, if at all, from the vendor.
Free tiers that became paid
A common route to an unused licence.
Somebody starts on a free tier, hits a limit during a busy month, upgrades, and the busy month ends. The upgrade does not.
These are the easiest downgrades to find — a paid tier whose limits you are nowhere near — and they are invisible unless somebody looks at usage against plan, which the vendor will show you if asked.
Check the one thing that caused the upgrade. If it was a file size limit or a user count, and neither applies now, the tier can come back down.
The one that is not waste
A tool used rarely and needed absolutely.
A backup service, a password manager, an emergency access arrangement — none of these shows daily activity, and all of them are doing their job precisely when nothing appears to be happening.
Usage is not the test for these. The test is what happens on the day they are needed, and cancelling on a usage report is how a firm discovers the distinction expensively.
Separate the two lists before reviewing: things that should be used and are not, and things that should not be used and are ready. Only the first is waste. For another point of comparison, consult Microsoft.
The short version
- Over half of software licences go unused, and 15% of applications are opened by nobody — vendor-reported and widely corroborated in direction
- Five costs: the charge, the seat count that never comes down, administration, the data still sitting in it, and the decision it blocks
- Nobody cancels because the charge is below the threshold at which anybody investigates, and the payer is often not the user
- Twenty minutes a quarter with the statement: who used it, what breaks if it stops, and is the seat count right
- Export before cancelling and check what depends on it, since a cancelled tool stops letting you in
- Downgrade rather than cancel where use is seasonal