Five numbers on SaaS waste — unused licences, portfolio sprawl, dollars burned, AI spend growth and cloud waste — and why visibility alone never fixes any of them.
In the early days of SaaS adoption the promise was clear: faster deployment, lower infrastructure cost, more flexibility. For the most part SaaS delivered. But a different picture has emerged alongside it — one marked by duplication, underuse and growing waste.
SaaS introduced new problems of visibility, governance and control. When licences are bought in silos, usage goes unchecked and ownership is unclear, organisations lose sight of what is actually running across the business. The result is waste, most of it hiding in plain sight.
Here are five numbers that describe the reality of SaaS waste in enterprise IT — and why fixing it takes more than better dashboards.
The average organisation uses 54% of the licences it pays for.
Nearly half the licence base is consumed by tools nobody is actively using. That comes from two directions. Overprovisioning — seats bought just in case, or bought in a block to hit a discount tier. And accounts never deactivated when people change role or leave.
SaaS is sold as flexible and scalable, but most enterprise contracts are fixed commitments that do not flex down when usage does. The commitment is annual; the usage decay is continuous.
Unused licences are close to impossible to spot without a single, accurate view of usage across every vendor and department. That is the specific gap: not “do we own this application”, but “did anyone open it in the last month”.
Very little of that is by design. Departments procure independently and often without involving IT or procurement. Marketing adopts a CRM add-on. HR implements a wellbeing platform. Sales trials a forecasting app. Each arrives with a different owner, a different contract and a different renewal date.
Decentralised buying does encourage teams to move quickly. It also produces duplicated tools, inconsistent data and a fragmented application landscape. Beyond a certain count, governance stops being a policy problem and becomes an arithmetic one: 305 renewal dates is not something a quarterly review can hold.
Against average annual SaaS spend of $55.7M.
That is more than a third of the SaaS budget returning nothing.
A large share of it never passed through procurement in the first place. Employees can sign up for a cloud tool with a corporate card and an email address, and most do it for entirely reasonable motives — they are trying to solve a problem quickly. But applications that live outside formal oversight bring compliance risk, duplicate functionality that is already paid for elsewhere, and spend that no budget owner recognises.
Multiply that across dozens of departments and you get a landscape that is hard to manage and effectively impossible to optimise, because the first step — knowing what exists — was never completed.
Worth being precise about this one: +393% is large enterprises specifically. Across all organisations, AI-native application spend grew 108%, and use of applications in the broader AI category grew 181%.
Whichever number applies to you, the direction is the same, and AI is arriving through exactly the route that produced the SaaS problem in the first place — individual sign-ups, expensed subscriptions, free tiers that quietly become paid ones. Except this time the tools frequently hold an OAuth grant against the company’s own document store.
This is the newest source of waste and the fastest-growing one, and it is compounding on a base that most organisations had not yet brought under control.
SaaS waste does not sit in isolation. The same pattern of committed spend outrunning governance shows up next door in cloud, and after five consecutive years of improvement the waste rate has gone back up.
That reversal is the warning sign. It suggests the growth in consumption — much of it AI-driven — is outpacing the maturity of the controls around it. Optimisation stops being a nice-to-have at the point where spend is compounding faster than the ability to govern it.
Here is the part that does not show up in any statistic: most organisations that identify waste act on only a fraction of it.
The problem is rarely a lack of information. Reports showing unused licences are the easy part. Acting requires a named owner, a defined process, and someone with the authority to remove a colleague’s access. Teams often lack the time or the mandate to reclaim licences, renegotiate a contract or reassign a seat, and so the finding is noted, carried forward and quietly re-noted next quarter.
This is why reclamation has to be built as automation rather than as a report. In CerteroX SaaS Management that means:
Discovery converges from three signals — identity provider sync, direct vendor connectors, and a browser extension that catches what neither of the other two can see. Forty-seven connectors ship today, resolving against a catalogue of more than 35,000 applications.
None of this suggests SaaS has failed. It remains one of the most useful shifts in enterprise IT in twenty years. But like any powerful resource it has to be managed deliberately.
The issue is not the number of applications, or the rate of spend growth. It is the absence of visibility, ownership and follow-through. Organisations that treat SaaS as part of one asset and cost position — alongside devices, licences, cloud and AI — rather than as a separate procurement problem are the ones that reduce waste and keep it reduced.
Other posts covering the same ground.
Most AI spend is not arriving as new invoices. It is arriving inside software you already own, priced per token, owned by nobody. Six commercialisation patterns, why per-seat governance cannot see them, and what to do instead.
FinOps began as a reaction to engineers being able to spend money faster than finance could see it. Here is how it got from there to a formal discipline that now covers SaaS, licensing and AI as well as cloud.
What a SaaS management platform has to do in 2026, how the market divides, and the six criteria to score your shortlist against — including where Shadow AI detection now sits as a first-class requirement rather than a tag.
Everything argued above is checkable. Name the publisher, the billing account or the platform you would argue with, and the session is built around it — the reasoning attached, not a summary slide.
No gated download at the end of it.