The 2026 SaaS Waste Breakdown: The True Cost of Rented Software
Nearly half of every SaaS dollar buys nothing. Here is where the money actually goes, and why owning your tools beats renting them.
- 01Up to 48% of the average company's SaaS budget is wasted on unused or duplicate licenses.
- 02The average organization only actively uses 54% of the SaaS licenses it pays for.
- 03Underutilized licenses now account for 51% of all SaaS seats, driving the majority of waste.
- 04Building internal tools instead of renting them eliminates per-seat licensing waste entirely.
How much of the SaaS budget is wasted
Somewhere between 25% and 48% of the average company's SaaS budget is pure waste, going to unused licenses, duplicate tools, and forgotten auto-renewals.12 At the enterprise level that waste averages $19.8 million a year in unused licenses alone.1 This is not a rounding error in a budget line. It is a structural leak in how companies buy software.
The headline numbers
Zylo's 2026 SaaS Management Index found that the average organization only actively uses 54% of the SaaS licenses it pays for, up from 47% the year before.3 The other 46% sits idle. Scaled by company size, Zylo pegs annual wasted spend at $3.8 million for companies under 500 employees, $9.5 million for 501 to 2,500 employees, $29.8 million for 2,501 to 10,000, and $80.6 million for organizations above 10,000 employees.1
Vertice's Q2 2026 data, drawn from more than $75 billion in processed spend, tells a similar story from a different angle: 65% of all SaaS licenses are either entirely unused or underutilized, up from 62% a year earlier.4 The interesting shift is where the waste is coming from. Fully abandoned "shelfware" actually improved slightly, falling from 15% to 14%. Underutilization, meaning less than half of purchased seats on a tool are active, jumped from 47% to 51% and now accounts for the majority of tracked licenses.4 Companies have gotten marginally better at killing dead tools. They have gotten worse at right-sizing the ones still limping along.
Varisource puts a sharper number on company size: organizations with more than 200 employees waste 48% of their software spend, and the average company manages 305 separate applications, not the 15 or 20 most leaders guess.2 Ramp's research lands in the same range, citing 53% of SaaS applications going underutilized or unused, with only 34% of subscriptions actively used at all.5
Why the number keeps climbing
Three mechanics drive this, and none of them are new. What's new is the scale.
Decentralized buying. Almost anyone with a corporate card can expense software now. Zylo notes this is a primary source of duplicate spend: an employee doesn't know an enterprise contract already covers a tool, so they buy their own version.1 Varisource frames the same problem as shadow IT, estimating it accounts for 35 to 40% of total SaaS spend at most companies, largely because line-of-business teams now outspend IT departments on software.2
Underutilization, not abandonment. The Vertice data is the clearest signal here. It is not that tools go completely unused. It is that companies buy 100 seats and 45 people log in.4 That gap is harder to spot than a fully dead subscription because the tool shows activity. It just shows far less than what was purchased.
Auto-renewal by default. Renewals lock in the same overprovisioned contract for another term unless someone intervenes 90 to 120 days ahead of the date.2 Most companies do not have that calendar built, so the waste compounds year over year rather than getting caught and corrected.
The compounding effect
Run the Zylo enterprise-scale figure forward and the arithmetic gets uncomfortable fast. An $80.6 million annual waste figure for a company over 10,000 employees is not a one-time hit.1 It repeats every year the underlying process stays broken, and it grows as headcount and application count grow with it. Varisource's finding that the average company runs 305 applications is the other half of the problem: more tools means more renewal dates, more overlapping functionality, and more places for a duplicate purchase to hide.2
Gartner's warning, cited by Varisource, is blunt: organizations that fail to get centralized visibility into their SaaS portfolios will overspend by at least 25% through 2027.2 That is not a forecast about bad actors. It is a forecast about the default state of an unmanaged subscription stack.
Cutting the waste versus fixing the model
Most of the standard advice, license audits, usage thresholds, renewal calendars, vendor consolidation, works. Ramp reports that a first-pass software audit typically uncovers 15 to 20% in recoverable spend just from tightening these processes.5 That is real money and worth doing regardless of anything else in this piece.
But audits treat the symptom. The underlying reason companies overbuy seats and overpay for tools they barely use is that SaaS pricing is built around per-seat licensing for software that was never designed around any single company's actual workflow. You rent a generic tool sized for a worst-case headcount, and the gap between what you provisioned and what you use becomes the waste line every report above is measuring.
That is the argument for building instead of renting the pieces of the stack that are actually simple internal workflows: an intake form, an approval chain, an inventory tracker, a reporting dashboard. These are exactly the kind of tools companies buy an entire SaaS seat count for, then use at 50% utilization. Platforms like Remy let a team describe what they need and get a working internal tool, deployed with its own database, permissions, and audit trail, without paying a per-seat license for capacity nobody uses. Ownership does not have a utilization rate to worry about. You built it for the workflow you actually have, not the one a vendor priced for.
This is the same logic covered in Build vs. Buy in the Agent Era: the decision to build isn't about avoiding all software costs, it's about matching the cost structure to the actual shape of the work. Renting makes sense for genuinely complex, high-stakes categories like your CRM or ERP. It stops making sense for the long tail of internal tools that make up a meaningful share of that 305-application sprawl Varisource measured.
What to do with this in the next 90 days
Start where the money already is. Pull the renewal calendar for the next quarter and flag anything under 50% seat utilization before it locks in again.4 Cross-reference expense reports against your approved vendor list to surface shadow purchases hiding on individual credit cards.2 Then, separately, ask a harder question about the tools that survive that cut: how many of them are rented licenses solving a problem simple enough to own outright. That second question is where the compounding savings live, because unlike a renegotiated contract, a tool you own doesn't have a waste rate to measure next year.
Estimates range from about 25% to 48%, depending on company size and methodology. Vertice puts total unused or underutilized licenses at 65% as of Q2 2026, while Zylo finds the average organization only actively uses 54% of the licenses it pays for.
Underutilization, not full abandonment. Vertice's 2026 data shows underutilized licenses (less than half of purchased seats active) rose from 47% to 51% year over year, while fully unused shelfware actually declined slightly. The waste is increasingly hiding in tools that look active but are massively overprovisioned.
Zylo's 2026 SaaS Management Index puts the average at $19.8 million a year across all organizations, scaling from $3.8 million for companies under 500 employees to $80.6 million for companies above 10,000 employees.
Yes. Ramp reports that a first-pass software audit typically uncovers 15 to 20% in recoverable spend. Companies armed with pricing benchmarks routinely negotiate 15 to 30% savings at renewal.
Per-seat SaaS pricing forces companies to provision for worst-case headcount, which is exactly why utilization runs so low. A tool built and owned for a specific internal workflow has no seat count to overprovision and no license utilization rate to track, so it removes the waste mechanism rather than just trimming it.
