Every year, finance teams sign off on SaaS renewals for tools that a significant portion of their workforce has never opened. The number we kept landing on, across 84 organizations and roughly 210,000 individual seat records, was 31 percent — nearly one in three licensed seats sitting dormant at any given 90-day window. That figure is not a rounding anomaly or a seasonal dip. It is a structural pattern, and it clusters in ways that are specific enough to be actionable. The organizations hit hardest are not the smallest or the largest; they sit in a particular band of headcount and IT maturity that makes them uniquely vulnerable to seat waste. What follows is an attempt to explain the shape of that vulnerability — where the waste concentrates, what the most-abandoned platforms share, and why the usual remedies mostly fail.
How we defined 'unused' and why the threshold matters ¶
Before the numbers mean anything, the definition has to be pinned down. For this analysis, a seat was classified as unused if the associated account recorded zero meaningful product events — not just logins, but feature-level interactions — over a rolling 90-day window. Login-only events were excluded because single sign-on configurations frequently generate automated login pings that have nothing to do with a human opening the tool. The 90-day window was chosen deliberately: it is long enough to exclude employees on leave or between projects, but short enough to capture genuine abandonment rather than occasional use. By that definition, 31 percent of seats across the full dataset qualified as unused. When the threshold was tightened to 30 days, the figure jumped to 44 percent. When loosened to 180 days, it fell to 19 percent. The 90-day mark is where the distribution had a natural inflection — the point at which 'not using it right now' starts to look indistinguishable from 'not using it at all.'
The mid-size anomaly: why companies between 200 and 2,000 employees waste the most ¶
The utilization curve across organization size is not linear. Small companies under 50 employees showed a 21 percent unused-seat rate — high, but explainable by founder-era tool sprawl where one person signed up for everything. Large enterprises above 5,000 employees came in at 27 percent, which is surprisingly manageable given procurement complexity. The worst band, consistently, was 200 to 2,000 employees: an average unused-seat rate of 38 percent, peaking at 43 percent for organizations in the 500-to-800 headcount range. The explanation appears to be a governance gap. Below 200 people, a single IT lead or ops manager can track what is actually being used. Above 2,000, enterprises have dedicated software asset management teams and vendor management processes. The middle band has neither. They have grown past informal oversight but have not yet built formal SAM infrastructure. Procurement is often decentralized — department heads buy tools independently — and renewal decisions are made by finance teams who look at invoice totals, not login telemetry.
What the most-abandoned platforms have in common ¶
Across the dataset, five platform categories accounted for 61 percent of all unused seats: project management tools, employee engagement survey platforms, sales intelligence add-ons, digital whiteboarding tools, and secondary communication channels (meaning any messaging platform that was not the organization's primary one). Three characteristics showed up repeatedly in the most-abandoned tools. First, they required behavioral change rather than workflow integration — platforms that asked users to go somewhere new, rather than embedding into existing processes. Second, they had been purchased at the departmental level for a specific initiative that concluded, but the license survived the initiative. Third, and most consistently, they lacked a 'forcing function': no workflow produced an error or a blocker if the tool was ignored. Tools that sat in the critical path of a process — a code review platform that blocked merges, a ticketing system tied to SLA reporting — showed utilization rates above 80 percent almost universally. Optional tools hovered near 50 percent at best.
The renewal cycle as a utilization trap ¶
One of the more counterintuitive findings was the relationship between contract length and waste. Annual contracts showed higher unused-seat rates (33 percent) than month-to-month arrangements (24 percent) — the opposite of what you might expect if you assumed that longer commitments created more pressure to extract value. The mechanism appears to be temporal distance: when renewal is 11 months away, there is no urgency to rationalize seats. By the time the 60-day renewal notice arrives, the budget cycle is already locked, and cancellation feels more disruptive than continuation. Multi-year contracts showed the highest unused-seat rates of all, at 41 percent, largely because the sunk-cost psychology is strongest when a deal cannot be unwound mid-term. The organizations with the lowest waste rates — consistently below 22 percent — were those that had decoupled their utilization reviews from their renewal cycles, running quarterly seat audits as a separate process rather than a pre-renewal scramble.
Onboarding length predicts long-term activation better than almost anything else ¶
Within platforms where we had access to activation event data — roughly 40 of the 84 organizations shared this level of telemetry — the single strongest predictor of whether a seat would still be active at 90 days was whether the user had completed a meaningful action within the first seven days of provisioning. Not a tutorial. Not a welcome email open. A real action: creating a project, running a query, publishing a post. Users who hit that threshold in week one were active at 90 days at a 74 percent rate. Users who did not were active at 90 days at only 19 percent. The implication for procurement teams is underappreciated: time-to-first-value is a vendor metric worth demanding during sales cycles. Vendors who cannot tell you what percentage of their new seats reach a meaningful action in week one are either not measuring it or not proud of the answer.
What the low-waste organizations actually do differently ¶
The organizations in the bottom quartile of unused-seat rates — those below 18 percent — were not all large, well-resourced enterprises. Several were in the 300-to-600 employee range, the same band where waste peaks in the broader dataset. What distinguished them was process, not headcount. All of them maintained a live software inventory updated at least quarterly, usually in a simple spreadsheet or a lightweight SaaS management platform like Torii or Productiv. All of them had a named owner for every tool — not a team or a department, a specific person whose performance review included a line about software ROI. And all of them had a documented deprovisioning trigger: a specific utilization threshold — typically fewer than two active sessions per month per seat — that automatically generated a review ticket. None of this is technologically sophisticated. It is organizational discipline applied to a problem that most companies treat as background noise.
The 31 percent figure is a floor, not a ceiling ¶
One caveat worth sitting with: the 84 organizations in this dataset are not a random sample. They are organizations that had enough telemetry available to make this kind of analysis possible — which means they were, on average, more instrumented and more aware of their software portfolios than a true cross-section would be. The actual unused-seat rate across all mid-size companies is almost certainly higher than 31 percent. It is simply harder to measure in organizations that lack the logging infrastructure to know what is being touched. The waste that shows up in spreadsheets is just the waste that is legible. The broader point is not that 31 percent is a precise number to anchor policy around — it is that the structural conditions producing it are consistent enough across organization types, contract lengths, and platform categories to suggest this is not a vendor problem or a user problem. It is a systems problem, and it has systems-level solutions.
The findings here are preliminary and the dataset is bounded — 84 organizations is a reasonable sample for pattern detection, not a definitive census. But the patterns hold tightly enough across organization size, industry, and contract type that the basic shape of the problem seems real. If anything, this analysis undersells the scale of the waste, because it only captures the waste that was measurable. The next phase of this work will look at whether vendor-side activation data matches the picture from the buyer side — early signals suggest it does not always, and that gap is its own story.