Shelfware is the single largest source of avoidable cost in most enterprise software estates. In our compliance assessments across 340+ engagements, the typical Fortune 500 carries 20–35% of total annual software spend in shelfware. The reasons are structural rather than situational — and the recovery levers are well understood for every major vendor.
Shelfware — software that is licensed but not deployed, or deployed but not used — is the single largest source of avoidable cost in most enterprise software estates. In our compliance assessments across 340+ engagements, the typical Fortune 500 estate carries 20–35% of total annual software spend in shelfware. That figure has been remarkably stable across industries, geographies and vendor mixes since 2016, which suggests it is not an artefact of any particular procurement style — it is a structural feature of how enterprise software is bought.
The reasons are predictable. Software is bought on multi-year terms, business priorities change inside the term, and the disposal mechanism (true-down, return, transfer) is harder to operate than the acquisition mechanism. The result is a slow accumulation: every renewal cycle adds licences that may be needed, and few renewal cycles remove licences that were not. Five years in, the shelfware ratio is structural rather than situational.
Shelfware comes in three forms and each has a different remediation path. The first is provisioned-but-unused: licences assigned to users who never log in. This is the easiest to find and the easiest to harvest. The second is deployed-but-overprovisioned: products where the feature edition exceeds actual usage (Microsoft E5 where E3 would do, Salesforce Enterprise where Professional would do, Oracle Enterprise Edition where Standard Edition 2 would do). The third is contracted-but-undeployed: licences purchased in a multi-year deal against a deployment plan that never materialised. Each of the three is invisible in standard procurement reporting; each requires a different data source to find.
Most CIOs are surprised by the first scan. The second scan is usually the one that drives action.
Provisioned-but-unused shelfware sits in the identity layer — Azure AD, Okta, Active Directory — joined against the application usage logs (Microsoft 365 audit logs, Salesforce login history, Oracle session records). Deployed-but-overprovisioned shelfware sits in the application feature telemetry (Microsoft 365 advanced auditing, Salesforce feature analytics, ServiceNow plugin usage). Contracted-but-undeployed shelfware sits in the entitlement register joined against the deployment record. Three different data sources, three different analytic patterns, three different cost-recovery levers.
Finding shelfware is not the hard part. Harvesting the savings is. The recovery levers are different for each vendor and each contract shape. For Microsoft, the levers are at renewal: reduce E5 seats to E3 (or vice versa where security workloads require it), true-down on non-renewal, and reshape the EA term. For Salesforce, the levers are the renewal license-type swap (Enterprise to Professional, full user to platform user) and the seat reduction at renewal. For Oracle, the levers are licence retirement at the support renewal moment (rare but possible) and metric migration. For ServiceNow and Workday, the levers are tied to the headcount band and the module mix.
The full shelfware programme toolkit covering discovery, recovery, and vendor-specific levers.
A shelfware elimination programme typically delivers in three phases. Phase one (months 1–3): discovery scan across the top five vendors, joining identity, usage and entitlement data. Phase two (months 3–6): recovery on the immediate wins — provisioned-but-unused licences harvested at the next renewal, edition right-sizing where feature usage supports it. Phase three (months 6–12): structural recovery on the contracted-but-undeployed shelfware, which usually requires a contract restructuring event. Total programme savings typically run 18–32% of the addressable spend, with the largest share captured in phase two.
The most useful programme metric is not the shelfware ratio at a point in time — it is the shelfware-creation rate per renewal cycle. The ratio tells you how big the problem is; the creation rate tells you whether the programme is closing the inflow. A managed estate operates below 4% shelfware-creation per renewal cycle. An unmanaged estate operates above 12%. The gap is recoverable, but only if the renewal cycle becomes the disposal mechanism rather than the accumulation mechanism.
We run discovery across identity, usage and entitlement data, and sequence the recovery against the renewal calendar.
Weekly compliance intelligence for IT leaders.
