Shelfware quantification — the rigor that survives a CFO challenge.

The definition has to be sharp.

A shelfware estimate that fails CFO challenge usually fails because the definition is too loose to defend. "Licenses we don't need" is not a defensible definition. "Licenses with zero usage in the last 90 days against a deployment that allows in-period reduction at renewal" is. Sharpness matters because the action that follows quantification — pulling licenses, declining renewal expansion, restructuring contracts — depends on each finding being defensible in vendor and internal challenge.

The four-axis definition we use consistently: usage threshold (typically zero or below-threshold in a defined period), license type (must allow reduction in the relevant contract), use-case status (the underlying business need is gone, not paused), and business owner attestation (the owning function confirms the license is unneeded). All four axes must be true for a license to count as shelfware in the rigorous sense.

What you have to reconcile.

Shelfware quantification requires reconciling three data sources that rarely agree out of the box. The reconciliation work is most of the project; the analysis is straightforward once the data is clean.

Entitlement data

What the contracts actually say you can use — license type, quantity, term, geographic scope. The entitlement view should come from a master contract repository indexed by vendor, with every Order Form, Amendment, and License Identifier document captured. In our experience, even disciplined enterprises find 8-15% entitlement variance between their procurement records and their actual contractual position.

Deployment data

What is actually installed, provisioned, or configured. The deployment view comes from the SAM tool, the IAM platform, the vendor console (for SaaS), and the asset management system (for on-prem). Deployment data is typically the cleanest of the three but most often miscategorized — a license assigned to a user is not the same as a license being used by that user.

Usage data

What is actually being consumed. The usage view comes from the application's own telemetry (login frequency, feature consumption, transaction volume) and is the source most often missing entirely. SaaS platforms generally provide usage data through admin consoles; on-prem and legacy applications often do not, requiring inference from supporting infrastructure (database connections, network traffic, authentication events).

What the math usually shows.

Across 340+ engagements, the consistent shelfware pattern looks like this: 14-22% of total software portfolio value qualifies as shelfware under the four-axis definition, with material variation by vendor.

By vendor archetype

• Microsoft — 12-18% shelfware, concentrated in E5 over-allocation (users licensed at E5 consuming only E3 capability), unused security and analytics SKUs, and Visual Studio entitlement bundled in MSDN.
• Oracle — 18-30% shelfware, concentrated in Database options never enabled, named users assigned at a metric that exceeds actual usage, and full-use entitlement on Enterprise Edition where Standard Edition would meet the requirement.
• SAP — 15-25% shelfware, concentrated in named user types over-classified (Professional users who consume Limited capability), engine licenses for retired use cases, and database licenses on retired systems.
• Salesforce — 8-15% shelfware, concentrated in inactive seats and over-classified license types.
• SaaS long tail — 25-40% shelfware, the highest concentration of all categories, because the long tail rarely has named ownership and renewals proceed by default.

What converts finding into saving.

Quantifying shelfware does not save money; pulling the licenses, restructuring the contracts and avoiding the next true-up does. The conversion path varies by contract type. Subscription licenses can be reduced at renewal if the contract allows in-period reduction; perpetual licenses produce ongoing maintenance savings if the maintenance is dropped; SaaS seats can typically be pulled mid-term if the contract permits.

The discipline that separates successful shelfware programs from unsuccessful ones is the linkage between quantification and the renewal cycle. Findings produced 30 days before renewal rarely convert; findings produced 9-12 months before renewal convert at much higher rates because the contractual mechanics — reduction provisions, true-down clauses, renewal repricing leverage — actually have time to be applied.