Contents
Key takeawaysWhat shelfware isWhere it hides by vendorThe six step auditTurning waste into savingsWhat we have seenHandling the account teamContract terms to ask forWhen to startWhat to do nextFAQShelfware is paid software that delivers no business value. A 90 day audit, started well before the renewal and sequenced by spend, finds it; only the share your contracts let you reduce comes back as savings.
- Shelfware has three shapes. Licenses bought and never deployed, deployed and never used, or left behind after a pilot, and each needs different evidence.
- It is common and concentrated. It ran at 12 to 30 percent of annual license spend in our audits, with about half of it in three vendors.
- The method has six steps. Scope, reconcile entitlement, prove deployment, pull usage telemetry, get owner sign off, then recommend.
- Only one layer is negotiable this year. Promise finance the portion recoverable at the next renewal, and that depends on the reduction rights in each contract.
- Cancellation is one option of five. Reduce, mix shift, sublease internally, cancel or redeploy, matched to the pattern behind each finding.
- Fix the cause or repeat the audit. Over commitment, missing offboarding and open ended pilots produce the same findings every year until the process changes.
What is shelfware, and what forms does it take?
Shelfware is the share of paid software licenses that delivers no business value. Almost every finding in a shelfware audit falls into one of three patterns, and the pattern tells you which recovery option will work.
- Purchased and never deployed. Licenses sit on the contract and the invoice but were never assigned to a user, server or workload.
- Deployed and never used. Licenses were assigned, usually at onboarding, and the person or system never logs in or calls the service.
- Abandoned after a pilot. A team deployed the product for a trial, the pilot ended, and the licenses stayed on the bill.
The distinction matters in the room with the vendor. Never deployed and deployed but idle need different evidence, and they point to different remedies, so an audit that lumps them together as "unused" gives the account team an easy way to dispute the whole list.
Why does the same shelfware come back every year?
Each pattern starts with a gap in process, and carelessness is rarely the cause. If you remove the waste and leave the process alone, next year's audit produces the same report with different names on it.
| Pattern | Root cause | Process fix |
|---|---|---|
| Purchased and never deployed | Over commitment in the discount window, where the vendor offers a deeper headline discount for a larger commitment | Size commitments to a deployment plan with named owners, and price the smaller volume before you accept the larger one |
| Deployed and never used | Provisioning without offboarding discipline, so licenses assigned at onboarding are never reclaimed | Tie license removal to the leaver and role change process in your identity system |
| Abandoned after a pilot | Pilots with undefined success criteria, bought on contracts that auto renew | Write success criteria and an end date into every pilot, and turn off auto renewal before signing |
Where does shelfware hide at each major vendor?
Shelfware concentrates. In the audits we have run, roughly 50 percent of the waste sat in three vendors, and the rest was spread thinly across the long tail of smaller publishers. The table shows where we usually find it at the large publishers and the recovery route that fits.
| Vendor | Common shelfware hotspot | Primary recovery route |
|---|---|---|
| Microsoft | Top tier suites over assigned; per user platforms sitting idle | Mix shift to a lower suite; per app licensing |
| Oracle | Unlimited agreement over deployed at certification; options enabled but unused | Right size at certification; deselect unused options |
| SAP | Indirect access named users; conversion buffer | Conversion credit; contract reduction right |
| Salesforce | Top tier users with no login in 90 days | Reduce at anniversary; downgrade the SKU |
| IBM | Measured capacity spread over unused virtual machines | Sub capacity rebaseline; deactivate workloads |
| Broadcom and VMware | Cores over committed; storage entitlement in excess | Reduction at renewal; alternative platform |
At Oracle, a ULA carries a fixed support fee however little you deploy, so products in the agreement that are barely used cost the same every year. Deal with them at the certification or renewal decision. Oracle's support policies reprice the licenses left on an order when you drop part of it, so a mid term cut often saves little.
For vendor detail, see our guides to Microsoft E5 shelfware, Salesforce shelfware, Oracle shelfware in the license position and IBM shelfware reduction.
Why should the audit start with the three largest publishers?
Because the concentration works in your favor if you sequence by spend. Rank publishers by annual cost and start at the top, and most of the recoverable money surfaces in the first two weeks of the cycle.
It also shrinks the list of people you need. You deal with three publisher relationships instead of 30, and each has a named business owner who can sign off an unused set without escalation. Audits sequenced by convenience, starting with whichever vendor exports data most easily, spend their first month on the long tail.
Enterprise FinOps and cost optimization brief
The audit method, quantification layers and recovery options in one download.
Get the white paper →How do you run a shelfware audit, step by step?
A shelfware audit has six steps, run in order, each with its own evidence. The method is mechanical on purpose: a finding that anyone can reproduce from the same records holds up when an account team challenges it.
- Scope. Agree the vendor list, the license types and the time window, typically the trailing 12 months, before anyone pulls data.
- Reconcile entitlement. Count what you contracted per SKU, per vendor and per cost center, taken from the signed contracts and order forms. The vendor portal shows what is provisioned, which is a different number.
- Prove deployment. Use provisioning records, identity provider exports and vendor admin console exports to establish what was assigned to users and systems.
- Pull usage telemetry. Collect login records, API call counts and feature usage flags across the window. This separates deployed and used from deployed and idle.
- Get business owner sign off. A named manager confirms, in writing, that each unused set is unused and not needed.
- Recommend. Each signed set becomes one of reduce at renewal, mix shift, sublease internally, cancel or redeploy.
We run this as part of our software spend management work, and the diagnostic side sits in the software spend assessment.
Where does the evidence for each step come from?
Most of the data already exists in your own systems. The records we ask for first:
- Contracts and order forms. The entitlement baseline for step 2, including any reduction or swap clauses.
- Microsoft 365 and Entra ID. License assignments from the Microsoft 365 admin center, usage reports under Reports, and the last sign in date per user from Entra ID sign in activity, which needs Entra ID P1 or above.
- Salesforce. The LastLoginDate field on the User object and the Login History report, filtered by license type.
- Oracle Database. The DBA_FEATURE_USAGE_STATISTICS view, which records which options and packs have been used on each database.
- IBM. IBM License Metric Tool (ILMT) reports, which show the sub capacity peak per product and the virtual machines behind it.
- SAP. System measurement (USMM) results and the License Administration Workbench consolidation, which show named users by type.
- SaaS generally. Last authentication dates from your single sign on provider, which cover every application behind it in one export.
A SAM tool helps with discovery on premises, but you do not need one to start. Our SAM tools guide covers when one pays for itself.
Why does business owner sign off decide the outcome?
It turns an analyst's spreadsheet into a decision a named person owns. When the account team disputes your usage data, a signed confirmation from the manager who runs that team carries more weight than any report.
The signature also gives procurement the authority to put a reduction request on the table and defend it. An audit that stops at the spreadsheet produces analysis and no savings.
How do you turn shelfware into a smaller renewal commitment?
Quantify it in three layers and promise finance only the middle one. Programs that mix up the layers lose credibility with the CFO quickly.
- Annual cost of the shelfware. The headline number, and the one you can do least about this year, because most of it is contractually committed for the current term.
- The portion recoverable at the next renewal. The only number worth promising. It depends on contract shape more than on the size of the waste: a license you can reduce at anniversary is recoverable, and one inside a multi year commitment with no reduction right is not.
- The multi year value of not carrying it forward. The number that justifies fixing the process instead of repeating the audit.
What does this look like on a $10 million software budget?
Take a hypothetical company spending $10,000,000 a year on licenses across 30 publishers. The audit finds $2,000,000 of shelfware, or 20 percent. The table shows how that becomes a recoverable figure.
| Finding | Annual cost of shelfware | Contract shape | Recoverable at next renewal |
|---|---|---|---|
| CRM: 250 top tier users with no login in 90 days, $2,000 per user per year. Owners sign off 200. | $500,000 | Reduction right at anniversary | $400,000 (200 x $2,000) |
| Productivity suite: 1,000 users on the top suite using only lower suite features, $300 price gap per user per year | $300,000 | Mix shift allowed at renewal | $300,000 |
| Virtualization: cores over committed | $250,000 | Multi year commitment, no reduction right, two years left | $0 this cycle |
| The other 27 publishers | $950,000 | Mixed; some renew in the next 12 months | $180,000 |
| Total | $2,000,000 | $880,000 |
The three largest findings hold $1,050,000, just over half the shelfware. The recoverable figure is $880,000, or 8.8 percent of the budget, less than half the headline number. On a three year renewal term, not carrying that $880,000 forward is worth $2,640,000, before counting the virtualization excess once its commitment ends.
Which recovery option fits which finding?
Five options cover the findings. Match the option to the pattern:
- Reduce at anniversary. Where the contract gives a reduction right. The cleanest option for never deployed licenses.
- Mix shift. Move users from a high tier they do not exercise to a lower one that covers their actual behavior. The saving is often larger than the seat count suggests, because every shifted user carries the price gap.
- Sublease internally. Reallocate entitlement to a team that would otherwise buy its own.
- Cancel. Only where the contract permits it and the business owner has signed off.
- Redeploy. Where the capability is wanted somewhere it was never provisioned, often the fix for pilot leftovers.
Should you cancel every license the audit flags?
The common advice is to cut everything unused at the first opportunity. We disagree with it as a default. Mix shift and redeploy keep the capability while removing the cost, which makes them easier to get through your own business units.
Blanket cancellation also carries a price risk. If the need returns, you buy back at whatever the vendor quotes that day, without the discount you gave up. Cancel where the owner has confirmed there is no need, and use the other four options everywhere else.
What have we seen in shelfware audits in 2024 and 2025?
Across roughly 60 to 80 software portfolios we audited for shelfware between 2024 and 2025, unused entitlement was never a rounding error. Three patterns recurred:
- Shelfware ran at 12 to 30 percent of annual license spend. It sat at the higher end on the large per seat platforms and lower on the infrastructure side.
- About half the waste sat in three vendors. The long tail of smaller publishers held the rest, in amounts that rarely justified the same effort.
- Reclaimed and renegotiated spend returned 8 to 18 percent of the budget within one renewal cycle. That funds the audit program many times over.
The difference between the shelfware found and the budget returned is contract shape. Licenses locked into a multi year term without a way to reduce stay on the invoice until it ends, which is why the recoverable figure is always the smaller one.
The audit finds the volume, quantification turns it into recoverable dollars, and the recovery options turn those dollars into a smaller renewal. None of it works in the other order.
What will the account team say when you ask to reduce?
Expect pushback on the evidence first and the commercial terms second. The lines we hear most often, and the replies that work:
- "Our portal shows those users as active." The portal shows assignment. Your telemetry shows no login across 12 months, and the business owner has signed off. Ask the vendor for its own login data if it disputes yours.
- "If you reduce volume, your discount drops." Ask to see the discount tiers in the contract, then price both options. A smaller discount on fewer licenses often costs less in total, and a price hold on the remaining units is a fair request.
- "Those users are seasonal and will need access next quarter." The manager who runs the team has confirmed otherwise in writing. Offer to add licenses back later at the same unit price.
- "Swap the unused licenses into credits for our new product instead." Credits for a product with no measured demand recreate the purchased and never deployed pattern. Accept a swap only where a named owner has a deployment plan.
- "The contract does not allow reductions until the term ends." Check the clause yourself. If it holds, record the finding for the next renewal and use it in that negotiation instead of conceding it.
What contract terms stop shelfware from building up again?
The next contract decides whether this year's audit has to be repeated. Ask for these terms at renewal:
- An anniversary reduction right. A stated share of licenses you can drop each year without repricing the rest. It is what makes shelfware recoverable at all.
- A swap or mix shift right. Permission to move users between tiers or products at the contracted discount, so a tier that is too rich can be corrected without a new negotiation.
- A price hold on additions. Licenses added later at the current unit price. This removes the vendor's argument that you should keep spare licenses just in case.
- Pilot terms with an exit. A fixed end date, written success criteria, and conversion to paid licenses only on your written order.
- No auto renewal without a quote. The renewal quote, with current counts, must arrive before your notice deadline.
When should a shelfware audit start before a renewal?
The cycle takes about 90 days from scope to action, so it has to start well before the renewal it is meant to inform. Starting alongside the renewal leaves you with findings the vendor will not accept in time. The schedule below is an illustrative plan for one cycle.
| Days | Work | Output |
|---|---|---|
| 1 to 10 | Scope: publishers ranked by spend, license types, trailing 12 month window | Scope agreed with finance and procurement |
| 1 to 14 | Contracts and admin console exports for the top three publishers | First view of where the waste concentrates |
| 11 to 40 | Entitlement reconciliation and deployment proof for the rest | Reconciled counts per SKU, vendor and cost center |
| 30 to 60 | Usage telemetry across the window | Deployed and idle sets, grouped by business owner |
| 55 to 75 | Business owner sign off | Signed unused sets |
| 70 to 90 | Recommendations and notices to vendors | Reduction and mix shift requests ready for the renewal |
Count back from the notice deadline, not the renewal date. If a contract requires reduction notice 90 days before renewal, the audit needs to start at least 180 days, about six months, before the renewal date. Our renewal program keeps that calendar across all your publishers.
What to do next
- Rank your publishers by spend. Start the audit with the top three and collect their contracts and admin console exports in the first two weeks.
- Pull the signed contracts. Reconcile entitlement per SKU, per vendor and per cost center from signed documents.
- Keep deployment and usage apart. Never assigned and assigned but idle are different findings with different remedies.
- Get sign off in writing. One named business owner per unused set, before anything goes to the vendor.
- Report the recoverable number. Tell finance what the next renewal can return, and show the multi year value separately.
- Fix the process behind each pattern. Commitments sized to deployment plans, offboarding tied to identity, pilots with exit criteria. Our spend management team can run the cycle with you.
Frequently asked questions
What counts as shelfware?
Any license you pay for that delivers no business value. That includes licenses bought but never assigned, licenses assigned to people or systems that never use them, and licenses kept after a pilot ended. Licenses that are used, but at a higher tier than the user needs, belong in the same review because a mix shift recovers them.
Where is shelfware concentrated?
In the largest contracts. Rank your publishers by annual spend and the top three will usually hold the bulk of the waste, while the long tail of smaller vendors holds small amounts each. Start at the top and the audit pays for itself early, even if the smaller publishers never get the same depth of review.
What does a shelfware audit actually involve?
Six steps in a fixed order: agree the scope, reconcile contracted licenses from the signed documents, prove what was deployed, measure usage over a trailing 12 month window, get each unused set confirmed by its business owner, and recommend an outcome. Most of the effort sits in the data collection, which is why scope comes first.
How much is realistically recoverable?
Across the audits we ran, 8 to 18 percent of the software budget came back within one renewal cycle. That is well below the shelfware found, because licenses inside a multi year commitment with no reduction right stay on the invoice until the term ends. Those findings still count: they shape the next negotiation.
Why does business owner sign off matter?
Usage data alone invites a debate about seasonality, planned hiring and data quality. A written confirmation from the manager responsible for the users closes that debate, gives procurement authority to request the reduction, and makes the decision the business's own, which matters if someone later asks for the licenses back.
What are the recovery options?
There are five: reduce at the anniversary, mix shift to a lower tier, sublease internally to another team, cancel, or redeploy where the product is wanted but was never provisioned. The pattern behind each finding decides which one applies. Pilot leftovers often suit redeployment, while licenses never deployed suit reduction.
Why do the same findings reappear each year?
Because removing waste does not change the processes that created it. Commitments sized for a discount instead of a deployment plan, licenses that stay assigned after people leave or change roles, and pilots that auto renew with no success criteria will rebuild the shelfware within a year.
Do you need a SAM tool to run a shelfware audit?
No. The first audit can run on signed contracts, admin console exports and your single sign on provider's last authentication dates. A SAM tool earns its cost mainly for on premises discovery, such as servers and virtual machines, and for keeping the numbers current between audits.