What two public audits prove about unused software licences
The Czech Supreme Audit Office found a ministry had left 28% of ~5,900 licences unused across six audited years, paying at least 10.5 million crowns just to maintain them. Germany’s federal auditor found more than €50 million spent on software nobody used — and one finding that matters more than the money.
Independent state auditors have twice put a hard, traceable number on unused software licences, and in both cases the same underlying failure appears: the organisation could not measure its own usage without asking the vendor. The euro figures are the headline; that second finding is the one worth acting on.
Do independent audits actually find unused software licences?
Yes — and the two clearest published examples come from national audit offices, not consultancies, which is what makes them citable. One is Czech, one is German. Both read the invoices themselves.
Two things they are not: they are public-sector cases, and each covers a single organisation over a long window. Neither percentage is a private-sector benchmark, and nobody should present them as one. They are existence proofs — evidence that this failure is real, measurable and survivable for years — not a rate you can apply to your own estate.
The Czech case: 28% of licences unused, six years running
The Czech Supreme Audit Office (Nejvyšší kontrolní úřad, NKÚ) examined the Ministry of Finance’s Integrated Information System of the State Treasury in control action KA č. 21/05, published 21 March 2022, covering 2015–2020. Its finding, in the auditor’s own words: *"v kontrolovaných letech 2015 až 2020 MF nevyužívalo v průměru 28 % z cca 5 900 zakoupených softwarových licencí, tedy zhruba 1 600 licencí"* — on average 28% of roughly 5,900 purchased licences went unused, about 1,600 seats.
The figure to hold onto is the smaller one. The ministry paid at least 10.5 million crowns in the audited period purely to maintain the licences it was not using. Not to buy them — to keep them alive.
Two handling notes, because the report contains several numbers that measure different things and mixing them produces a false claim. The 10.5 million crowns is maintenance on the unused portion; it is not the same as the total IISSP software maintenance for the period, and it has nothing to do with the cumulative acquisition figure the report also carries. Quote each one for what it measures, or quote neither.
What makes this case useful is not the size. It is the duration: the gap held for six consecutive audited years. This was not a purchasing error someone caught the following quarter. It was a steady state.
Source: NKÚ, control action KA č. 21/05
The German case: more than €50 million on software nobody used
Germany’s Bundesrechnungshof published *Mehr als 50 Mio. Euro für ungenutzte Software* (Bemerkungen 2021, supplementary volume, item 50) in 2022, auditing the Bundeswehr’s use of one vendor’s software between 2006 and 2019.
The findings, each measuring its own thing:
- At least €50 million spent on unused software over the period.
- €25 million of that was maintenance alone.
- At least €5 million per year in ongoing cost to maintain software that was not being used, under unchanged conditions.
- At least 20 cases in which software was bought and then never used, or used for years at a *"deutlich geringeren Umfang als lizenziert"* — a markedly smaller scope than licensed.
The auditor does not name the vendor, and neither do we. It is also worth reporting the disagreement honestly: the defence ministry argued it had acted lawfully throughout and rejected the finding of a budget-law breach; the Bundesrechnungshof reviewed that response and maintained its assessment.
The finding that matters more than the money
Both audits describe the same mechanism, and the German report states it plainly:
*"Die Bundeswehr verschaffte sich unabhängig vom Anbieter und dessen Ergebnissen keinen eigenständigen Überblick darüber, inwieweit sie dessen Software nutzte. Hierzu fehlten ihr Analyseinstrumente und die nötigen Kenntnisse."*
The organisation had no independent view of its own software usage. The only usage data it possessed came from the vendor’s own contractual audits — a right the supplier exercised three times, most recently in 2019. It lacked, in the auditor’s words, both the analytical instruments and the necessary knowledge to produce that picture for itself.
Sit with what that means commercially. Every decision about renewal, expansion and consolidation was being taken on a number supplied by the party whose revenue depended on the answer. Nobody was lying. The buyer simply had no second opinion.
Why the maintenance fee doesn’t care whether you use it
The German report also explains why unused licences are more expensive than they look. The annual maintenance fee depends solely on which and how many licences were purchased. In the auditor’s words: *"Ob die Bundeswehr die Software nutzt, ist hingegen nicht maßgeblich."* — whether the software is actually used is not relevant.
Worse, the ministry stated that the fee cannot be reduced after purchase and that maintenance on individual licences cannot be cancelled; licences can only be replaced by at least equivalent new ones. That is the shape of a great many enterprise agreements, and it is why "we will clean it up at the next renewal" is often not available as a plan.
An over-purchase is not a one-time mistake in that structure. It is an annuity paid to the vendor.
What this does and does not prove about your company
What it proves: software waste at scale is real, it is measurable by an independent party, it persists for years inside competent organisations, and its ongoing cost is driven by contracts that ignore usage entirely.
What it does not prove: that 28% is your rate, that any of this is typical of the private sector, or that your suppliers behave as the ones in these reports did. Two public bodies, two long windows, two extreme cases.
Three questions to ask before your next renewal
- Where does our usage data come from? If the honest answer is "the vendor’s report", you are negotiating with their account of your business. That is the single finding both audits share.
- What are we paying in maintenance on licences nobody opened this year? Not the purchase price — the recurring fee. In the Czech case that was the whole story; in the German case it was half the loss.
- Which contract terms actually let us act on the answer? Commitment length, billing frequency and downgrade rights decide whether a finding turns into recovered money or into a note for 2027.
We do not sell licences and have nothing to upsell, which is why the utilisation number we produce has no interest attached to it — and why we will tell you to cancel something.
Both auditors found the money. What they could not find was an organisation that knew what it was using.