Half the Software Licenses Companies Pay For Go Unused.

In plain terms: Two companies that track software spending measured what businesses actually use of what they pay for. One found organizations use about 54% of their software licenses, so roughly 46% sit paid for and untouched, averaging about $19.8 million a year wasted at large companies. The other found 15% of applications are completely unused and another 51% are barely touched. You are probably paying, right now, for tools nobody opens.

This week has been a pile of failure data: 95% of AI pilots with no measurable payback, companies quitting their AI projects at more than double last year's rate, and three out of four small businesses stuck "exploring" instead of running anything.

Today is the plainest version of the same disease. It has nothing to do with AI, it has been happening for years, and unlike the others, you can check it yourself in about ten minutes.

What the research says

Zylo runs an annual SaaS Management Index on the software spending it manages. The finding, repeated across years: organizations use only about 54% of the licenses they buy. The other 46% are paid for and unused, which works out to roughly $19.8 million in average annual waste at the enterprises they measure. Prior years landed at 49% to 51% used. The number moves a little. The pattern does not.

Vertice, a different company doing similar work, benchmarked it from another angle and found 15% of applications are completely unused, plus another 51% underutilized, meaning the company is using less than half the licenses it pays for.

Two independent vendors, two methods, same shape: about half of what gets bought never really gets used.

The honest note

Both of these are vendor benchmarks. Zylo and Vertice sell software-spend management, so "you are wasting a fortune on software" is a number that happens to help them. Their methodology is only partly public, and the data leans enterprise. That $19.8 million is emphatically not your number.

So treat this as directional, not gospel. What makes it worth your attention is not the precision. It is that two independent companies with different data landed on the same shape, and that every owner reading this already knows the drawer. You have subscriptions right now you would struggle to justify out loud.

Buying feels like progress

Here is why this connects to the rest of the week.

Buying is a decision you can make in an afternoon. You compare a few options, you put in a card, and something real happens: an invoice, a login, a tool in your stack. It feels like progress. You can tell your team you did something about the problem.

Wiring is the opposite. It is slow, unglamorous, and nobody claps for it. It means deciding who owns the thing, what job it does, and how you would know if it worked.

So companies buy tool number nine while tools one through eight sit there collecting dust. That is the same instinct that produced the 95%: the purchase is the moment of hope, and the wiring is the actual work. Acquiring is not adopting. It never was.

What this means for you

Before you buy the next tool, audit the ones you have. Pull up your card statement, list every subscription, and ask two questions about each: who actually opens this, and what does it do for money? Most owners find two or three they are paying for that nobody has touched in months. Cancelling those is the fastest money you will make this week, and it costs you nothing but ten minutes.

Then run the bigger version of the same idea, because software is not the only asset you own and ignore. The largest one in most small businesses is the customer list. You already paid for every name on it, through ads, referrals, and hours of your own selling. And most of it sits there, untouched, exactly like shelfware. Database Reactivation is that same move in reverse: instead of cutting what you pay for and do not use, you switch on the asset you already bought and never went back to.

Before you buy anything else, use what you already own. That is the whole lesson, and it has been true a lot longer than AI has.

FAQ

I am a small business. $19.8 million in waste is not my number. Does this even apply? The dollar figure does not, the ratio does. If you spend $600 a month across a dozen subscriptions and half go unused, that is a few thousand dollars a year for nothing. Same shape, smaller stage. The advantage is that you can actually check yours this afternoon, where a company with 400 apps cannot.

How do I tell shelfware from something seasonal I only use sometimes? Ask who opened it in the last 90 days and whether anything would break if it vanished. Seasonal tools have an owner and a date. Shelfware has neither, and the honest answer is usually "we bought that for a project that never happened."

Should I just cancel everything I am not using? Cancel the ones with no owner and no job. But the better habit is the rule going forward: before any new tool, name the person who owns it and the one number it should move. A tool with an owner becomes a system. A tool without one becomes next year's shelfware.