The Software Stack Nobody Decided To Build
Somewhere in your company, a subscription renews this month for a tool that two people relied on in 2023 and nobody has opened since. The charge is small enough to clear the corporate card without a question. Multiply that by the number of times it has happened, and you start to see the real shape of most software stacks in the lower middle market. Nobody sat down and designed them. They accumulated.
Firms that track software usage put the average company at around 106 applications in 2024, down from a peak near 130 2 years earlier. For mid-market companies the count runs higher, closer to 254 apps in a portfolio by 2025. The studies measure things differently, so treat the exact figures loosely, but the direction is consistent, and it does not match how any owner I know would describe their own operation. Ask them how many pieces of software the business pays for and the guess is usually low by half.
How the stack builds itself
No single decision creates sprawl. A manager needs a scheduling tool and expenses it. A new hire brings a favorite app from her last job. Marketing signs up for a trial that quietly converts to an annual plan. Each of those is reasonable on its own. Added together over 5 years, with turnover moving people in and out, you get a stack that no one person understands and no one is accountable for.
The tools also overlap. You are very likely paying for two products that do project tracking, three that store files, and a couple that send email campaigns, because different teams solved the same problem at different times without checking. Finance sees line items, not functions, so the duplication never surfaces on a budget review. It surfaces when someone finally lists every login the company owns and reads it out loud.
What the sprawl actually costs
The wasted subscription money is the part you can see, and it is the smaller part. Industry estimates for unused or redundant software routinely land near a third of total spend. Worth recovering, but not the reason to care.
The real cost is that your information scatters. When customer data lives in one system, orders in another, and the true picture only exists in a spreadsheet somebody rebuilds every Monday, you have bought yourself slow answers and arguments about which number is right. Every extra system is also a door. It holds a copy of your data, it has a password, and it is one more thing that can be breached or simply forgotten when an employee leaves with access still live. A stack you cannot list is a stack you cannot secure.
There is a quieter tax too. People spend real hours moving information by hand between tools that were never connected, retyping into one screen what already exists in another. That work feels like productivity. It is mostly friction you are paying for twice.
Rationalizing without ripping everything out
This does not call for a transformation. It calls for an inventory and a couple of afternoons.
Pull every recurring software charge from the last 12 months. Put each one next to the job it does and the number of people who actually log in. The list alone will embarrass you a little, which is the point. You will find the ghost subscriptions, the three tools doing one job, and the app one former employee championed that the rest of the company quietly ignored.
Then make someone own it. Not to police every purchase, but to hold the map, so that adding a tool means retiring one and every login has a person attached to it. Consolidate where two systems genuinely do the same work, and resist the urge to replace five imperfect tools with one giant platform you will spend a year installing. The goal is a stack you can name, not a monument.
Do the inventory this quarter. Most owners who run it for the first time cut their tool count by a quarter or more and lose nothing they miss. The money back is nice. The clarity is the prize, because you cannot run a business well on systems you did not know you had.