ADP, the payroll company, started out in 1949 with a van. It came around to your business, picked up your time sheets and punch cards, drove them off to a computer you could never afford, ran the payroll there, and drove the results back. For decades that was the arrangement, and for a small business it was the only arrangement on offer. The computer belonged to somebody else.
Punching business records onto cards, around 1940. The machine that read them lived somewhere else. U.S. National Archives, public domain.
It sounds quaint now, but that van is an honest picture of how small business software has worked for most of the last seventy years. A computer of your own cost millions, and the custom software to run on it cost as much again. Owning the system your business ran on was a Fortune 500 luxury. Everybody else rented time on somebody else’s machine, or ran the whole operation on paper.
An IBM 704 at the NACA Langley lab, 1957. This is what owning a computer looked like. NASA, public domain.
Even as computers shrank, the price wall barely moved. In 1988 IBM launched the AS/400, the machine that was supposed to bring computing to smaller companies, at $36,000, call it $95,000 after inflation, and that was before anyone wrote a line of software for it. Mid-size distributors and dealers bought them. The plumber, the dental practice, and the shop on the corner never got close.
The golden window
Then, for about twenty years, the wall came down.
The PC did it. By the late eighties the machine on the desk cost a couple thousand dollars, and most businesses already owned one, bought for the spreadsheet and the word processor. On top of it appeared a new kind of tool: database builders like dBase, FoxPro, and later Microsoft Access, which let one skilled person build in weeks what used to take a team of programmers months.
A personal computer fair in Amsterdam, 1987. The booth sign behind them sells custom software for optician shops, plus programmer consultancy. Bart Molendijk / Anefo, Nationaal Archief, CC0.
A whole trade grew up around that fact. By 1984 there were already more than a thousand companies selling custom dBASE development, and most of them were one or two people working their own town. The profession was so local it never picked up a household name. I think of it as the local app guy, the same way a town had its local electrician. He would sit down with the owner of a plumbing company, learn how the jobs actually flowed, and build a job tracker that fit like a glove. A patient scheduler for the practice. Work orders for the shop. Adjusted to 2026 dollars, a system like that usually landed somewhere between five and thirty thousand. Paid once. The business owned it outright and ran on it for years.
I want to underline what those systems were, because it matters. The local app guy built the operation itself: invoicing, dispatch, inventory, patient records, the screen your employees opened every morning to do the work. The web hadn’t happened yet, so there were no websites to sell and no funnels to tune. His entire job was making your business run better, and hundreds of thousands of small businesses bought exactly that. The only thing that had ever stood between a small business and its own system was the price, and for twenty years the price was right.
What closed it, and what filled the hole
The web closed it.
A desktop database ran on the PCs you already had, plus a shared file on the office server, administered by nobody. A web application needed hosting, security, backups, a database server, and someone to keep all of it alive. The one-person build became a team build again, and the price snapped back to fifty thousand and up. Expectations moved at the same time: work stopped happening in one room, and a file on the office server can’t follow you to the second location or out to the truck. The tools were orphaned too. Microsoft bought FoxPro and wound it down, and Access never got a real web version. So the trade faded. Some of those consultants retired, and a lot of them started selling websites, because that’s what the web made sellable. The guy who once built your dispatch system became the guy doing your marketing.
The hole he left got filled by subscriptions, which was the sensible outcome. Thirty dollars a month with nothing upfront beats a ten-thousand-dollar bid in almost every conversation a busy owner has time for. But rented tools are built for the average business, so each one covers a slice, and today a company with fewer than fifty employees runs on about sixteen of them. At fifty-person companies, software spend now runs around $580 per employee, per month. Even a twenty-person shop at $150 to $300 a head is paying $36,000 to $72,000 a year, every year, for a pile of tools that mostly don’t talk to each other, with somebody on the team spending afternoons moving the same numbers between them.
None of that was a mistake. Renting has been the rational choice for twenty years, the same way mailing your punch cards off was rational in 1960, and if that’s your stack, you’re in good company with nearly every business your size. It’s just worth seeing the old shape in it. Your business bends to fit the tool, the data lives on somebody else’s computer, and the meter never stops. The van never went away. It became a subscription.
The window just reopened
The 1985 window opened because three costs collapsed at once. The machine got cheap, since the PC was already on the desk. The tool got simple, since dBase and its cousins replaced the programming team. And with those two down, the labor collapsed to one local person you could afford.
The same three costs have now collapsed again. The machine is the browser and the cloud, already on every desk and in every pocket you employ. The tool is AI, which has changed what one experienced developer can build and, just as important, maintain. Across the industry, skilled development cost has come down by roughly a third to a half, and small builds have fallen the furthest. The custom system that needed a six-figure budget ten years ago is back inside a small-business number. I put current, real-world numbers on that in what custom software actually costs now.
And this version is better than the one we lost. The 1994 job tracker lived on one office machine, got fragile the moment two people opened it at once, and could never leave the building. Its 2026 equivalent is a real web application. The crew opens it on their phones, the office watches the same live numbers the owner does, and it gets maintained like a serious piece of software instead of rotting on a PC under someone’s desk.
Rent the commodity, own the operation
The old trade had its limits figured out, and they still hold. Nobody should own everything. Accounting is a solved, commodity problem and QuickBooks is genuinely good at it, so keep renting that. Payroll is solved too. ADP is still here all these years later, minus the van, and still worth it. When a rented tool fits a commodity job, rent it gladly.
The part worth owning is the operation: the quoting, the job tracking, the scheduling, the handoffs between the field and the office, the numbers you run the business by. In the businesses I work with, that is where the hurt shows up first, in the gaps the rented tools leave between each other. No vendor can build that part in advance, because it’s the part that makes your business your business. That is the piece the local app guy used to build, and it is affordable again.
And there’s no deadline on any of it. If your stack works and nothing hurts, change nothing, and don’t let anybody make you feel behind for running on spreadsheets and subscriptions in the meantime. A reopened window just means owning your own system is an option again. Options can wait until you’re ready.
Why I’m telling you this
I run a small shop called Allen App Tools, and my job is the one the FoxPro consultant had in 1994: sit with the owner, learn how the work actually flows, build the system that fits it, and hand over the keys. I learned this history only recently myself, and it was oddly encouraging. Small businesses owning their systems has been done before, at real scale, for twenty years. The only thing that ever made it rare was the price.
If you want to see what one of these looks like in real life, read the story of a fence and deck company that went from six rented tools to one app it owns. And if you’re wondering whether your business is anywhere near that line, book a call and we’ll talk it through honestly, including the real possibility that you should change nothing yet.
A small business owning its own system was normal once. The price is what made it rare, and the price just changed.