The number you're holding was right
If you've been quoted $150,000 for a custom system, or told it would take a year, or watched an ERP implementation go sideways at your business or a mate's, you did not misunderstand anything. That was the market. For roughly twenty years, those numbers were accurate, and anyone who told you otherwise was either inexperienced or about to disappoint you.
So the first thing worth saying is that the scepticism is earned. We are not going to argue that you were wrong. We are going to argue that the thing you were right about has changed, which is a different claim and a much easier one to check.
Why it was genuinely that expensive
Building software used to be almost entirely labour. That is the whole explanation, and everything else follows from it.
A custom operations system is not one program. It is a few hundred small ones. Every screen has to be drawn. Every form has to validate what someone typed into it. Every list has to sort, filter, page and export. Every table in the database has to be designed, then connected to the four other tables it relates to. Every report has to be assembled from those tables and then made to print properly. Every permission has to be checked in every place it matters.
None of that work is clever. Most of it is close to clerical. But it had to be typed, by hand, by someone on a day rate, and there was a great deal of it. A competent developer might produce two or three finished screens in a good week. A manufacturing system needs sixty of them. Multiply that out, add a project manager and a business analyst because a team that size cannot coordinate itself, and $150,000 stops looking like a markup and starts looking like arithmetic.
You weren't paying for software. You were paying for hours. The software was almost a by-product.
What changed, in three parts
1. The tooling collapsed the timeline
The repetitive scaffolding, the database layer, the forms, the CRUD screens, the standard reports, is now generated and then reviewed, rather than typed from scratch. Work that took a developer a week takes an afternoon.
It matters what that sentence does and doesn't say. Nobody is cutting corners and nothing is being left out. The output still gets read, tested and corrected by a person who is accountable for it, and that review is real work. What has gone is the typing, which was never the valuable part and was always most of the bill.
The judgement about what to build takes exactly as long as it always did. Only the execution got faster, and execution was where the money was going.
2. Infrastructure got absurdly cheap
Running a business system used to mean owning a server. Then it meant renting a virtual one, which you still had to patch, monitor, back up, secure and replace. Either way there was a machine somewhere that was your problem, and a line item to match.
Your system now runs on Cloudflare's edge network, the same infrastructure that carries a substantial share of the world's internet traffic. There is no server. There is nothing to patch. Capacity that would have required a rack and a systems administrator fifteen years ago costs a small fraction of a single per-user subscription, and it is more reliable than either.
3. The bottleneck moved
This is the part that actually matters and the part most people miss.
The constraint on building a business system used to be "can we build this?" It isn't anymore. The constraint is "do we understand this business well enough to build the right thing?"
Understanding a factory is not a technical problem. It is a matter of sitting with the person who does the purchasing and working out why they keep a private spreadsheet next to the ERP, then finding out from the storeman that the reason is a receiving rule that was never written down. That work has always been the difference between a system people use and a system people work around. It was also, historically, the part that got compressed, because the build was so expensive that nobody wanted to spend money on talking first.
Now it's the main event. That is a better arrangement for everyone, and it is the single biggest reason these builds succeed.
What that does to the buy-versus-build decision
The old economics forced a bad trade, and it was so universal that most operators stopped seeing it as a trade at all.
Custom was too expensive, so you bought something off the shelf. Off the shelf was built for a general case, which meant it fitted your business somewhere between sixty and eighty percent. The remaining twenty to forty percent became spreadsheets, workarounds, a person whose actual job is reconciling two systems, and a set of things everybody knows the software can't do so nobody asks anymore. Then you paid for that, per user, per month, forever, with the price rising every time you hired somebody.
That trade made complete sense when the alternative was $150,000 and a year of upheaval. It does not make sense when the alternative is under $35,000 once.
Run the comparison the way you'd run it on a piece of plant. A business paying $1,500 a month for inventory software that doesn't fit is spending $18,000 a year to keep the compromise alive. Over five years that's $90,000, and at the end of it they own nothing, the fit hasn't improved, and the per-user price has gone up twice.
When custom costs less than a few years of subscription fees, buying a compromise stops making sense. That's the whole argument. Everything else on this site is detail.
What hasn't changed
An explanation that only lists advantages isn't an explanation, it's a pitch. So here is the other side, and it's the part we'd want to read if we were you.
Understanding your business still takes real conversations with the people doing the work. There is no tooling that shortcuts it. If we skip it, we build something confident and wrong, quickly, which is worse than building it slowly.
Getting your data out of a system that doesn't want to give it up is still difficult. Some vendors export cleanly. Others export something technically complete and practically useless. Cleaning fifteen years of accumulated product records, duplicate customers and half-retired SKUs is unglamorous, and it takes as long as it takes.
Getting a team to change how they've worked for fifteen years is still the hardest part of any implementation. Software doesn't fix that. Building the system around how they already work, rather than how a vendor decided they should, makes it dramatically easier, but it doesn't make it free.
That's why we do discovery properly, why we lean hardest on the industries we already know, and why the timeline depends as much on you as on us.
How to check any of this
You should not take a page like this on trust, and you don't have to. The claim is falsifiable, which is unusual in this category.
We'll build a working demo of your business, in a real system, with your products and your terminology, in 48 hours. Free, before you've spent anything and before there's a proposal in front of you. You click through it.
If custom software still took a year, we could not do that. The demo is the argument, and it's the reason we'd rather show you than keep writing.
The judgement about what to build takes exactly as long as it always did. The typing doesn't.