The Most Expensive Line Item on Your Budget Is a Decision You Made Ten Years Ago
We all know the sunk cost fallacy. We can all recite it at dinner parties. And yet, walk into any boardroom and you'll find it alive and well, wearing a lanyard, sitting in the seat labeled "our ERP investment."
"We've spent eight years and eleven million on this implementation. We can't walk away now."
Yes, you can. In fact, the eleven million is precisely the reason you should be asking whether to walk away — not the reason to stay. The money is gone. The licenses are gone. The consultant hours are gone. The only question that matters is the one nobody wants to ask out loud: if we were starting today, with today's tools, would we buy this again?
For the first time in decades, the honest answer might be no.
What changed
Until very recently, "build vs. buy" wasn't a real debate for most companies. Building meant hiring a development team, waiting eighteen months, and praying. Buying meant Salesforce, SAP, Oracle, or whichever generic platform your industry defaulted to — plus an army of external consultants to bend it into something resembling your actual business.
That equation has been quietly rewritten. LLMs like Claude, AI coding tools, and modern automation platforms have collapsed the cost and time of building custom software by an order of magnitude. A workflow that once required a six-figure consulting engagement can now be prototyped by a small internal team in weeks. Internal tools that would have been "nice to have someday" are now a sprint away.
The moat around enterprise SaaS was never really the software. It was the cost of the alternative. That cost is evaporating.
Let's be honest about the limits
Before anyone accuses me of selling a fantasy: no, you cannot vibe-code your way to a replacement for SAP's financial core by Friday. Nothing quite like that has actually been built yet. No AI tool today can replicate twenty years of accumulated domain logic, compliance certifications, and battle-tested edge cases that live inside mature, specialized platforms. Payroll, regulated financial reporting, mission-critical supply chain — these are not weekend projects, and anyone telling you otherwise is selling something.
But here's the uncomfortable follow-up question: how much of what you're paying for actually is that deep, specialized core?
The 80% problem
Look honestly at your SaaS stack and ask how much of it your teams actually use. Most enterprises pay for sprawling, one-size-fits-all platforms and use a fraction of the functionality — while simultaneously fighting the platform daily because the 20% they genuinely need doesn't quite fit how they work.
So what happens? You hire consultants. Expensive, external, generic consultants whose entire job is to configure, customize, and contort a legacy platform into an approximation of your real processes. You're not paying for innovation. You're paying rent on rigidity — and then paying again for the privilege of softening it.
Read that back: we pay for software that doesn't fit our routines, then pay specialists to make the software slightly less unfit. And we call the alternative — building tools around our actual workflows — "risky."
The real question for decision makers
I'm not arguing every company should rip out its ERP tomorrow. I'm arguing that the default has flipped, and most leadership teams haven't noticed.
The old default: buy the platform, adapt the company to it.
The new default worth considering: define what your company actually does, and ask whether a lean, custom-built layer — assembled with AI-assisted development and automation tools, by people who know your business — serves it better than a generic giant plus a consulting bill.
Three questions to bring to your next budget review:
- Can our legacy platforms adapt at the speed our business now changes? If every process change requires a change request, a consultant, and a quarter of lead time, the platform isn't a foundation. It's an anchor.
- What are we actually paying for? Separate the truly specialized core (keep it, it's earned its place) from the generic workflow layer (CRM views, approvals, dashboards, reporting glue) that AI-era tooling can now replicate — tailored to you, owned by you.
- Are we deciding based on the future, or defending the past? If the main argument for a system is how much you've already spent on it, that's not a strategy. That's the sunk cost fallacy with an enterprise license.
The companies that win the next decade won't be the ones with the biggest SaaS budgets. They'll be the ones that stopped asking "how do we fit our business into the software?" and started asking "how fast can we build software that fits our business?"
The tools are on the table. The only thing still locked in is the mindset.
What's your take — is your organization still defending decade-old platform decisions, or already building around them?