Systems & Data

The "Frankenstack": why your data, not your ambition, is capping your growth

The ceiling on most manufacturers isn't money or market. It's the tangle of systems nobody planned.

Most owners assume their growth limit is something external: capital, headcount, demand. Often it's something internal and a lot quieter. It's the pile of disconnected systems the business grew into over the years. Researchers at MIT Technology Review Insights gave it a name that fits perfectly: the "Frankenstack." A stack stitched together from whatever solved yesterday's problem, now a monster you spend real energy just keeping alive.

Here's the number that should stop you. By one estimate, only about a third of companies say they have a unified data core. The rest run on splintered sources: an ERP over here, a separate quality system over there, spreadsheets filling the gaps, and a few critical things that live only in one person's head. Each piece works. The business still runs. But the seams between them are exactly where growth leaks out.

Why this matters more now than it did five years ago

The whole promise of modern automation, and of AI in particular, is software that can move across your workflows and act on current information. A Frankenstack blocks precisely that. If your systems don't talk to each other, the AI can't either. It can only see the one corner it's plugged into, which means it can only ever be a clever tool in a single department, never the thing that lifts the whole operation. As the researchers put it, these stacks "lack the continuity required for agentic AI to deliver outcomes across the business."

The cost is real, it's just hidden

A Frankenstack rarely shows up as a line item, which is why it survives for years. It shows up as friction:

• Swivel-chair work. People re-keying the same data from one system into another, all day, because the systems won't share it.
• Reports that take weeks. Because a person has to assemble the numbers by hand from four places before anyone can decide.
• Decisions on stale data. By the time the picture is complete, it's already out of date.
• Errors that compound. One wrong entry early on quietly poisons everything downstream.

None of that is dramatic. All of it caps how fast you can grow.

The fix isn't a bonfire

You don't beat a Frankenstack by ripping everything out and starting over. That's how you turn a slow problem into an expensive one. You beat it by knowing exactly what you have first, then connecting and rationalizing on purpose. The manufacturers who do this well start with an honest map: every system, how data actually moves between them (or doesn't), and where the real single source of truth lives for each part of the business. Once that map exists, the plan tends to make itself obvious. Connect what should talk. Retire what shouldn't exist. Standardize the data so the next system you add can actually use it.

There's a bonus most people miss. A cleanup like this usually becomes the catalyst for fixing the deeper problems that were holding the business back anyway. The report notes the same thing: sorting out the stack forces the kind of clarity that gives a company a real roadmap for the first time.

Where this leaves you

That map is where we start. Before anyone talks about buying new tools, we get you a clear, current picture of your stack and your data: the swivel-chairs, the silos, the spreadsheets quietly doing a system's job. Then we connect and standardize, so your information can finally move, and so anything you add next is built on solid ground instead of bolted onto the monster.

Your ambition probably isn't the problem. Your plumbing is. And the ceiling lifts the moment your data can move.

Sources: MIT Technology Review Insights, The Connected Customer (2025), drawing on Metrigy research.