Every manufacturer of a certain size has the same conversation eventually. The ERP is old. It has been customised past recognition. The people who understood the customisations have left. Someone proposes a replacement, and within a month there are three vendors in the building, each of whom has independently concluded that the answer is their platform.
Sometimes it is. More often the platform is adequate and the problem is somewhere else, and a replacement spends seven figures and eighteen months arriving at the same frustration with a newer logo on it.
Here is how to tell the difference before you sign anything.
Separate the four failure modes
When people say "the ERP is the problem," they usually mean one of four things, and only one of them is solved by a new ERP.
The platform is genuinely at end of life. The vendor has announced a sunset date, security patches have stopped, or the version you are on cannot be upgraded without a reimplementation anyway. This is a real forcing function, and the date on the vendor letter is the date you plan against.
The configuration was never finished. Phase one went live years ago. Phase two, the part that would have automated the thing everyone still does by hand, never happened. Nobody decided to stop; attention moved. The business has been operating in the gap ever since and has come to believe the gap is the software.
The process is the problem. The system enforces a workflow that made sense when it was configured and does not match how the plant runs now. That is a process and configuration question, and it survives a migration intact if nobody addresses it.
Nobody was trained. People use eleven per cent of the platform and route around the rest with spreadsheets. Replacing it moves the same habits to a new interface.
Only the first genuinely requires a replacement. The other three are cheaper to fix, and if you do not fix them first, they come with you.
The questions the vendors will not ask
An honest evaluation of your current platform starts with questions nobody selling a replacement has any incentive to raise.
What percentage of the modules you already own are actually in use? What did the original implementation scope include that never got delivered? Which customisations are load-bearing and which were workarounds for a configuration nobody revisited? What would it cost to finish the implementation you already paid for, and how does that compare to starting again?
That last number is the one that changes decisions. We have seen companies discover that eighteen months of a replacement programme could be avoided by six weeks of configuration and two days of training, and we have seen companies confirm that yes, the platform is finished and it is time. Both are good outcomes. What is not a good outcome is finding out afterwards.
If it is time, negotiate before you fall in love
If the replacement is genuinely necessary, the decision you make in the next quarter determines your cost base for the next decade. Two things matter more than the feature comparison.
First, the total cost of ownership over seven years, not the licence quote. Implementation, integration, data migration, the internal time nobody counts, and the annual uplift buried in year three of the agreement.
Second, your exit. Every ERP contract has switching costs designed into it, and the time to negotiate data portability and termination assistance is before you sign, not when you next want to move.