Choosing a manufacturing ERP is a decision a business lives with for a decade. Most selection processes go wrong the same way: they start with software demos and end with whoever demoed best. This is a more reliable process, written for manufacturers who want to choose well rather than choose fast.
Start with your operation, not the software
Before looking at a single product, write down how your business actually manufactures. Are you discrete assembly, or process and batch? Make-to-stock, make-to-order, or engineer-to-order? How deep are your bills of materials, two levels or eight? Do you subcontract operations? Do you need lot or serial traceability, and to what standard? How many sites, and do they feed each other?
This is the most important step and the one most often skipped. A manufacturing ERP is not good or bad in the abstract; it is good or bad at the way you make things. A system built for repetitive discrete assembly can be genuinely weak at process manufacturing with recipes and co-products, and the reverse is just as true. If you cannot describe your operation precisely, you cannot judge a system against it, and you will end up judging the demo instead, which is exactly the trap this process exists to avoid.
Separate the must-haves from the nice-to-haves
From that description, build two lists. Must-haves are the things that, if the system cannot do them, end the conversation: the BOM structure you actually use, your manufacturing mode, the traceability your industry demands, the integrations you genuinely cannot live without. Nice-to-haves are everything else.
The discipline is keeping the must-have list short and honest. A long must-have list usually means wants and needs have blurred, and it quietly pushes you toward overbuying a system far heavier and more expensive than the operation needs. A useful test for each item: if the system could not do this, would the project genuinely fail? If the answer is no, it belongs on the nice-to-have list.
Build a real shortlist
With the must-haves defined, narrow the field to three or four systems that plausibly clear them. Match the tier to the operation: a twenty-person job shop and a two-thousand-person multi-plant manufacturer are not shopping in the same market. Resist the urge to shortlist on brand recognition alone. A well-known enterprise system on a small manufacturer's shortlist is usually there because the name is familiar, not because it fits, and it will distort the rest of the process by making everything else look small.
Test demos against your reality
When you see demos, do not watch the vendor's script. A vendor demo is built to impress, on clean data, along the happy path. Make them show your situation: your kind of BOM, your exception cases, what month-end looks like, what happens when a customer changes an order mid-production, how a component shortage surfaces. Hand them a sample of your own messy data if you can.
A system that handles your awkward cases in a demo will handle your ordinary days in production. A demo that only shows the smooth path is telling you nothing you can use. Ask, specifically, to see the things you were told would be hard, and watch how the vendor responds to being pushed off-script.
Weigh the total cost honestly
The licence price is the small, visible number. The real cost is implementation, data migration, customisation, training, and the years of operating the system afterward. A cheap licence on a system that is expensive to implement and hard to maintain is not cheap; it just moves the cost to where the proposal does not show it.
Ask every shortlisted vendor for an honest estimate of the full first-year cost and the recurring annual cost after that. Treat vague answers as a finding in themselves: a partner who will not estimate the real cost early is unlikely to be precise about it later, when the work is underway and the leverage has shifted.
Weigh the partner as hard as the software
The same ERP, implemented by a team that understands manufacturing and implemented by one that does not, produces a success and a failure from identical software. The implementation partner decides the outcome at least as much as the system does.
Ask who specifically will run your project, not who is in the sales meeting. Ask what comparable manufacturers they have delivered for, and ask them to describe one in detail. Ask how they handle a project that goes wrong, and what happens after go-live, whether stabilisation is part of the engagement or billed later as a separate phase. A partner who answers those concretely is worth more than a slightly longer feature list.
Make the decision on evidence
By the end you should be choosing between two or three systems that all clear your must-haves, on the basis of fit, total cost, and the partner. That is a decision made on evidence, and it is one you can still defend in three years. The selection that started with demos and ended with the best presenter is the one the business quietly regrets. For how we run this process, see our manufacturing ERP work.