The wrong way to choose an ERP is to call three vendors, watch their demos, and pick the one whose salesperson you liked best. The right way is to answer six questions before you talk to a single vendor — and then use those answers to filter the field from twenty candidates to two.
Question 1: What is the specific operational pain?
Name it precisely. Not "we need a system" — that is not a pain. "Our sales team takes four hours per quote because pricing is in three Excel files and two people need to approve before we can send" is a pain. "We cannot see production status without calling the supervisor on the floor, which happens twelve times a day" is a pain. "We have missed three delivery dates this quarter because no one knew the job was blocked on materials" is a pain.
Once you have the specific pain, ask: is this a finance problem or a workflow problem? Finance problems (consolidation, multi-currency, complex revenue recognition) point toward packaged ERP. Workflow problems (quoting, production visibility, supplier coordination, customer portals) often point toward a custom build or a very narrow packaged deployment.
Question 2: Is your workflow standard or non-standard?
Standard workflow means: your quoting logic follows quantity tiers, your production follows a linear BOM, your supplier relationships are straightforward purchase orders. Packaged ERP systems model this well.
Non-standard workflow means: your pricing depends on substrate type, customer category, order volume, and delivery timeline simultaneously; your production involves substrate-specific setup times and yield rates that vary by batch; your supplier relationships involve consignment, partial deliveries, and quality holds. Packaged ERP systems model this poorly — or require expensive customisation to model it at all.
The test: find three Singapore manufacturers in your sub-vertical using the packaged ERP you are evaluating. Ask them specifically if the quoting and production modules handle their non-standard cases without customisation. If they cannot name three, or if they all say "we customised it heavily," that is your answer.
Question 3: What is the five-year total cost?
Every ERP has a five-year TCO, not a one-time cost. Calculate: implementation cost (year one) + annual licence or SaaS subscription × five years + annual support retainer × five years + customisation and upgrade costs years two to five. For a 20-user Singapore manufacturer, this typically lands at:
Odoo (via SG partner): SGD 120,000–280,000 over five years. ERPNext: SGD 130,000–350,000. SAP Business One: SGD 250,000–500,000. Dynamics 365: SGD 200,000–500,000. NetSuite: SGD 350,000–700,000. Custom build (Start Canyon): SGD 10,000–30,000 build + SGD 0 licence + support as needed. The custom build number looks dramatically lower because it is — there is no ongoing licence and the system is built precisely for your workflow, minimising support.
Question 4: What does your team actually need to adopt?
The best ERP that nobody uses is worse than the worst ERP that everybody uses. Assess your team realistically: are they comfortable with complex software, or do they need something that works exactly like their current process? Production floor staff and operations teams typically have low tolerance for interfaces that require training — they will revert to Excel within six months if the system is not intuitive for their specific tasks.
This question often resolves the standard-vs-custom debate. A custom build can be designed exactly around how your team currently works — same vocabulary, same sequence, same outputs — which dramatically reduces training time and reversion risk.
Question 5: What are the Singapore-specific requirements?
Every Singapore manufacturer must verify: InvoiceNow/PEPPOL compliance (is it supported natively, via a partner module, or not at all?), GST handling at 9% for standard, zero-rated exports, and import GST, CPF payroll if HR is in scope, and MOM foreign worker levy tracking if applicable. These requirements disqualify some vendors and add significant implementation cost for others.
Question 6: What happens when something breaks?
ERP support in Singapore is partner-dependent for most platforms. Ask: how many SG-based consultants does this partner have? What is the SLA for production-down issues? What happens if this partner closes or pivots? For packaged ERP, the platform continues but your local support degrades. For a custom build, the source code is yours — any competent developer can support it, and Start Canyon provides ongoing retainer support.
The decision matrix
Custom build wins when: workflow is non-standard, team needs high adoption simplicity, five-year TCO matters, and the problem is operational not financial. Packaged ERP wins when: workflow is standard, multiple departments need broad coverage simultaneously, the business is scaling toward multi-entity complexity, or a PSG-eligible solution is the budget constraint.
If you have answered these six questions and still cannot determine which path fits, Start Canyon's paid discovery is designed for exactly this situation. The discovery deliverable gives you a written recommendation — custom build vs packaged ERP, scoped to your specific workflow — that you can take to your board.