Cloud ERP Not Enough? Keep It, Integrate or Go Custom

When a cloud ERP stops being enough, it usually breaks in three places: complex BOMs and production scheduling, integration with e-commerce, POS, warehouse systems or LINE, and reports that cannot be built or documents keyed in twice. Most of the time you do not need to replace the ERP. Keep the package and fill the missing step; go custom only when the core process itself is your competitive edge.

Subscription cloud ERPs go live fast and have clear pricing. Based on prices published on vendor websites in November 2026, there are plans at about NT$1,500 per month for a whole company with unlimited users, and plans billed per user at a few hundred NT dollars per user per month. That price band suits companies whose processes resemble their industry and that are systemizing for the first time. The trouble usually starts once the company grows and its processes drift away from the software's defaults.

When does a cloud ERP start falling short?

The warning signs are staff opening spreadsheets outside the ERP, the same data typed into two systems, or management reports assembled by hand. At that point, find where the process is stuck before rushing to replace anything.

An ERP's value is that order intake, purchasing, inventory, production and finance share one set of linked data (Gartner's definition of ERP also centers on integrating core business processes). Once any step runs outside the ERP, that link breaks, and month-end reconciliation, inventory figures and costing start drifting apart.

In practice, three situations come up most often:

• Production: multi-level product structures, many substitute parts, or capacity-based scheduling that the packaged BOM and scheduling features cannot express.
• Integration: orders arrive from e-commerce, store POS or LINE, shipping happens in a warehouse system, and data moves by manual import and export.
• Management: the reports managers want cannot be built, or one document is filled in separately in the ERP, a spreadsheet and an approval system.

The next three sections cover two paths for each situation.

Complex BOMs and scheduling: add a module or go custom?

If the BOM simply has many levels or substitute parts, or you only lack work-order reporting and a scheduling board, keep the ERP and add a production module beside it. Custom production management only makes sense when the process itself is your edge, such as adjusting formulas or routings per order.

A BOM (bill of materials) lists which parts make up a product and in what quantities; MRP then uses the BOM and orders to work out what to buy and when (the ASCM Supply Chain Dictionary defines both). Most cloud ERPs support standard multi-level BOMs, but they tend to struggle when parts change by customer specification, outsourced processing is priced by stage, or scheduling must account for the real capacity of machines and people.

Path one: keep the package and add a module. This fits when the ERP's part numbers, inventory and costing still work well and only the shop-floor step is missing. Build work-order reporting, a scheduling board or a quality inspection module alongside it, and write completed quantities and material usage back to the ERP. A single-process module is typically NT$500,000 to 1 million; a small add-on such as a reporting form or scheduling board starts at NT$150,000.

Path two: go custom. This fits when formulas, routings or pricing rules are your company's expertise, the package needs many workarounds to handle them, and every ERP upgrade means reconfiguring. Custom work usually turns production management into its own system while finance stays in the ERP; a mid-sized system spanning several departments is typically NT$1 million to 2.5 million, depending on scope.

The ERP will not open up for e-commerce, POS, warehouse or LINE integration. What now?

First check whether the ERP has an API or scheduled import and export. With an API, integrate directly, at about NT$50,000 to 200,000 per integration point. With file exports only, use scheduled file exchange. Consider a custom order hub only when even exports are limited and integration points keep multiplying.

Common integration needs:

• E-commerce: online orders become ERP sales orders automatically, and stock levels flow back to the store to prevent overselling.
• POS: store sales and e-invoice data flow back to the ERP. Issuing, voiding and allowances must reconcile with the books; check the process against the Ministry of Finance's directions for electronic invoices.
• Warehouse (WMS): shipping instructions go from the ERP to the warehouse, and picking and shipping results are written back.
• LINE: order confirmations, shipping or payment reminders are sent to customers through a LINE Official Account. The LINE Messaging API provides interfaces for push and reply messages, but push volume is billed by the Official Account plan, so estimate monthly message counts first.

Path one: keep the package and integrate. This fits when the ERP itself works well and the data just needs to flow in and out automatically. When JoinX built the website and e-commerce platform for 比漾百貨, we kept the existing ERP, POS and Digiwin accounting system and used integration to connect online and offline operations. Cost is counted per integration point, about NT$50,000 to 200,000 each, with the ERP vendor's API module billed separately. See ERP integration services for more.

Path two: go custom. This fits when you need to connect more than three or four systems that all sync with each other, or the ERP's API limits force a workaround for every integration. A custom "order hub" can then act as the center: e-commerce, POS, warehouse and LINE connect to it, and the ERP handles only accounting. A mid-sized order system is typically NT$1 million to 2.5 million.

Reports cannot be built and documents are keyed in twice. How do you fix it?

Missing reports usually mean the data is in the ERP but in the wrong shape, so exporting it to a reporting tool or adding a report module solves it. Double data entry means the process breaks outside the ERP, so connect that form or approval step back to the ERP, or rebuild the whole step as custom.

The cost of retyping documents is often underestimated: a dozen documents a day, each entered twice, adds up to real staff time over a year, plus errors in the books. Count which documents are retyped and by whom, and you will know which step to fix.

Path one: keep the package and add a module. This fits when the ERP's data is complete and only the presentation or the front-end form is missing. For example, turn quotations and purchase requests into online forms that become ERP documents once approved, or pull ERP data into a reporting tool on a schedule to produce the margin and inventory-turnover analysis managers want. Small add-ons like these start at NT$150,000.

Path two: go custom. This fits when the data the report needs is not in the ERP at all, such as shop-floor hours, field dispatch or member behavior, and those processes change every day. Rather than piling more forms around the ERP, build that operational process into its own system and reconcile it with the ERP. A simple custom ERP starts at about NT$450,000.

Keep, integrate or go custom: how do cost and timeline compare?

Based on 2026 custom projects in Taiwan, integrating one system costs about NT$50,000 to 200,000, small add-ons start at NT$150,000, a single-process module is typically NT$500,000 to 1 million, and a simple custom ERP starts at about NT$450,000. Actual cost depends on how many systems you connect and how open the ERP is.

Cloud ERP subscriptions and the ERP vendor's API modules and licenses are billed separately. Maintenance after a custom system goes live is covered by a separate contract.

How do you decide which path to take?

Work through the checks below in order; most companies can decide by step three. If only the data flow is stuck, integrate. If a whole process is missing, add a module. Go custom only when even the core process needs workarounds.

1. List the data retyped every day: which documents are entered in two or more places, by whom, and how many per day.
2. Check how open the ERP is: whether it has an API, whether scheduled import and export is possible, and how the vendor charges for API modules.
3. Decide whether the gap is data flow or process: if the data exists but does not sync, integrate; if a whole process sits outside the ERP, add a module.
4. Check whether configuration can handle the core process: consider custom only when what must change is how the company competes, and every upgrade forces reconfiguration.
5. Go live in phases: fix the most painful step first, extend to the next once it is stable, and never replace everything at once.

For a fuller comparison of packaged, cloud and custom ERP, see How to Choose an ERP: Packaged, Cloud ERP or Custom Development?.

Companies that do go custom rarely get everything in one release: 欣晨快遞 grew its logistics management system and driver APP in phases from 2025 to a fifth version, and 驊曖體育 has released its yoga studio management system up to a ninth version since 2020. Both launched core processes first, then extended them as operations required.

Frequently Asked Questions