The short answer: there is no rule that building a management system is always better or that SaaS is always cheaper. Build or deeply customize only when the process is a source of differentiation, no product fits, and the company can sustain engineering and operations over time. When the process is standard, speed matters, and there is no product team, SaaS or low-code is the better first move. Decide on five-year TCO, integration, data boundaries, and exit cost, not on the first-year quote.
This article is for companies choosing ERP, CRM, OA, MES, or a department-level management system. Research cut-off: 25 October 2026. YiDa prices are official public list prices; IDC figures are global market forecasts; the TCO ranges are editorial estimates, not quotes. Customer cases and claims such as "70% choose this route" that could not be verified in the earlier version have been removed.
Buy standard processes. Build only what differentiates you. A five-year budget has to include people, integration, data, and exit, not just software fees.
Build, SaaS, low-code, custom development: what is the difference?
The difference is not only price. It is delivery speed, process fit, data boundaries, and who owns long-term maintenance. The four routes suit different stages. They cannot be ranked as simply better or worse.
| Route | Best for | Main cost | Main risk |
|---|---|---|---|
| Standard SaaS | Finance, HR, collaboration, standard CRM, and other common processes | Subscription, implementation, interfaces, expansion, internal administrator | Price rises by user or data volume, limited customization, data export |
| Low-code or no-code | Forms, approvals, department-level apps, lightweight data management | Platform annual fee, implementation, integration, data-volume expansion | Limits on complex processes and high concurrency, platform lock-in, resource ceilings |
| Custom development | Industry-specific processes that need deep integration with existing systems | Requirements, development, testing, integration, launch, maintenance | Scope creep, vendor dependency, documentation and source-code delivery |
| Full in-house build | Core differentiated products, very large scale, no product fits | Product, R&D, testing, operations, security, cloud resources | Long cycles, hiring difficulty, key-person risk, continuous iteration |
The route comparison is an editorial summary. Routes can be combined: SaaS for standard finance, custom work for core scheduling, and low-code for peripheral approvals.
Which systems should you not build in-house?
General finance, HR, collaboration, email, and standard approvals should not be built in-house. They vary little between companies, carry high compliance requirements, and need long maintenance cycles. Buying a mature product is usually faster and cheaper.
The value of building comes from a process no one else has, not from rewriting standard features. Financial accounting, tax, social insurance, email, instant messaging, and standard OA approvals already have mature products and compliance work behind them. Building in-house means writing features and also tracking policy changes, patches, audits, and security for years.
| Criterion | Leans buy | Leans build or customize |
|---|---|---|
| Process differentiation | Industry-standard process, differences mostly configuration | Core process is a competitive advantage and no product on the market matches it |
| Time to launch | Results needed in weeks to months | Can accept a longer build and commit to iteration |
| Internal capability | No product, engineering, or operations team | Stable product, development, testing, and operations owners |
| Data boundary | Standard data, a compliant SaaS is acceptable | Core data must stay on-premises or in a private deployment, and existing products cannot meet that |
| Integration complexity | Standard interfaces, a few system connections | Deep ERP, MES, or OT integration, with interfaces that need custom work |
| Long-term cost | Subscription and expansion are predictable; low team burden | Can carry three to five years of R&D, operations, and security investment |
The decision table is an editorial recommendation. "Leans build" does not mean full in-house development. It can also mean custom development, a low-code extension, or a hybrid.
How do you calculate five-year TCO so you do not just save in year one?
Five-year TCO = first-year build or purchase + four years of annual operations + change and integration + migration and exit. For a build, include people and operations. For SaaS, include expansion and interfaces.
Five-year TCO = first-year build or purchase + 4 × annual running cost + change and integration + migration and exit.
Annual running cost = subscription or license + cloud resources + operations staff + security and backup + internal administrator.
Do not omit: data migration, interface rework, training, process change, and end-of-service or migration assistance.
The comparison below uses the same "department-level management system" assumption. The figures are editorial estimates, not quotes.
| Route | First year | Annual running | Five-year TCO | Key assumptions |
|---|---|---|---|---|
| SaaS | RMB 80,000 | RMB 50,000 | ~RMB 280,000 | Standard product, light integration, one internal administrator |
| Low-code | RMB 150,000 | RMB 80,000 | ~RMB 470,000 | Platform annual fee, implementation, 2–3 interfaces, business-built apps |
| Custom development | RMB 500,000 | RMB 150,000 | ~RMB 1.1 million | Year-one requirements, development, testing, and launch; annual maintenance at 15–20% |
| Full in-house build | RMB 900,000 | RMB 450,000 | ~RMB 2.7 million | 2–3 person team, cloud and security, ongoing product and operations investment |
The model excludes hardware, a full ERP rebuild, and business benefits. SaaS and low-code annual fees vary with users, data volume, interfaces, and edition. In-house labor is estimated at fully loaded cost; actual figures vary widely by city and team.
The table does not prove that building is always expensive. If the system is a core differentiated product, generates external revenue, or significantly reduces long-term cost, building can be justified. But it requires treating product, engineering, testing, operations, security, and knowledge transfer as long-term investments, not a one-time development fee.
Can low-code replace in-house builds or ERP?
Low-code can replace some forms, approvals, and department-level apps. It cannot automatically replace ERP finance, supply chain, production planning, or complex permissions. It is better suited to peripheral systems and transitions.
Using DingTalk YiDa's public prices as an example: Professional starts at RMB 5,988 per year, Dedicated from RMB 60,000 per year. At the 200-user tier, Professional is about RMB 21,988 per year and Dedicated about RMB 90,000 per year. Prices vary with headcount, data volume, attachment capacity, and edition. Check the official site and your actual usage before purchasing.
| Low-code fits | Low-code does not fit |
|---|---|
| Approvals, forms, registers, inspections, department-level data collection | Complex financial accounting, cost roll-up, master production scheduling |
| Light integration and reports within standard interfaces | High-concurrency core transactions, complex OT or equipment control |
| Business users building and iterating quickly | Strong compliance, complex permissions, cross-organization data isolation |
| Replacing Excel and offline processes, then expanding | Assuming a platform purchase replaces all ERP and MES capability |
YiDa prices come from the official pricing page (accessed 25 October 2026). Editions, resource quotas, and private-deployment options follow the official site or a sales quote. This article is not procurement advice.
What should the contract and data boundary cover?
Write down data export, interface permissions, account and attachment migration, source code and deployment documentation, annual fee caps, expansion unit prices, and end-of-service and exit support. These terms determine your negotiating position five years from now.
- Scope and acceptance: modules, user count, interfaces, data migration, training, and go-live criteria in the contract. A feature list is not acceptance.
- Data ownership: business data belongs to the company. Specify the export format, frequency, attachments, and logs.
- Interfaces and integration: interface list, call limits, field mapping, retry behavior, and who handles integration testing. Avoid a separate interface activation fee after go-live.
- Pricing mechanism: subscription or license, expansion, interfaces, storage, training, changes, and annual fee caps. Model the five-year budget at the cap.
- Source code and documentation: whether custom development includes source code, deployment documentation, and database schema. At minimum, the system must be migratable and maintainable.
- Exit: advance notice of service termination, data-export assistance, migration window, and costs. Do not discover that the data cannot leave only when you change vendors.
Who is LingMOE for, and who is it not for?
It suits manufacturers that want ERP, MES, OA, and quality traceability integrated, deployed privately, or replaced in phases. Standard office collaboration, simple forms, or a general CRM do not need a manufacturing integration platform.
LingMOE, our AI manufacturing platform, covers ERP, MES, and OA in one system. It fits single- and multi-plant manufacturers whose production planning, reporting, quality, and approvals are fragmented, and who need data boundaries and private deployment. It is designed for core manufacturing processes plus system integration. If the company only needs approvals, forms, or standard finance, SaaS or low-code is usually faster.
For a budget breakdown of ERP, MES, and OA integration, and for replacing a legacy ERP, see the related guides in our Insights archive. If you want a build-or-buy assessment, talk to us. We do not promise a fixed timeline or a minimum price.
FAQ
Is building a management system always cheaper than SaaS?
No. Software licensing may be zero, but product management, development, testing, operations, security, and continuous iteration are long-term costs. SaaS is cheap in year one, but can rise quickly with users, data, and interfaces. Compare five-year subscription, implementation, integration, internal labour, and exit cost in one table.
Can low-code replace ERP?
Low-code is better for forms, workflows, department-level apps, and lightweight data management. It cannot automatically replace ERP finance, supply chain, production planning, or complex permissions. Core transactions, cost accounting, and MES integration in manufacturing still need professional systems. Low-code can supplement or bridge them.
How should five-year TCO be calculated?
Five-year TCO = first-year build or purchase + four years of annual operations + change and integration + migration and exit. For a build, include product, development, testing, operations, cloud resources, security, and knowledge transfer. For SaaS or low-code, include subscription, implementation, interfaces, data-volume expansion, internal administration, and exit. Comparing first-year quotes alone is misleading.
Does sensitive data mean we must build in-house?
No. Private deployment, a dedicated cloud, a hybrid deployment, and contractual data-boundary terms may satisfy compliance better than building. Building is more reasonable only when existing products cannot support the core process and the company has long-term engineering and operations capability.
How should a company of about 50 people choose?
Usually start with SaaS or low-code for one or two high-frequency processes, then assess whether customization is needed. Do not build a complete platform from day one. Consider custom development or a build only when the process is a core differentiator, existing products cannot support it, and the company can carry long-term maintenance.
How does the contract prevent vendor lock-in?
Specify the data-export format, interface permissions, account and attachment migration, source code or deployment documentation, annual fee caps, expansion unit prices, end-of-service and exit support, and migration assistance. Do not look only at the feature list. Whether you can take the data with you and keep operating determines your bargaining power five years from now.
Buy standard processes; build only what differentiates you. Low-code fits peripheral systems and transitions, not the core of ERP. Put five-year TCO, integration, data boundaries, and exit cost on one page. Otherwise a low first-year price, or the promise that building gives more control, will make the decision for you.
Research method and limitations
This article is public price research plus editorial modeling. It is not a project audit, a vendor quote, or procurement advice. YiDa prices are from the official public page (accessed 25 October 2026) and change with headcount, edition, data volume, and promotions. The IDC figure is a global forecast for the low-code, no-code, and intelligent developer technology market, not data on Chinese manufacturing or a single company. The RMB 80,000/50,000, 150,000/80,000, 500,000/150,000, and 900,000/450,000 first-year and annual assumptions illustrate structure; they are not quotes for any company. The final build-or-buy decision should rest on real requirements, a proof of concept, the contract, and internal capability.
Data sources and scope
Research cut-off: 25 October 2026. All sources are public pages. Prices, dates, and scopes follow the originals.
- DingTalk YiDa official pricing page (accessed 25 October 2026): aliwork.com/pricing. Professional from RMB 5,988 per year, Dedicated from RMB 60,000 per year; at the 200-user tier, Professional about RMB 21,988 per year and Dedicated about RMB 90,000 per year. Follow the official site or a sales quote.
- IDC, "Worldwide Intelligent Developer Technologies and Low-Code and No-Code Forecast, 2025–2029" (December 2025): idc.com. The public summary shows global market revenue rising from USD 32.3 billion in 2024 to USD 225.4 billion in 2029. This is a global forecast, not purchase data for a single company.
- Alibaba Cloud price overview (accessed 25 October 2026): aliyun.com. Cloud servers, databases, storage, backup, and security for a self-hosted or private deployment are priced separately according to the official site and actual configuration. This article does not use specific cloud prices in its model.