The short answer: for most single-plant manufacturers, first-year ERP, MES, and OA integration budgets fall into three bands: RMB 100,000–300,000 for a light, standardized deployment; RMB 500,000–1,500,000 for single-plant integration; and RMB 3–8 million or more for multi-plant, group, or deeply customized work. It is not the price of one software package. It is five separate accounts: software, implementation, integration, data and training, and operations.
This is a cost analysis and procurement reference for owners, CIOs and IT leaders, and procurement managers at manufacturers with 50–1,000 employees. Research cut-off: 24 October 2026. Prices come from public vendor articles, industry quote articles, and IDC market data, plus our own editorial estimates. They are not a formal quote from Lingyi or any other vendor, and not a promise of return on investment.
Integration does not mean one vendor supplies everything. A first-year quote is not a five-year TCO. And implementation usually costs more than software. Count plants, interfaces, and master-data domains before choosing a tier.
How much does ERP, MES, and OA integration actually cost?
Across the three tiers — light single-plant, integrated single-plant, and multi-plant — first-year budgets commonly run RMB 100,000–300,000, RMB 500,000–1,500,000, and RMB 3–8 million or more. Reusing existing systems pushes you toward the lower end. Deep customization and private deployment push you toward the upper end.
| Tier | Typical company | First-year budget (editorial estimate) | Usually includes | Not included / may be added | Following years |
|---|---|---|---|---|---|
| Light single-plant | 50–200 people, one plant, relatively standard processes, basic finance or inventory system in place | RMB 100,000–300,000 | Light ERP/MES/OA subscription or standard edition, 1–3 interfaces, basic training, go-live coaching | Deep customization, private hardware, internal labor, production-line downtime | RMB 20,000–60,000 per year |
| Single-plant integration | 200–500 people, multiple workshops or process steps, ERP, MES, and OA siloed | RMB 500,000–1,500,000 | Core ERP, MES, and OA modules, 3–8 interfaces, master-data review, some customization, training and pilot | Multi-plant replication, full private hardware, internal labor, downtime losses | RMB 100,000–200,000 per year |
| Multi-plant / group / deep customization | 500+ people or multiple plants and entities, complex processes, high compliance or data-boundary requirements | RMB 3–8 million+ | Multi-entity, multi-plant, 10+ interfaces, private or hybrid deployment, deep customization, permission auditing, phased rollout | Private hardware (commonly RMB 500,000–2 million+ extra), internal team, plant downtime, long-term operations | RMB 500,000–1 million+ per year |
The three tiers are editorial estimates based on public quote articles and industry TCO ranges, not a universal price list. Actual prices vary with user count, plant count, modules, interfaces, customization, and deployment model. Buying only MES or only OA should cost less than the matching tier; reusing an existing ERP or OA also pushes you to the lower end. AI modules and agents are usually priced separately and are not included in these tiers.
Why can the same “integration” quote run from RMB 100,000 to RMB 5 million?
The difference is scope: users, plants, modules, interfaces, customization, deployment model, and data quality. An RMB 100,000 quote and an RMB 5 million quote can both be honest.
| Variable | Low end | High end | Effect on budget |
|---|---|---|---|
| Users / concurrency | 20–50 accounts, light concurrency | Hundreds of accounts, many roles, multiple devices | Higher license or subscription, training, and support costs |
| Plants / organizations | One plant, one entity, one set of books | Multiple plants, entities, books, and currencies | Higher permissions, master data, consolidation, and rollout costs |
| Module scope | Work orders, reporting, quality inspection, approvals | Full finance, supply chain, production, quality, equipment, and collaboration modules | More modules mean more implementation and interfaces |
| Interface count | 1–3 standard APIs | 10+ systems, including legacy systems or equipment with no API | Higher interface development, integration testing, and maintenance |
| Customization depth | Standard configuration plus a few reports | Unique processes, complex approvals, deep secondary development | Higher development, testing, and upgrade-compatibility costs |
| Deployment | Public-cloud SaaS | Private or hybrid, domestic IT stack, cybersecurity grading | Higher hardware, security, operations, and delivery time |
This table explains quote differences; it is not a pricing formula. The same variable carries different weight across vendors and industries.
Before comparing quotes, align the scope: how many plants, how many users, which modules, which systems, what must be customized, and where it will be deployed. Without a common scope, comparing total prices is meaningless.
Where does the money actually go? Five cost lines
Software, implementation, integration, data and training, and operations. In manufacturing, implementation and integration often cost more than software. Data work and internal labor are the items most often left out of the first-year budget.
| Public source | Scope | Key ratios / figures | What it means for the budget |
|---|---|---|---|
| Zhiyuan Internet (13 November 2025) | Growing manufacturers with RMB 50–200 million annual revenue and 100–300 employees | ERP TCO RMB 300,000–1,500,000; software 20–40%, implementation 40–60%, customization/integration 5–15%, hardware 5–10%, annual operations 3–5%; implementation often 1–3× software; perpetual-license maintenance often 15–22% of license | Implementation is the largest cost; comparing only license fees badly underestimates the total |
| Digiwin (20 March 2026) | ERP cost structure for a 100-user mid-sized manufacturer | Cloud ERP three-year TCO RMB 450,000–800,000; subscription 22%, implementation 35%, customization 18%, cloud resources 10%, integration/operations 15%; on-premises ERP three-year TCO RMB 1.1–2.6 million; hardware 30%, software 20%, implementation 25%, customization 12%, operations 13% | Cloud versus on-premises is not simply “which is cheaper”; it is a difference in cash flow and long-term ownership |
| Blacklake (2 June 2026) | MES software fees and first-year total investment for light, mid-sized, and large plants | Light: RMB 10,000–45,000 in year one; mid-sized: RMB 100,000–380,000; large: RMB 510,000–1.9 million; mid-sized MES implementation often 1–3× software; hidden costs include data migration, secondary development, training, and downtime | A low MES unit price does not mean a low total; implementation and data determine the real go-live cost |
| Tencent Cloud Developer Community / Blacklake (13 April 2026) | SaaS versus perpetual license for 50–150-person small and mid-sized manufacturers | SaaS about RMB 150,000–300,000 over three years; perpetual license about RMB 450,000–750,000 over three years including maintenance; hidden implementation fee for 30 days on site about RMB 50,000–150,000; annual maintenance commonly 20% of the perpetual price | Over three years SaaS is often lower; over the long run a perpetual license may amortize better, but maintenance and upgrades must be included |
All of these are public vendor or industry articles and may carry a marketing bias. They are not an independent audit or a universal price list, and the figures cannot be added together into a “standard quote.” We use them only as budget anchors and check the ranges with the editorial model below.
After reconciling those public figures, we use a five-line model for the first-year budget. It is not a quote sheet; it is a checklist against omissions.
| Line item | Typical first-year share | What it covers | Most common omission |
|---|---|---|---|
| Software / subscription | 20–40% | ERP, MES, and OA licenses or annual subscriptions, priced by module, user, or plant | Only base modules are quoted; quality inspection, scheduling, and reports cost extra |
| Implementation services | 25–40% | Discovery, process design, configuration, testing, go-live, project management | The quote excludes on-site work, travel, training, and go-live support |
| Integration / interfaces | 10–25% | ERP–MES–OA, WMS, PLM, equipment, finance, and other interfaces | Legacy systems have no API, and the number of interfaces is underestimated |
| Data / training | 5–15% | Master-data cleanup, migration, evaluation, key-user training | Duplicate BOM, material, and customer codes; training happens only once |
| Operations / annual fees | 10–20% in year one and every year after | Subscription renewals, maintenance, cloud resources, monitoring, support | Annual fees treated as one-off; no allowance for upgrades and changes |
Shares are editorial estimates and vary widely by project. First-year shares may add up to slightly more than 100% because implementation overlaps with integration, and data overlaps with training. The table excludes internal labor, production-line downtime, and hardware replacement.
Five-year TCO example: put the first-year quote back on a timeline
Assume no inflation, no hardware replacement, and no internal labor. Use first-year investment plus four years of annual operations. The value of this example is not a standard answer; it is a consistent basis for comparing vendors.
| Scenario | First year | Annual operations | Five-year TCO | Calculation and assumptions |
|---|---|---|---|---|
| Light single-plant | RMB 200,000 | RMB 40,000 | RMB 360,000 | 200,000 + 40,000 × 4; SaaS or light MES, one plant, few interfaces, excludes internal labor |
| Single-plant integration | RMB 800,000 | RMB 120,000 | RMB 1,280,000 | 800,000 + 120,000 × 4; ERP + MES + OA, 3–8 interfaces, includes data review, excludes private hardware |
| Multi-plant / group | RMB 3,000,000 | RMB 450,000 | RMB 4,800,000 | 3,000,000 + 450,000 × 4; multiple entities, 10+ interfaces, private or hybrid, excludes hardware and downtime |
The examples are editorial estimates, not quotes. A real budget should add hardware, internal labor, downtime, upgrades, and change requests. If a vendor's first-year quote is very low but annual fees are high, the five-year TCO can reverse.
An IDC article published in 2026 reported that China's AI-enabled ERP market reached USD 315.7 million in 2025, up 96.1% year on year. The figure counts only new revenue from AI products and modules in ERP scenarios; it excludes traditional ERP. AI modules are clearly becoming a new budget line, but do not treat 96.1% as the markup you should expect to pay. The three tiers above exclude AI modules. Intelligent scheduling, quality inspection, or a knowledge base should have a separate budget.
Is ERP, MES, and OA one software product or three systems connected?
Both count as integration. The key is not the number of vendors. It is consistent master data, maintainable interfaces, and unified permissions and audit. One vendor supplying everything is not necessarily cheaper, and three systems are not necessarily more expensive.
| Path | Best for | Advantages | Risks | Cost profile |
|---|---|---|---|---|
| Single integrated platform | Relatively standard processes, all three systems aging, a wish to reduce interfaces and vendor coordination | One master-data model, one permission system, one accountable party | Vendor lock-in; some modules are less deep than specialist products; high switching cost | Software and implementation may be higher; integration and coordination lower |
| Multi-system integration | A stable ERP or OA already in place, and MES needs specialist depth | Preserves existing investment; best-fit system in each domain | Many interfaces, complex responsibility boundaries, upgrades affect each other | Software may be lower; integration and operations higher |
| Self-built / low-code | Highly unique processes, a stable IT team, and the ability to maintain the system long term | Fits the business closely; flexible iteration | Key-person dependency, weak documentation and testing, high long-term maintenance cost | First year may be low; three-to-five-year total cost is often underestimated |
The choice of path is an editorial judgement, not an industry consensus. A company can move from multi-system integration to a platform as it matures.
For a 200–500-person single plant with a stable ERP, the usual priority is to keep the ERP and add MES, OA, and interfaces. If all three systems are old, master data is chaotic, and the vendor has ended support, a single integrated platform may cost less. For multiple plants, unifying the data model and governance rules matters more than insisting on one brand.
Which tier fits which company size?
Companies with 50–200 people should start with a light, standardized deployment. Companies with 200–500 people should aim for single-plant integration. Companies above 500 people or with multiple plants should move to a private or group edition. Do not buy a large platform first and work backwards.
| Company size | Recommended tier | What to do first | What not to do | First-year budget |
|---|---|---|---|---|
| 50–200 people, single plant | Light single-plant | Run work orders, production reporting, quality inspection, and one or two approval flows; keep interfaces to 1–3 | Replace the core ERP immediately; buy “full-suite” modules you will not use | RMB 100,000–300,000 |
| 200–500 people, single plant with multiple workshops | Single-plant integration | Assign a master-data owner, map 3–8 critical interfaces, and pilot on one line | Skip the pilot and roll out plant-wide; change organizational processes and systems at the same time | RMB 500,000–1,500,000 |
| 500+ people / multiple plants | Multi-plant / group | Define the data model, permission audit, and phased rollout plan; replicate by plant | Big-bang go-live; accept the project by “number of modules” instead of a closed business loop | RMB 3–8 million+ |
| Existing stable ERP | Add MES/OA plus integration | Keep the ERP, add shop-floor execution and collaboration, integrate read-only first and write back later | Replace a working ERP in the name of “integration” | Assess from the top of the light tier or the bottom of the single-plant tier |
Budgets are editorial estimates, not quotes. Take the lower end when systems can be reused, interfaces are standard, and processes are simple; take the upper end or higher for multiple entities, deep customization, and private deployment.
Who Lingyi's LingMOE fits: discrete manufacturers with one or more plants, an existing ERP but fragmented MES and OA, a need to connect work orders, production reporting, quality inspection, and approvals on one platform, and requirements around data boundaries or private deployment. Who it does not fit: companies with fewer than 50 people that only need invoicing, inventory, or simple approvals; companies without a data owner or IT administrator; and companies that expect every module live in two weeks. A light SaaS product is a better deal for them.
How long does implementation take, and when does it slip?
Four to eight weeks for a light deployment, three to six months for single-plant integration, and six to eighteen months for multi-plant or deep customization. Delays usually come from master data, interfaces, approval chains, and production cooperation, not slow coding.
| Phase | Light single-plant | Single-plant integration | Multi-plant / custom | Key dependency |
|---|---|---|---|---|
| Requirements / blueprint | 1–2 weeks | 3–6 weeks | 6–10 weeks | Business owner, process list, plant scope |
| Configuration / development | 2–4 weeks | 6–12 weeks | 3–6 months | Module scope, customization, vendor resources |
| Data / interfaces | 1–2 weeks | 4–8 weeks | 2–4 months | Master-data quality, legacy APIs, equipment protocols |
| Pilot / go-live | 1–2 weeks | 4–8 weeks | 2–4 months | Production window, training, parallel run |
| Total (reference) | 4–8 weeks | 3–6 months | 6–18 months | Excludes approvals, tendering, and budget queues |
Timelines are editorial references, not delivery commitments. Actual duration depends on scope, data, interfaces, decision speed, and the production window. Public articles also cite about 1.5–3 months for standardized cloud ERP and 6–12 months for custom on-premises ERP, which is consistent with these ranges.
The most common causes of delay: inconsistent BOM or material codes, legacy systems without APIs, repeated changes to custom reports, approval chains that cross departments, production lines that can only switch at weekends, and insufficient vendor consultant capacity. Write these into a joint responsibility list before signing. It is more useful than “go live as soon as possible” in the contract.
Which hidden costs most often blow the budget?
Master-data cleanup, legacy interfaces, secondary development, permission auditing, training, and downtime or parallel running. These are rarely in the first-year software quote, yet they determine whether the project goes live.
| Hidden cost | When it occurs | Impact | How to control it |
|---|---|---|---|
| Master-data cleanup | Peaks 1–2 months before go-live | Duplicate BOMs, materials, customers, and suppliers make reports and inventory inaccurate | Assign a data owner; freeze coding rules first; accept by sampled accuracy |
| Legacy system interfaces | Blueprint and integration testing | No API, inconsistent fields, repeated interface changes; adds 2–10 weeks | Build an interface list and responsibility matrix first; use read-only integration where possible |
| Secondary development | After configuration reveals that standard functions do not fit | Change orders accumulate; budget and schedule both overrun | Standard first, low-code or configuration second; quote every change order before development |
| Permissions / audit | Multi-plant, multi-role, compliance scenarios | Permissions added after go-live; rework disrupts operations | Define the role matrix, SSO, logs, and audit during blueprint |
| Training / change management | Pilot and rollout | The system goes live but nobody uses it; data returns to Excel | Key-user training, SOPs, performance measures, and on-site support |
| Downtime / parallel running | Cutover and month-end or year-start | Production stops; duplicate data entry; high hidden cost | Choose a low-peak window; pilot on one line; prepare a rollback plan |
| Annual maintenance / upgrades | From year two onward | Perpetual-license maintenance often 15–22% of license; upgrades extra | Budget over 3–5 years; cap annual fees and define upgrade scope in the contract |
The 15–22% maintenance range comes from the Zhiyuan Internet article; the 20% perpetual-license maintenance figure comes from the Tencent Cloud / Blacklake article. Actual percentages depend on the contract and cannot be applied to every vendor.
How should we accept the project and avoid low-price lock-in?
Write acceptance as measurable targets: master-data accuracy, work-order closure rate, inventory accuracy, month-end close time, production-reporting timeliness, and user adoption. The contract should cover scope, interfaces, data export, source code or exit, and annual fees.
| Acceptance item | Definition | Suggested target (adjust by scenario) | Data source |
|---|---|---|---|
| Master-data accuracy | Share of sampled BOMs, materials, customers, and suppliers that match physical stock or contracts | ≥95–98% | Sample audit, ERP master-data tables |
| Work-order closure rate | Work orders reported, received, or closed ÷ work orders that should be closed | ≥90–95% | MES / ERP work orders |
| Inventory accuracy | Sampled SKUs or batches that match the physical count ÷ total sampled | ≥95–98% | ERP / WMS stocktake |
| Month-end close time | Time to complete month-end close, compared with the pre-go-live basis | 30% shorter or an agreed number of days | Finance close records |
| Production-reporting timeliness | Work orders reported in the same shift or day ÷ work orders that should be reported | ≥90% | MES reporting records |
| User adoption | Weekly active target users ÷ total target users | ≥80% three months after go-live | Login and operation logs |
The targets are editorial recommendations, not an industry standard. Adjust them by process, shift, and data maturity. Acceptance should define the measurement period, sampling method, and dispute process; “the system runs stably” is not enough.
At minimum, the contract should specify: module, user, and plant scope; the interface list and responsibilities on both sides; the boundary of data migration and cleanup; pricing for customization and change orders; acceptance metrics and payment milestones; annual maintenance fees and caps on increases; data export formats; whether source code and deployment documentation are delivered; exit and migration support; SLAs and support hours; and security and compliance responsibilities. The most common low-price lock-in is “cheap software, interfaces and changes billed separately.” Before signing, get a total price that includes interfaces and five years of annual fees.
FAQ
Can we do ERP, MES, and OA integration for under RMB 100,000?
Yes, if the scope is a light, standardized, single-plant deployment with a small number of users and 1–3 interfaces. It is usually not enough to replace the core ERP, do deep customization, connect multiple production lines, or cover multiple entities. With a low-cost proposal, confirm whether implementation, interfaces, data work, and annual fees are included.
For ERP, MES, and OA integration, is SaaS or a perpetual license cheaper?
Over three years, SaaS is often cheaper. Over five to ten years, a perpetual license may be cheaper after amortization, but maintenance, upgrades, hardware, and IT labor must be included. Public quote articles give a reference for 50–150-person companies: about RMB 150,000–300,000 for SaaS over three years and RMB 450,000–750,000 for a perpetual license. That is not a universal conclusion.
Can we start with MES and add ERP or OA later?
Yes, and it is common. The order should follow the pain point and the data foundation: start with MES if work orders, production reporting, and quality inspection hurt most; add ERP first if finance and supply chain hurt most; start with OA if approvals and collaboration hurt most. The key is to appoint a master-data owner first, so materials, BOMs, and customer codes are not built three times.
Does ERP, MES, and OA integration require private deployment?
Not necessarily. Companies with standard processes, a single plant, and ordinary data sensitivity can start with SaaS or a hybrid deployment. Private deployment suits multi-plant companies, companies whose core data cannot leave the plant, companies with intranet-only interfaces, and companies with high compliance requirements. Private deployment adds hardware, operations, and security costs; it should not be treated as a synonym for integration.
Our old ERP still works. Should we replace it?
If the old ERP is stable, interfaces are available, and maintenance is acceptable, keep it and add MES, OA, and integration. Evaluate replacement when the vendor has ended support, maintenance is too expensive, critical processes cannot be configured, or data cannot be exported. Calculate the five-year TCO before replacing; do not compare only the new software quote.
For a more detailed comparison of keeping, adding modules, and replacing, a five-year TCO example, and a migration checklist, see “Should you keep or replace your legacy ERP in 2026?”.
What is most often missing from an ERP, MES, and OA integration quote?
Implementation, interfaces, master-data cleanup, training, annual maintenance, internal labor, and downtime or parallel-running costs. Before signing, get a total price and five-year TCO that include these items, and write down the interface list, change-order pricing, data export, and exit terms.
The budget question is not “how much does one suite cost?” It is “how many plants, how many interfaces, how much customization, and who maintains it?” Calculate five years first, choose a tier second, and prove it on one plant or one line.
Sources and caveats
Research cut-off: 24 October 2026. The figures are drawn from public vendor and industry articles and may carry a marketing bias. IDC data is used only to show the direction of the AI-enabled ERP market and does not validate the three budget tiers. The tiers, five-line ratios, five-year TCO examples, timelines, and acceptance targets are editorial judgements and should be adjusted to the company. Lingyi does not publish a single price for its products because price depends on modules, users, plants, interfaces, customization, and deployment; any formal price is subject to a written quotation or contract.
- Digiwin (鼎捷数智), “2026年ERP管理系统多少钱？云部署与本地部署成本全解析,” 20 March 2026. digiwin.com. Vendor article with cloud and on-premises ERP cost structures and a 100-user TCO range.
- Zhiyuan Internet (致远互联), “企业管理软件erp系统多少钱？总成本预算与定价模式剖析,” 13 November 2025. seeyon.com. Vendor article with TCO, implementation, and maintenance ratios for 100–300-person manufacturers; the page notes that some content was AI-generated.
- Blacklake (黑湖), “MES系统多少钱一年？2026年最新价格透明盘点,” 2 June 2026. blacklake.cn. Vendor article with first-year investment and hidden costs for light, mid-sized, and large MES deployments.
- Tencent Cloud Developer Community, reposting “Blacklake Laohai” (黑湖老黑), “MES软件价格对比｜2026年MES系统报价单完整公开,” 13 April 2026. cloud.tencent.com. Repost with a three-year SaaS versus perpetual-license comparison and references for hidden implementation and maintenance fees.
- IDC, “ERP 的下一站，是‘会执行的 AI’,” 8 June 2026, citing 2025 AI-enabled ERP market data. idc.com. China's 2025 AI-enabled ERP market was USD 315.7 million, up 96.1% year on year; the figure excludes traditional ERP revenue.