The short answer: an old ERP is neither automatically due for replacement because it is old, nor safe to keep because it can still issue invoices. Check four things first: whether vendor support has ended, whether business and compliance requirements are still covered, whether integration and data are maintainable, and whether the five-year TCO is clearly higher than replacement. Only when two of the four are flashing red should you move into a replacement evaluation. Otherwise, keep the ERP and add MES, OA, and interfaces.
This is a procurement and migration analysis for manufacturers with 100–1,000 employees, and for CIOs, IT leaders, finance, and operations. Research cut-off: 25 October 2026. It cites SAP and Kingdee official lifecycle pages and IDC market data. Prices are editorial estimates, not a quote from Lingyi or any vendor. Specific support dates, contracts, and migration scope are subject to the latest documents from the software vendor or service provider.
End of support is not a shutdown, but it does mean narrower patches and support. Keeping the system also has a cost. Replacement has to pay back the five-year gap through business benefits.
Why reassess a working ERP now?
Support cycles, compliance and tax rules, system integration, and AI data requirements are all changing. SAP ECC mainstream maintenance runs to 2027. Kingdee K/3 WISE has ended standard support, and some KIS versions end service in 2026. You can keep running, but risk and maintenance cost will rise.
| Trigger | Public fact / example | What it means for the company |
|---|---|---|
| Vendor mainstream support expires | SAP Business Suite 7 / ECC 6.0 mainstream maintenance to end-2027, with optional extended maintenance to end-2030 at extra cost | Not a shutdown, but patches, support, and upgrade paths narrow |
| Old version end of service | Kingdee K/3 WISE has ended standard support; some historical KIS Cloud versions end service on 30 June 2026 or 31 December 2026 | Continued use may lack security patches, tax adaptation, and vendor support |
| Business / compliance change | Multi-plant, multi-entity, tax or audit, cross-border data, or industry regulation requirements | Modifying the old system is expensive, or it cannot meet the new rules |
| Integration and data | MES/OA/CRM/WMS interfaces rely on manual work or Excel; master data is duplicated | Integration and maintenance costs rise every year; data is not trustworthy |
| AI / analytics needs | IDC: China's AI-enabled ERP market was USD 315.7 million in 2025, up 96.1% year on year | AI features are a new budget line, but old systems may not support them |
SAP and Kingdee dates come from official lifecycle pages. IDC counts only new revenue from AI products and modules in ERP scenarios, excluding traditional ERP; it does not mean every legacy ERP must be replaced.
When should you keep, and when should you replace?
Keep the system when the vendor still supports it, processes fit, interfaces are stable, data is clean, and maintenance is affordable. Consider replacement only when support has ended, critical processes are unsupported, interfaces cannot be maintained, or the five-year TCO is higher than replacement.
| Decision area | Keep / optimize | Add modules / hybrid | Replace / rebuild |
|---|---|---|---|
| Vendor support | Still in mainstream or extended support | Support is ending soon, but MES/OA can be added first | Support ended, no security patches, vendor no longer supports it |
| Business fit | Core processes are at least 80% covered | ERP works, but shop floor and office are disconnected | Critical processes cannot be configured and depend on manual work or add-ons |
| Data and interfaces | Master data is stable; interfaces are maintainable | Interfaces can be reworked; data can be cleaned | Data is seriously inaccurate; interfaces cannot be maintained |
| Cost | Maintenance plus internal IT is manageable | Adding modules costs less than replacement | Five-year TCO is clearly lower, or risk cost is high |
| Recommendation | Keep; add reports, interfaces, permissions | Keep ERP; add MES/OA and integration | Replace in phases; no big bang |
The decision table is editorial advice, not a vendor or regulatory conclusion. “Clearly lower” should include business benefits, downtime, and risk, not just the software quote.
Keep, add modules, or replace: what is the five-year TCO gap?
Put the maintenance, internal IT, and interface or patch costs of keeping on the same table as the software, implementation, migration, integration, training, and parallel-run costs of replacing. Do not compare only the new software quote.
A hypothetical example: 100 users, RMB, five years, excluding tax. The numbers are not a quote. Keeping the old ERP: annual maintenance estimated at 15% of the original RMB 300,000 license (public articles commonly cite 15–22%), plus RMB 80,000 internal IT and RMB 30,000 interfaces and patches, for about RMB 155,000 a year. Replacing: first-year software, implementation, data, integration, and training about RMB 500,000, then RMB 100,000 a year.
| Option | First year | Each following year | Five-year TCO | Calculation and assumptions |
|---|---|---|---|---|
| Keep the old ERP | RMB 155,000 | RMB 155,000 | RMB 775,000 | 155,000 × 5; maintenance 45,000 + internal IT 80,000 + interfaces/patches 30,000 |
| Replace the ERP | RMB 500,000 | RMB 100,000 | RMB 900,000 | 500,000 + 100,000 × 4; first year includes software, implementation, data, integration, and training |
| Difference | — | — | +RMB 125,000 | Replacement costs RMB 125,000 more and must be justified by business benefit or risk reduction |
The example is an editorial estimate, not a quote; it excludes downtime, hardware replacement, internal project labor, and business benefits. If the old system is out of support and carries high security risk, the risk cost may exceed RMB 125,000. If the old system is stable and fits the process, replacement may not pay off.
The decision is not “the new system has more features.” It is: can the extra RMB 125,000 over five years be recovered through inventory, lead time, month-end close, labor, compliance, or downtime risk? If the answer is unclear, keep the ERP, add modules, and clean up the data first.
What is the biggest risk in replacing an ERP?
Not software features. Data, interfaces, customization, downtime, and user habits. The older the system, the more undocumented rules it contains, and migration most often fails on reports, exception processes, and master data.
| Risk | Typical manifestation | How to control it |
|---|---|---|
| Master data | Duplicate materials, BOMs, customers, and suppliers; old and new codes do not match | Clean and freeze codes first; accept by sampled accuracy |
| Historical data | Only balances migrate, not detail; audit or traceability breaks | Define migration scope, retention period, and archiving |
| Customization / add-ons | Many reports, plug-ins, and Excel macros; documentation missing | Inventory the custom list; standardize what can be standardized |
| Interfaces | MES/WMS/CRM/finance interfaces have no API and inconsistent fields | Build an interface list and responsibility matrix; read-only first, write-back later |
| Downtime / parallel running | Cutover window too short; duplicate entry; month-end close collides | Choose a low peak; pilot on one plant; prepare a rollback |
| User habits | The system goes live but staff return to Excel; data is incomplete | Key-user training, SOPs, performance measures, on-site support |
| Scope creep | “While we are at it,” add features, change processes, add plants | Freeze scope; quote change orders before development |
The risk list is editorial advice. Real projects should extend it by industry, system version, and plant count.
How many steps does migration take, and how do you avoid stopping production?
Baseline and scope, data and interface inventory, POC and selection, migration rehearsal, pilot and parallel run, cutover, and a four-week post-go-live review. Each step needs an exit condition. Do not cut over the whole plant at once.
| Stage | What to do | Exit condition |
|---|---|---|
| 1. Baseline and decision | Inventory the old system version, support window, processes, costs, and risks | A written decision to keep, add modules, or replace |
| 2. Scope and process | Define plants, modules, users, interfaces, and what is out of scope | Scope frozen; changes go through change orders |
| 3. Data and interface inventory | Master data, historical data, reports, interfaces, permissions | Data accuracy and interface list signed off |
| 4. POC and selection | Run key scenarios with real work orders, BOMs, and month-end close | Key scenarios pass; quotes and contracts are comparable |
| 5. Migration rehearsal | Full and incremental migration; reconcile balances, inventory, and open items | Differences explained; rollback plan works |
| 6. Pilot and parallel run | Pilot on one plant or line; run old and new in parallel for a period | Task closure, inventory accuracy, and month-end close meet acceptance |
| 7. Cutover and review | Cut over at a low peak; review defects, interfaces, month-end close, and adoption for four weeks | Issues closed before rolling out to the next plant |
The steps are editorial advice, not a fixed methodology; trim them to the ERP/MES scope. The four-week post-go-live review can borrow the checklist logic from a website migration, but the objects here are orders, inventory, month-end close, and interfaces.
What should acceptance and the contract cover?
Accept master data, order-to-cash, procure-to-pay, inventory accuracy, month-end close, interfaces, performance, and adoption. The contract should cover scope, data, interfaces, source code or export, annual fees, downtime, and exit.
- Master-data accuracy: sampled materials, BOMs, customers, and suppliers ≥95–98%.
- Order-to-cash: the full chain from sales order to invoice and receipt runs end to end; exceptions are traceable.
- Procure-to-pay: requisition, purchase, receipt, reconciliation, and payment close the loop.
- Inventory accuracy: sampled count match ≥95–98%; batch or serial numbers traceable.
- Month-end close: close time, variance analysis, and report definitions comparable with the old system.
- Interfaces: MES/WMS/CRM/finance success rate and retry mechanism; read-only before write-back.
- Performance and adoption: P95 response time, concurrency, and weekly active key users.
- Contract: module, plant, and user scope; interface list; data migration; change orders; acceptance payments; annual fee cap; data export; source code and deployment documentation; downtime and rollback; exit support.
Who LingMOE is for, and who it is not for
It fits manufacturers whose old ERP still works but MES and OA are fragmented, and who want integration or private deployment. It is not a fit for companies with very simple processes, a stable old ERP, and only a few missing reports.
Lingyi AI Manufacturing Platform (LingMOE) covers ERP, MES, and OA integration for single-plant and multi-plant manufacturers that already have an ERP but have fragmented shop-floor execution, production reporting, quality inspection, or approvals, and that need permission and data boundaries. It can be an option for “keep the old ERP + add MES/OA/integration” or for phased replacement. We do not recommend replacing the whole system when the old one is stable and only a few reports are missing.
For a more detailed three-tier budget and hidden costs, see “What ERP, MES, and OA integration really costs in 2026”. For manufacturing AI procurement timing, see “Manufacturing AI in 2026: policy targets vs. plant reality”.
FAQ
Does SAP ECC stop working in 2027?
No. The end of 2027 is the end of mainstream maintenance for SAP Business Suite 7 / ECC 6.0, not a system shutdown. SAP offers optional extended maintenance to the end of 2030 at extra cost. After that, the system moves to customer-specific maintenance, and patches and support coverage narrow. The system can still run, but security, compliance, and upgrade risk rise.
Can we keep using Kingdee K/3 WISE after support ends?
An on-premises legacy system can usually still start and issue invoices, but the vendor has stopped standard support and may no longer provide defect fixes, security patches, or version updates. Long-term use increases security, tax-adaptation, and maintenance risk. Whether to continue depends on business criticality, patch needs, and migration cost; “it still opens” is not enough.
Our old ERP still works. Why do a five-year TCO?
Because keeping also costs money: annual maintenance, internal IT, interfaces and patches, hardware, downtime, and compliance risk. Replacing also costs money: software, implementation, data migration, integration, training, and parallel running. Comparing only the new software quote leads to the wrong conclusion. Put five years of cash outlay and risk for both options on the same table.
Is replacement always more expensive than keeping?
No. Maintenance, internal labor, and interface costs may rise year after year if you keep the system. Replacement has a high first-year cost, but long-term subscription or operating costs and business benefits may be better. The key question is whether the inventory, lead-time, month-end close, labor, or risk benefits of replacement cover the five-year TCO gap.
Should we replace ERP first, or add MES/OA first?
It depends on the pain point and the data foundation. If the ERP is stable but shop-floor execution, production reporting, and quality inspection are fragmented, add MES/OA and interfaces first. If the ERP is out of support, critical processes are unsupported, or data cannot be maintained, evaluate replacement. Do not replace every system at once in the name of “integration.”
How long does an ERP migration usually take?
There is no universal timeline. A single-plant replacement with clear scope commonly takes 3–6 months. Multiple plants, deep customization, and complex historical data can take 6–12 months or longer. Most of the time goes into master data, interfaces, reports, parallel running, and user training, not software installation.
The legacy ERP decision is not “new or old.” It is support, risk, the five-year total, and business benefit calculated together. If the benefit is unclear, keep the system, add modules, and fix the data. Do not replace for the sake of replacing.
Sources and caveats
This article is public-source research plus editorial estimates, not a first-hand project audit. SAP and Kingdee lifecycle dates come from official pages. IDC data represents only the AI-enabled ERP segment. The five-year TCO is a hypothetical example, not a quote. Support dates, contracts, migration scope, and actual costs vary by version, industry, and region; companies should rely on the latest documents from the software vendor or service provider. We do not promise that replacement will reduce cost, and we do not promise a migration timeline.
- SAP Support, “Strategy / Maintenance commitments” (page updated 1 October 2026). support.sap.com. SAP Business Suite 7 / ECC 6.0 mainstream maintenance runs to end-2027; optional extended maintenance runs to end-2030 and adds 2 percentage points to the maintenance base.
- Kingdee (金蝶), “产品生命周期” (accessed 25 October 2026). kingdee.com/plc. K/3 WISE standard support has ended; KIS Cloud historical versions have end-of-service arrangements on 30 June 2026 and 31 December 2026; refer to official announcements.
- Kingdee (金蝶), “KIS 云历史端产品生命周期停售、停服公告,” 12 May 2026. kingdee.com. KIS Cloud Flagship and Professional V7.0/V16.0 end service on 30 June 2026; Standard and Mini V14.0 end service on 31 December 2026.
- IDC, “ERP 的下一站，是‘会执行的 AI’,” 8 June 2026, citing 2025 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.
- 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. The public article states that annual maintenance for perpetual licenses is commonly 15%–22% of the license fee; the TCO example uses 15% as an assumption, not an official ratio.