When Revlon migrated its North Carolina cosmetics plant to SAP S/4HANA in 2018, the botched go-live left roughly $64 million in orders unfulfilled and contributed to a $70.3 million net loss that quarter. The company also spent an additional $53.6 million trying to mitigate the customer-service fallout. Shareholders sued. Customers rerouted. And the plant, one of Revlon’s core US manufacturing facilities, essentially stopped serving demand for weeks.
Revlon isn’t an outlier. According to Panorama Consulting’s 2025 ERP Report, roughly 68% of ERP implementations fail to meet their original objectives, and the number climbs to 73% for discrete manufacturing specifically. Cost overruns in manufacturing average 215%. Only about a third of implementations finish on time.
If you run a US manufacturing operation on an aging ERP (whether that’s a heavily customized SAP ECC, a 15-year-old Epicor install, or a Sage system no one on your current team originally scoped) you already know why you need to move. What you probably don’t have is a clear plan for getting off it without stopping the line. That’s what this guide covers.
Why Legacy Replacement Is So Dangerous in Manufacturing
Most ERP failure case studies focus on retail and consumer goods. But manufacturing is where a bad go-live hits the P&L hardest, because the downstream effects are physical.
The 2024 Siemens True Cost of Downtime report found that unplanned downtime now consumes about 11% of annual revenue at the world’s 500 largest companies, or roughly $1.4 trillion collectively. Aberdeen Research puts the average cost of unplanned downtime at a manufacturing plant at approximately $260,000 per hour. For mid-sized discrete manufacturers, MaintainX’s 2024 State of Industrial Maintenance survey pegs the number closer to $25,000 per hour. Automotive plants can hit $2.3 million per hour.
When your ERP handles production scheduling, materials planning, and shop-floor data collection, a botched cutover doesn’t just create bad reports. It stops metal from being cut.
Three factors make manufacturing migrations uniquely risky:
- Deep customization. Legacy ERPs have been shaped over 10 to 20 years to match specific routing rules, BOM structures, and MES integrations. Most of that logic isn’t documented anywhere.
- Physical dependencies. Barcode scanners, weigh scales, PLCs, and MES systems all hit the ERP for master data. Break one connection and a whole line goes down.
- Real-time inventory demands. Unlike finance or CRM, a two-hour outage in production means missed shipments, expedited freight fees, and unhappy Tier-1 customers.
The Modern ERP Options for US Manufacturers
Before mapping out a migration, it’s worth knowing what you’re actually migrating to. The manufacturing ERP market has shifted meaningfully in the last five years, and the shortlist most plants are working from today looks different than it did in 2019.
The main categories worth evaluating:
- Enterprise suites (SAP S/4HANA, Oracle Cloud ERP). Deepest manufacturing functionality, especially for process manufacturers and multi-site global operations. Implementation costs regularly cross seven figures and timelines run 18 to 36 months. Best fit for plants above roughly $500M in revenue with complex regulatory requirements.
- Mid-market cloud suites (Microsoft Dynamics 365 F&SCM, NetSuite, Infor CloudSuite Industrial). Solid discrete manufacturing coverage with faster implementations (typically 9 to 18 months). NetSuite has real limitations for complex routing and MES-heavy operations. Dynamics 365 sits closer to the enterprise suites in capability and cost.
- Open-source and open-core platforms. Have matured significantly since 2020. Odoo ERP for manufacturing in US deployments, for example, now handle MRP, quality management, shop-floor scheduling, and MES integration for mid-market manufacturers at a fraction of the licensing cost of the enterprise suites. Best fit for plants under about $250M in revenue that want to avoid multi-year lock-in.
- Industry-specific systems (Epicor Kinetic, Global Shop, IQMS/DELMIAworks, Plex). Purpose-built for specific manufacturing modes (job-shop, plastics, food, automotive tier). Often the strongest functional fit out of the box but with smaller partner ecosystems.
One thing worth stating plainly: a cheaper license doesn’t mean a cheaper implementation. Data migration, integration work, and change management dominate total cost regardless of platform. Panorama’s research shows that 35% of ERP failures trace back to inexperienced implementation partners, so vet the humans as carefully as the software. The right question isn’t which ERP is best in the abstract, but which one fits your production model, integrates cleanly with your existing plant systems, and can be delivered by a partner who has done it before in your industry.
The Hidden Cost Math Most Vendors Skip
When ERP vendors pitch total cost of ownership, they show you licensing, implementation services, and hardware. They rarely quantify the operational cost of the transition itself. That’s the number that kills projects.
A realistic cost-of-migration calculation includes:
- Software and implementation fees (the visible cost)
- Data migration and cleanup, typically 15 to 30% of project cost
- Custom integration work to preserve connections to MES, WMS, PLM, and shop-floor equipment
- Change management and training, the source of 42% of implementation failures per Panorama’s 2025 analysis
- Productivity loss during the parallel-run period, often 3 to 6 months
- Contingency for go-live disruption, typically 4 to 8 weeks of degraded output even in well-run projects
If your plant produces $5 million a month and your line runs 15% below normal for 6 weeks during cutover, that’s roughly $1.1 million in lost margin (assuming a 15% contribution margin) before anything actually “fails.” Plan for it explicitly, because it will happen.
Big Bang vs. Phased: The Honest Trade-Off
There are two mainstream migration approaches, and both work in the right context.
Big Bang cuts over the entire ERP in one weekend. Everything old goes off, everything new comes on. The Hershey’s 1999 disaster (a $112 million implementation that crippled candy shipments right before Halloween) is the textbook cautionary tale. Big Bang is faster and cheaper if it works, but a bad go-live can shut down production plant-wide.
Phased rollout goes live one module or one site at a time. Finance first, then procurement, then production, then multi-plant rollout. It’s more expensive because you’re running two systems in parallel, but a problem in one module doesn’t halt the plant floor.
For most US manufacturers, phased is the safer choice. There are two conditions where Big Bang can actually work:
- Single-plant operations under about $50M in revenue with simple product routings
- Companies replacing an ERP so broken that parallel operation isn’t feasible
If neither applies to you, run phased.
A 6-Phase Playbook That Protects Production
This is the sequence experienced implementation teams follow when the goal is zero unplanned production downtime.
Phase 1: Audit Before You Scope (2 to 4 weeks)
Map every integration point touching the current ERP: barcode scanners, MES, quality systems, tax engines, shipping labels, EDI feeds to major customers. Anything undocumented has to be discovered now, not during cutover weekend.
Phase 2: Clean the Data Before You Migrate It (6 to 12 weeks)
Bad master data is the single most common cause of go-live disasters. Deduplicate customer records, standardize UOM codes, verify BOM accuracy against the shop floor, and archive obsolete SKUs. Panorama’s data attributes 35% of implementation failures to inexperienced project teams underestimating this step.
Phase 3: Configure and Integrate in a Sandbox (8 to 16 weeks)
Build the new system against a full copy of production data. Test every integration. Run mock month-end closes. Simulate a real production day, end to end. If a scenario fails in the sandbox, it will fail worse in production.
Phase 4: Parallel Run Finance First (4 to 8 weeks)
Move general ledger, AP, AR, and reporting to the new system while production stays on the legacy. This shakes out master data issues without touching the plant.
Phase 5: Cut Over Production in Controlled Waves (6 to 12 weeks)
Bring one product line or one work center live at a time. Keep the legacy system as a fallback for at least two weeks per wave. Have MES and finance teams on-site for each cutover weekend.
Phase 6: Sunset the Legacy System (2 to 4 weeks)
Only decommission the old ERP once every process has completed a full month-end close on the new system without exceptions. Keep read-only access for at least 12 months for audit and historical reporting.
The Three Things Manufacturers Consistently Underestimate
Across published post-mortems of failed manufacturing ERP projects, the same three issues show up repeatedly:
- The master-data backlog is bigger than you think. Plan for roughly 3x the cleanup effort you initially scope. Bad BOMs and duplicate part numbers are the most cited root cause of post-go-live production issues.
- The MES and PLC integration work is undocumented. Whoever built those connectors 8 years ago is likely gone. Budget for reverse-engineering.
- Change management isn’t optional. If your shop-floor supervisors don’t trust the new system, they’ll revert to spreadsheets within a week. Train them before go-live, not after. This one variable explains why the same ERP succeeds at one plant and fails at another.
What to Do This Quarter
If your legacy ERP is genuinely holding back production and you’re planning a replacement, the concrete actions worth taking in the next 90 days are:
- Commission an integration audit of every system touching your current ERP
- Pull your last 12 months of production incident logs and quantify the cost of your current ERP’s failures
- Interview at least three implementation partners who have delivered your target ERP inside a US manufacturing plant, and ask each for two references you can call
- Build a realistic cost model that includes the productivity loss of the transition period, not just software fees
The manufacturers who make this transition without stopping production aren’t the ones with the biggest budgets. They’re the ones who plan for the messy middle instead of pretending it won’t happen. Every failure case in the public record (Hershey, Nike, HP, Revlon, MillerCoors) traces back to leadership underestimating one of the phases above. The playbook is knowable. The discipline to follow it is what separates the plants that keep shipping from the ones that spend a quarter apologizing to customers.
Drevian Quenvale writes the kind of ai algorithms and machine learning content that people actually send to each other. Not because it's flashy or controversial, but because it's the sort of thing where you read it and immediately think of three people who need to see it. Drevian has a talent for identifying the questions that a lot of people have but haven't quite figured out how to articulate yet — and then answering them properly.
They covers a lot of ground: AI Algorithms and Machine Learning, Tech Innovation Alerts, Expert Tutorials, and plenty of adjacent territory that doesn't always get treated with the same seriousness. The consistency across all of it is a certain kind of respect for the reader. Drevian doesn't assume people are stupid, and they doesn't assume they know everything either. They writes for someone who is genuinely trying to figure something out — because that's usually who's actually reading. That assumption shapes everything from how they structures an explanation to how much background they includes before getting to the point.
Beyond the practical stuff, there's something in Drevian's writing that reflects a real investment in the subject — not performed enthusiasm, but the kind of sustained interest that produces insight over time. They has been paying attention to ai algorithms and machine learning long enough that they notices things a more casual observer would miss. That depth shows up in the work in ways that are hard to fake.
