ERP for Manufacturing: What Off-the-Shelf Gets Wrong

I spent three days in a manufacturing facility last month watching their production process. Twelve different factories. Complex multi-step processes that vary by product family. Custom tooling that requires setup time that's unpredictable. Vendor-supplied components with quality variance. Rules about what can be produced where, what requires manual inspection, what can be batched together. Rules layered on top of rules, exceptions to exceptions.

Then I looked at their SAP implementation, which cost them $2 million and took two years. SAP models manufacturing well for textbook scenarios: linear production flows, standard bill of materials, deterministic lead times. But this company's actual production was nothing like the textbook.

The gap between how manufacturing actually works and how generic ERP systems model manufacturing is enormous. And that gap is where all the value of custom ERP lives.

The Problem With Generic Workflows

Most ERP systems start with a standard manufacturing workflow: demand → plan production → procure materials → manufacture → deliver. It's a nice linear story that makes sense as a default. But real manufacturing is Shakespearean—full of plot twists, unexpected delays, quality problems, and people making on-the-fly decisions.

The distributor I mentioned in March had a production facility that made custom components. Customers ordered non-standard configurations. The facility could sometimes combine orders into a single production run (saving time) or sometimes had to produce them separately (more expensive but necessary). The decision was made by a production manager based on factors that existed nowhere in the ERP's data model: gut feel, recent quality issues with combined runs, tooling availability three weeks out.

SAP could model "combined runs" versus "separate runs" as two different process types. But it couldn't model "sometimes combined based on factors that change weekly." So the process became: use SAP for standard orders, use spreadsheets and manual tracking for custom orders. Within six months, they had real data in both systems with no way to reconcile them.

That's what breaks ERP implementations in manufacturing: the assumption that process is deterministic and can be modeled upfront. Real manufacturing is probabilistic and emergent. It adapts based on current conditions.

Where Custom ERP Wins

When we build manufacturing ERP, we start by understanding that the actual process is more important than the theoretical process. We model reality first, then add constraints on top of that reality rather than trying to force reality into pre-built constraints.

The cost allocation example from March is a perfect illustration. A manufacturing company has multiple factories, different overhead structures, custom allocation rules based on products and volumes. Generic ERP systems have cost allocation rules that work for the first six months, then require constant exception-handling as real-world complexity emerges.

With a custom system, you model cost allocation as data, not as hard-coded business logic. The finance team defines rules: "Product line A allocates overhead at rate X. Product line B allocates at rate Y. If monthly volume exceeds threshold Z, use alternative rate W." Now when the business changes—which it always does—the rules update without touching code.

Similarly, scheduling. Most ERP systems model scheduling as a data input: you define how long an activity takes, and the system calculates dates. But real manufacturing scheduling is constraint optimization: what's the minimum time to complete all work given machine availability, quality hold times, setup requirements, and labor availability? Generic systems punt on this and pass it back to humans using spreadsheets.

A custom system can embed constraint optimization, learn from historical data about what estimates are accurate, and proactively flag when a timeline is impossible before you've committed to it.

The Economics of Custom Versus Off-the-Shelf

This is where I have to be honest: custom ERP is not cheaper upfront. A $2 million SAP implementation versus a $1.5 million custom ERP sounds like a win for custom. But if the SAP system breaks six months post-launch because it can't handle real processes, the real cost of SAP is $2 million plus eighteen months of process workarounds plus eventual replacement.

The question isn't "What's cheaper?" It's "What delivers value for the actual business model?"

For a company running textbook manufacturing—standard products, predictable processes, clear workflows—off-the-shelf ERP is absolutely the right choice. You get support, upgrades, and a system that's been battle-tested by thousands of companies.

For a company running non-standard manufacturing—customized products, complex processes, rapid iterations on how things are built—custom ERP is the only choice that actually works. Off-the-shelf will break under the weight of reality.

I've audited fifteen manufacturing companies running SAP. Three of them are genuinely happy. Twelve of them are frustrated because the system works fine until reality diverges from the model. Then it becomes an obstacle instead of a tool.

The Decision Framework

Here's what I recommend for manufacturing companies evaluating ERP in November 2018:

First, map your actual process versus textbook process. If they're similar, buy commercial. If they're different, document the differences and estimate the cost of workarounds. If workaround cost is significant, custom ERP becomes attractive.

Second, think about how much your process changes. If your manufacturing process hasn't changed fundamentally in five years, commercial ERP can probably handle it with configuration. If your process changes quarterly because you're adapting to market demand or customer requirements, that volatility favors custom systems that can adapt without complete reimplementation.

Third, assess your technical capacity. Custom ERP requires ongoing partnership with your development team. You need to evolve the system as the business evolves. If you're a manufacturing-focused company without strong technical organization, that becomes a liability.

The companies winning in specialized manufacturing aren't trying to force their unique processes into commercial software. They're building or buying systems that start with their actual processes as the foundation.