How to Choose ERP Software for Manufacturing
Turn this article into takeaways for your work.
Each assistant summarizes the article only for you and suggests best practices for your work.
Updated September 2026.
The right manufacturing ERP depends far less on your revenue than on what you make and how you make it. A 40-person job shop building engineer-to-order weldments and a 400-person food processor running batch recipes need genuinely different systems, and neither is well served by the finance-first ERP a generic shortlist hands them. Company size sets your budget. Manufacturing type sets your candidate list, and getting that second one wrong is the expensive mistake.
So this guide is narrower than how to choose ERP software, which covers the category as a whole, and than how to choose ERP for small business, which sizes the decision by headcount. Here the ground is manufacturing-specific: discrete versus process versus mixed-mode, how deep your MRP and bill of materials really need to go, shop floor control, traceability, costing method, and where the ERP stops and MES, PLM, or CAD begins.
Key Facts:
- More than a quarter of organizations exceeded their project budgets, with additional technology the most-cited reason, per Panorama Consulting Group's 2026 ERP Report.
- Food companies covered by the FDA's FSMA Section 204 traceability rule must hand FDA an electronic sortable spreadsheet of traceability records within 24 hours of a request. Congress directed FDA not to enforce before July 20, 2028.
- The FDA's Quality Management System Regulation took effect February 2, 2026, amending 21 CFR Part 820 to incorporate ISO 13485:2016 by reference.
- Manufacturing is what separates the tiers in the one mainstream ERP that publishes list prices: Microsoft Dynamics 365 Business Central charges $110.00 user/month paid yearly for Premium, which adds manufacturing, against $80.00 for Essentials.
Why manufacturing ERP is a different purchase
A general ERP is a finance system with operations bolted alongside. A manufacturing ERP is a production system with finance attached. That sounds like marketing until you look at the data model, where the difference actually lives.
| Capability | Finance-first ERP | Manufacturing ERP |
|---|---|---|
| Bill of materials | Flat kit or assembly list | Multi-level, revision-controlled, with effectivity dates and scrap factors |
| Planning | Reorder points, min/max | MRP netting demand against supply across the full BOM explosion |
| Routings and capacity | Not modeled | Operations, work centers, setup and run times, capacity |
| Shop floor | Not modeled | Work order dispatch, labor and material reporting, scrap and rework |
| Costing | Purchase cost plus overhead | Standard or actual cost by material, labor, and machine, with variances |
| Traceability | Item-level, sometimes lot | Genealogy both directions, plus expiry and quality holds |
If none of those gaps are yours, you may not need this category. Light assembly with stable SKUs is usually better served by an inventory platform, covered in how to choose inventory management software.
Discrete, process, or mixed-mode
This is the first real fork, and the one buyers most often get wrong, because both kinds of vendor say yes to the same demo questions. A discrete BOM assumes fixed inputs producing one countable output. A process formula has to express variable yield, co-products and by-products, potency, catch weight, and unit-of-measure conversions between how you buy, make, and sell. You can force a discrete system to fake some of that with custom fields and scripting, then own that customization through every upgrade.
| Type | Primitives the system needs | You are here if |
|---|---|---|
| Discrete | BOM, routing, work order, serial number | You can disassemble a unit back into parts |
| Process | Formula, batch ticket, yield, by-product, potency, catch weight | You cannot un-bake the cake, and output weight varies |
| Mixed-mode | Both models plus UoM conversion between them | You blend or cook, then fill and count |
| Repetitive or flow | Rate-based schedules, backflush, kanban | You plan in units per hour, not work orders |
Mixed-mode is more common than buyers assume. A supplement company that blends powder in batches and then fills and cartons bottles is mixed-mode, and a pure discrete system handles the bottling fine and the blending badly. Ask which of the four each vendor was built for. Products acquire the other modes later, and the seams show.
How deep your MRP and bill of materials need to go
"Has MRP" is a checkbox nearly every vendor ticks. The useful question is how deep, because buying one rung too low is the most common reason a spreadsheet survives go-live.
| Rung | You need it if |
|---|---|
| Reorder points | Stable demand, short lead times, few components |
| Single-level MRP | Simple assemblies, purchased components only |
| Multi-level MRP | You make sub-assemblies, or sell a part you also consume |
| MRP II with capacity | Machine or labor capacity, not material, is the real constraint |
| Finite scheduling or APS | Changeover cost or setup sequence materially affects throughput |
Bill of materials depth deserves the same scrutiny: revision control with effectivity dates, phantom BOMs for intermediates you never stock, alternates and substitutes, component-level scrap factors, and configurable BOMs if you sell options rather than fixed SKUs. The demo product is always three levels deep and clean. Yours is not.
Build type: make-to-stock, make-to-order, engineer-to-order
Where the customer order enters your process decides which parts of the ERP you live in daily. Most manufacturers run more than one, so the system has to handle the mix rather than the average.
| Build type | What the ERP must do well | Mismatch symptom |
|---|---|---|
| Make-to-stock | Forecasting, safety stock, master production schedule | Project overhead on routine stock runs |
| Make-to-order | Order-driven MRP, promise dates from capacity | Dates quoted from forecast the plant cannot keep |
| Assemble or configure-to-order | Configurator, generated BOMs and routings, CPQ link | Sales quotes builds the plant cannot make |
| Engineer-to-order | Project accounting, estimating, CAD and PLM links | Margin invisible until after delivery |
Configure-to-order is where CRM and ERP have to agree, since a configurator living only in the quoting tool eventually ships an order the plant cannot fulfil as quoted. How to choose a CRM for manufacturing covers that handoff.
Shop floor control and machine data capture
The honest version most demos skip: in many ERPs, "shop floor control" means a screen where an operator types a quantity and a clock time. That is a real capability, and for plenty of shops it is enough. But it is not machine data capture, and vendors use the two phrases interchangeably.
| Method | What you get | Watch out for |
|---|---|---|
| Paper travelers keyed in later | A record, eventually | Data is days stale, so scheduling runs on fiction |
| Operator terminal or tablet | Live work order status, labor and material reporting | Adoption dies past a few taps |
| Barcode or RFID scanning | Fast, low-error material and labor transactions | Label design and scanner ruggedness are real project work |
| Direct machine connection (OPC UA, MTConnect) | Counts, cycle times, downtime reasons, OEE | Few ERPs do this natively; ask what is included |
| Separate MES or MOM platform | A full execution layer feeding summary data to ERP | A second implementation; set the boundary first |
If you want OEE and downtime reasons by machine, assume you are buying an execution layer. Ask vendors to name the protocol and show a live connection, not a screenshot.
Traceability and compliance by sub-industry
Traceability is where a manufacturing ERP earns its cost or exposes you. The test is a recall drill: pick a finished lot, walk backward to every raw material lot in it and forward to every customer who got it. If that means exporting to a spreadsheet, you have records, not traceability.
| Sub-industry | Regulatory driver | What the ERP must carry |
|---|---|---|
| Food and beverage | FDA FSMA Section 204 | Traceability lot codes, key data elements at each critical tracking event, the 24-hour spreadsheet |
| Medical devices | FDA QMSR, incorporating ISO 13485:2016 | Device history records linked to design controls, nonconformance and CAPA, UDI |
| Pharma and nutraceuticals | cGMP, and 21 CFR Part 11 for e-records | Batch records, potency adjustment, quality holds, audited e-signatures |
| Aerospace and defense | AS9100, customer flow-down, export control | Serial genealogy, certificates of conformance, mill certs held through assembly |
| Automotive | IATF 16949, customer-specific requirements | PPAP documents, cumulative-quantity EDI release accounting, label standards |
| General industrial | Customer contracts and warranty exposure | Lot or serial tracking only where the risk sits |
Traceability is often an add-on even in systems built for makers: Katana prices Traceability at $249/month on top of Core. And if you are regulated, put the recall drill in the contract as an acceptance test, not in the evaluation as a demo question.
Costing method: standard, actual, and the argument in between
Manufacturers argue about this more than any other ERP setting, and the argument is worth having before you sign rather than during configuration.
| Method | Fits | Failure mode |
|---|---|---|
| Standard costing | Repetitive production, stable BOMs, mature engineering data | Stale standards make every variance look like noise |
| Actual costing | Job shops, make-to-order, volatile material prices | Needs disciplined floor reporting or the numbers are garbage |
| Average cost | Distribution-heavy operations with light production | Hides the difference between a good run and a bad one |
| Job or project costing | Engineer-to-order and contract manufacturing | Without it, ETO margin is unknowable until close |
Whichever you pick, ask how the system reports variance: purchase price, material usage, and labor and overhead efficiency. A system that computes a job cost but cannot say why it missed standard is doing half the job.
The integration boundary: MES, PLM, CAD, and shop floor hardware
This decision quietly determines how much the ERP has to do, and therefore what it costs. The more the ERP absorbs, the fewer integrations you own and the more compromise you accept in each layer.
| System | Buy a separate one when |
|---|---|
| MES or MOM | You need OEE, machine data, or paperless work instructions |
| PLM | Engineering changes are frequent and audited, or you run multiple CAD sources |
| CAD | Always separate; the ERP integration is the deliverable |
| WMS | Multi-zone warehouses, directed putaway, or high pick volumes |
| QMS | An auditor expects a controlled system, not a document folder |
The trap is buying an all-in-one on the promise it covers all five, then finding in month seven that its PLM is a document folder. Write the boundary down first: which system masters the part number, which masters the BOM revision, and where a change order gets approved. The software integration requirements checklist turns that into something you can hand a vendor.
What to look for: evaluation criteria
Score finalists on these before the demo, not during it. Demos avoid the columns where a product is weak.
| Criterion | What to check |
|---|---|
| Native mode fit | Whether it was originally built discrete, process, or mixed-mode |
| BOM and routing depth | Multi-level, revisions with effectivity dates, alternates, configurable BOMs |
| Planning engine | Multi-level MRP, capacity planning, and how often the run can execute |
| Traceability model | Genealogy both directions, expiry, quality holds, recall reporting |
| Costing flexibility | Standard, actual, and job costing, plus real variance reporting |
| Shop floor reality | What operators touch, on a plant tablet, with gloves on |
The SaaS vendor evaluation scorecard turns these into a scored comparison, and ERP evaluation criteria goes deeper on scoring the implementation partner separately from the platform.
Key questions to ask before a demo
- Was this built for discrete or process manufacturing? Not "does it support," but what it was built for. The answer predicts where the seams are.
- Can you build my hardest real BOM and routing in a sandbox before we sign? Give them revisions and alternates, not a clean sample.
- Show me a recall drill on lot data. Pick a finished lot, walk it both directions, and time it.
- What do operators touch, and on what hardware? Ask to see it on the device you would actually deploy.
- Which capabilities are separately priced modules? Traceability, quality, planning, and warehouse management often are.
- Who implements this, and how many plants like mine have they done? Ask for two references in your sub-industry, and call both.
- What is the metering unit, and what happens at the next threshold? Per named user, per organization, and per resource are all live models here, and mixing them up is the classic budgeting error.
How manufacturing ERP is actually priced
This is the least transparent category in the collection, and it helps to know why rather than read it as evasion. Deals get scoped to plant count, module mix, transaction volume, and a partner's estimate, so a rate card would be wrong for most buyers.
Tier-one manufacturing ERP mostly publishes nothing. Oracle NetSuite publishes no figure anywhere, and its own modules guide says only that module licensing fees vary. Epicor, Infor, SAP Business One, IFS, and QAD all route to a demo request. Acumatica publishes no number either but explains its model on its pricing page: "Pay only for the functionality you need, not for user seats." Plex describes one annual subscription fee "without the complexity of concurrent or named user license."
| Billing unit | Real example |
|---|---|
| Per named user, per month | MRPeasy: $49, $69, $99, $149 per user per month |
| Per organization, usage-metered | Katana Core from $299/month, add-ons on top |
| Per plan with a seat allowance | Fishbowl Inventory Essentials $229/month annually, 2 users |
| No seat charge at all | Acumatica, priced by application and resources; Plex, one annual subscription |
| Modules priced separately | Katana: Manufacturing Management $199, Traceability $249, Warehouse Management $149, all per month |
Two rules travel with every quote. Annual is essentially always cheaper than month-to-month here, so if a rep's math implies otherwise, ask them to rerun it: MRPeasy states plainly that annual billing gives one month free at the same per-user rate. And a computed number is not a quoted number, since module mix and implementation move a total far more than seat count does.
Manufacturing ERP shortlist
Grouped by what each vendor publishes, not ranked. Every price came from the vendor's own page.
Systems with published list prices:
| System | Best for | Published price and term |
|---|---|---|
| Katana | Small makers wanting live production planning | Free $0 (30 SKUs); Core from $299/month per organization, usage-metered; Advantage custom |
| MRPeasy | Real MRP without an implementation project | Starter $49, Professional $69, Enterprise $99, Unlimited $149, per user per month; annual gives one month free |
| Fishbowl | QuickBooks-adjacent shops needing warehouse depth | Inventory Essentials $229/month (2 users), Growth $429, Scale $729; Advanced Manufacturing from $675/month, billed annually |
| Odoo | Manufacturers happy to configure their own system | Standard $8.95 and Custom $13.60 per user per month yearly ($11.20 and $17.00 monthly); One App Free $0 |
| Dynamics 365 Business Central | SMB to mid-market already on Microsoft | Essentials $80.00, Premium $110.00 (adds manufacturing), Team Members $8.00, per user per month paid yearly |
Systems that publish no list price:
| System | Best for | What we confirmed |
|---|---|---|
| Oracle NetSuite | Multi-entity mid-market wanting one cloud suite | No figure anywhere on netsuite.com; its modules guide says only that module fees vary |
| Epicor Kinetic | Mid-market discrete, especially fabrication | ROI material but no price; demo request only |
| Infor CloudSuite | Industry-specific mid-market and enterprise | No published rate on Infor's product pages |
| SAP Business One | Smaller manufacturers wanting SAP fit | No figure on its Business One pages; the dedicated pricing URL no longer resolves |
| Acumatica | Cost driven by usage, not headcount | Its own page states it does not charge for seats |
| Plex (Rockwell Automation) | ERP and MES from one vendor | One annual subscription, no user licensing |
"No published price" is not the same as expensive. It is a sales-process norm in this tier, so get a written quote before ruling anyone out on opacity alone. Best NetSuite alternatives goes tool by tool.
Total cost of ownership
Licence is the smallest number in a manufacturing ERP project and the one vendors quote first. Panorama's 2026 research found additional technology was the most-cited reason projects went over budget, which is another way of saying buyers price the software and not the system.
| Cost layer | Why manufacturing budgets miss it |
|---|---|
| Implementation and partner fees | Scoped by the partner rather than the vendor, and commonly a multiple of first-year licence |
| Data cleanup | Routing standards are often years stale, and MRP output is only as good as its routings |
| Modules bought mid-project | Traceability, quality, planning and warehouse modules surface during configuration |
| Shop floor hardware | Nobody budgets Wi-Fi in a metal building until the scanners drop connection |
| Integration build | CAD, PLM, MES, and OEM EDI are each a project with ongoing maintenance |
| Internal time | The planners and supervisors you need most already run production |
The software total cost of ownership guide has a fuller model, and the ERP implementation guide covers phasing, cutover, and the parallel run.
What actually breaks in manufacturing ERP implementations
The demo always works. These failures surface in month four, when historical data is loaded and the plant is trying to run on it.
| What breaks | What actually helps |
|---|---|
| BOMs migrate incomplete, living in spreadsheets, CAD, and someone's head | Freeze and audit BOMs first; pick one master for part numbers |
| Routing times are fiction, so capacity planning is confidently wrong | Re-time your top jobs before go-live; MRP II is worthless on bad routings |
| Operators bypass a shop floor module built for a desk, not a changeover | Pilot on real hardware with real operators; count taps per transaction |
| Inventory accuracy collapses at cutover, exposing absent counting discipline | Fix count accuracy first; software does not create discipline |
| Traceability has gaps nobody notices until a recall | Run a recall drill in UAT and treat gaps as blocking defects |
| Scope grows to cover a mode the system was not built for | Why the mode question comes first; it is the hardest thing to fix later |
How to choose: a decision framework by manufacturer profile
Match your profile to a direction, not to a single vendor.
| Your profile | Direction | Why |
|---|---|---|
| Light assembly, stable SKUs, no real routings | An inventory platform with BOM support | You would buy planning machinery you never switch on |
| Small discrete maker under roughly 20 users | Katana or MRPeasy | Published pricing, self-implementable, real multi-level planning |
| SMB to mid-market standardized on Microsoft 365 | Business Central Premium | Manufacturing sits in Premium, published at $110.00 user/month paid yearly |
| Mid-market discrete, multi-plant, complex routings | Epicor Kinetic, Infor CloudSuite, or Acumatica | Built for this profile; expect quote-only pricing and a partner-led build |
| Process manufacturer: food, chemical, nutraceutical | A process-native or mixed-mode system | Retrofitted discrete systems accumulate customization debt |
| Regulated production needing full genealogy | Screen on traceability and quality first, price second | The recall drill is the requirement; the rest is negotiable |
| Engineer-to-order or capital equipment | Real project accounting plus PLM links | Job costing alone will not show ETO margin before delivery |
Frequently asked questions
What is the difference between manufacturing ERP and MRP software?
MRP is the planning engine that nets demand against supply and explodes a bill of materials into what to buy and make. Manufacturing ERP wraps that engine in finance, purchasing, inventory, quality, and shop floor control. MRPeasy and Katana sit closer to the MRP end, while a full ERP earns its cost once finance, multiple plants, or regulated traceability enter the picture.
Can a discrete manufacturing ERP handle process manufacturing?
Usually not well, and this mismatch is the most expensive one to fix. Process manufacturing needs formulas with variable yield, co-products and by-products, potency adjustment, and catch weight, none of which a discrete bill of materials expresses natively. You can customize your way there, then carry that customization through every upgrade.
Why do so few manufacturing ERP vendors publish prices?
Because deals are scoped to plant count, module mix, transaction volume, and a partner's estimate. NetSuite, Epicor, Infor, SAP Business One, Acumatica, and Plex all route to a quote. Systems built for smaller manufacturers, including Katana, MRPeasy, Fishbowl, and Odoo, do publish real numbers.
Do we need a separate MES, or is ERP shop floor control enough?
It depends whether you need machine data or just work order status. If operators reporting quantities, labor, and scrap on a terminal covers you, the ERP module is usually enough. If you need OEE, downtime reasons by machine, or paperless work instructions, that is an execution layer.
How long does a manufacturing ERP implementation take?
Longer than the equivalent finance-only project, because BOM and routing cleanup sits on the critical path and rarely compresses. Small cloud MRP systems can go live in weeks when the item master is clean. Mid-market ERP across multiple plants is a months-long program, paced by data readiness and floor adoption rather than configuration.
Making the call
Answer the mode question before you look at a single price. Discrete, process, or mixed-mode decides which vendors can serve you at all, and it is the one choice customization cannot rescue later. After that, size your MRP and BOM depth against your hardest real product, and decide where the ERP stops and MES, PLM, or QMS begins.
Then narrow to two finalists, hand each your messiest BOM and a real recall drill, and get the metering unit, module list, and implementation scope in writing. How to build a software shortlist and how to avoid software buyer's remorse cover the steps either side.
Related reading
- How to Choose ERP Software: the category-wide guide, if manufacturing is only part of your requirement
- How to Choose ERP for Small Business: the same purchase sized by headcount, not production model
- ERP Evaluation Criteria: scoring the platform and the implementation partner separately
- ERP Implementation Guide: phasing, data migration, cutover, and the parallel run
- How to Choose Inventory Management Software: the lighter option if you assemble rather than manufacture
- How to Choose a CRM for Manufacturing: where quoting and CPQ hand off to the ERP
Head of Enterprise Solutions
On this page
- Why manufacturing ERP is a different purchase
- Discrete, process, or mixed-mode
- How deep your MRP and bill of materials need to go
- Build type: make-to-stock, make-to-order, engineer-to-order
- Shop floor control and machine data capture
- Traceability and compliance by sub-industry
- Costing method: standard, actual, and the argument in between
- The integration boundary: MES, PLM, CAD, and shop floor hardware
- What to look for: evaluation criteria
- Key questions to ask before a demo
- How manufacturing ERP is actually priced
- Manufacturing ERP shortlist
- Total cost of ownership
- What actually breaks in manufacturing ERP implementations
- How to choose: a decision framework by manufacturer profile
- Frequently asked questions
- What is the difference between manufacturing ERP and MRP software?
- Can a discrete manufacturing ERP handle process manufacturing?
- Why do so few manufacturing ERP vendors publish prices?
- Do we need a separate MES, or is ERP shop floor control enough?
- How long does a manufacturing ERP implementation take?
- Making the call
- Related reading