Is ERPNext reliable enough for a real manufacturer? An honest limits list

At a glance

If ERPNext is free, what is the catch? Here it is — written by a certified ERPNext partner whose business depends on these implementations succeeding. Six real limits, what each means on your floor and in your close, who it blocks, and the realistic workaround: manual consolidation eliminations, no turnkey US bank feeds, coarse capacity planning with no native APS, calendar-based rather than predictive maintenance, no PLM, and a tamper-evident rather than cryptographically immutable audit trail. Then the counterweight — what ERPNext genuinely does well for $2-20M manufacturers, and the proof that serious operations run on it.

Why a partner that sells ERPNext is publishing its weaknesses

Somewhere in your evaluation, somebody asked the question that put you here: if ERPNext is free, what is the catch? It is the right question. Review sites answer it vaguely, Reddit answers it anecdotally, and vendors do not answer it at all. So here is the catch — all of it that we know, written by a certified ERPNext partner whose business depends on these implementations succeeding.

Simple self-interest, honestly stated: implementations fail when the gap between expectation and reality surfaces in week six instead of week zero. Every limit below will be discovered by your team eventually — during evaluation if you are lucky, after go-live if you are not. We would rather lose a prospect at the article stage than rescue a failed project later. If any item on this list is a hard requirement for your business, ERPNext is the wrong choice and we will tell you so — that is literally the service at our free expert audit.

Is ERPNext reliable enough for a real manufacturer?

Yes, for most $2-20M discrete and process manufacturers: it reliably runs BOMs, work orders, MRP, inventory, and full accounting in one open-source system. It is not the right fit if you need finite-capacity scheduling (APS), PLM, predictive maintenance, or automated consolidation eliminations — those are real gaps, listed below.

Reliable has two meanings and ERPNext passes both differently. As software — does it run, does it post correctly, does it survive an audit — it is mature: nearly two decades of development, GPL-licensed, maintained by Frappe Technologies with a large contributor community, running production workloads worldwide. As a functional fit — does it do everything your most demanding process needs — the answer depends entirely on whether your requirements intersect the six limits below. So let us go through them.

Limits one and two: manual consolidation eliminations, no turnkey US bank feeds

ERPNext handles multiple companies in one instance well — separate books, chart of accounts, and currency per entity, with a Consolidated Financial Statement report that aggregates balances up the tree. What that report does is aggregation. It does not detect and reverse intercompany activity. Elimination entries — intercompany sales, payables and receivables, unrealized profit in inventory — are manual journal entries you book as a defined step in your month-end close. This blocks groups with many entities, minority interests, complex ownership percentages, or statutory consolidation audits that demand an elimination engine. For typical 2-15 entity groups, templated elimination journals plus a close checklist make this a routine step measured in hours, and inter-company invoices are linked documents so the pairs to eliminate are traceable rather than hunted. We cover the full workflow at /solutions/multi-entity. It works; it is not push-button, and nobody should sell it to you as push-button.

Out of the box, bank reconciliation in ERPNext runs on statement imports — OFX and CSV files you download from your bank and load. There is no QuickBooks-style flip-a-switch live feed for US banks. Live feeds via Plaid or similar connectors are achievable, but as scoped integration work per client, not a checkbox. This blocks almost nobody, honestly — but it annoys nearly everyone coming from QuickBooks, so it belongs high on this list. The workaround is weekly statement imports with ERPNext's bank reconciliation tool, or a scoped Plaid integration. Related honesty while we are in the accounting module: there is no US payroll tax engine either — the standard pattern is Gusto or ADP for payroll, with summary journals posted to ERPNext, which also means labor is absorbed into job costs at workstation rates, not payroll actuals. Details at /platform/accounting.

Limits three and four: coarse capacity planning, calendar-based maintenance

ERPNext plans capacity in workstation-hours: it knows each workstation's available hours and can flag overloads. What it does not do is finite-capacity scheduling — automatic sequencing around constraints, changeover optimization, drag-and-drop rescheduling with ripple effects, constraint-based what-if analysis. There is no native APS module. This blocks high-mix shops where the schedule is the competitive advantage — heavy changeover costs, shared bottleneck resources, promise dates quoted off a finite model. If you are evaluating ERPNext specifically to get APS, stop; it is not in there. Most $2-20M shops schedule from ERPNext's work order and capacity data plus a planner's judgment — and that is often a massive upgrade from the status quo (see Texas Stone Designs below: the win was getting all scheduling inputs into one database, not an optimization algorithm). Shops needing true APS pair ERPNext with a dedicated scheduling tool or a custom Frappe app. See /platform/manufacturing for what the native module does cover.

ERPNext's asset maintenance supports scheduled preventive maintenance — tasks on a calendar, assigned, logged. It does not do condition-based or predictive maintenance: no sensor or IoT ingestion out of the box, no vibration or runtime-hours triggers, no failure prediction. This blocks plants running reliability-centered maintenance programs on critical rotating equipment, where maintenance timing is driven by condition data. Calendar-based preventive maintenance in ERPNext covers what most $2-20M plants actually run today, which is often less than that. Condition-based programs keep their CMMS or build sensor ingestion on Frappe as custom work.

Limits five and six: no PLM, tamper-evident not immutable audit trail

ERPNext manages BOMs with versioning and has basic engineering-change support. It is not a product lifecycle management system: no CAD integration, no managed engineering-change workflow with impact analysis across open orders, no design-document lifecycle control, no requirements traceability. This blocks engineer-to-order and regulated-design manufacturers whose revision control burden lives upstream of the BOM — if your pain is managing the design process, ERPNext addresses the manufacturing consequence of that process, not the process itself. Keep your PLM or PDM as the engineering system of record and treat ERPNext's BOM as the manufacturing BOM, synced at release. That boundary is normal even in tier-1 ERP deployments.

Submitted documents in ERPNext cannot be silently edited — changes go through cancel-and-amend, which preserves the cancelled original and links the amendment, and a version log records edits. That is tamper-evident: alterations leave a visible trail. It is not immutable: a user with sufficient system privileges operates the database underneath. No mainstream ERP on a conventional database is truly immutable, but some vendors imply it; we will not. This blocks organizations under regulation that mandates cryptographically immutable, WORM-style records at the application layer. For standard financial audits, cancel-and-amend plus role permissions plus database backups is the accepted control set — auditors work with it routinely. Where stricter guarantees are required, append-only log shipping and restricted database access are infrastructure-level controls, scoped per client.

That is the list. If none of the six blocked you, ERPNext deserves its shortlist slot. If one did, book the free audit anyway and we will confirm it is a real blocker — our honest answer includes don't buy ERPNext from us.

Who should NOT use ERPNext?

Skip ERPNext if you require automated multi-entity elimination engines, native APS or finite scheduling, PLM with engineering change control, predictive maintenance, or regulator-mandated immutable (not just tamper-evident) audit logs. Businesses with those hard requirements are better served by specialized or tier-1 systems — and should budget accordingly.

One addition that is not a feature gap: if your organization expects a vendor to own every problem end-to-end and has no appetite for a partner-delivered open-source model, the cultural fit matters as much as the functional one.

What ERPNext does genuinely well for $2-20M manufacturers

The counterweight, held to the same standard of specificity:

• One database for GL, MRP, and inventory. A work order consumes stock and books WIP in the same transaction; a shipment relieves inventory and posts COGS in the same posting. No sync layer between operations and accounting, because there is nothing to sync — this single property eliminates the reconciliation labor that defines the QuickBooks-plus-inventory-add-on stage of growth.
• No per-user, per-module license fees. GPL-licensed. Twenty users or two hundred, the software cost is identical: zero. Your money goes to hosting, implementation, and support — see our published pricing. The per-user line most ERP budgets are built around simply does not exist.
• Real manufacturing depth at this size. Multi-level BOMs with operations, scrap, and costing; work orders; MRP; subcontracting; lot and serial traceability; multi-warehouse stock with valuations.
• Open schema, open code. Every table is inspectable, every workflow extensible via the Frappe framework. When your process does not match the system, the system can be changed — without renegotiating a license.
• Speed to live. Our typical implementation for this segment runs 6-8 weeks. (Typical for scoped projects at this size — not a universal promise; multi-entity and heavy-customization projects run longer, and the parallel run is roughly 1.5x workload for finance while it lasts.)

Does anyone serious actually run on ERPNext?

Yes. Our clients include Texas Stone Designs (scheduling cut from 24 hours to 5 minutes) and Mr Goli Soda (70 franchises, 45 factories). The Frappe ecosystem documents manufacturers like beadedstream and Piper Plastics — verifiable at frappe.io. ERPNext runs real production floors, not just demos. Sorted by what we can personally vouch for — a distinction most vendor proof sections blur, so we will keep it sharp.

Our clients, first-hand ERPNova and VGTS implementations: Texas Stone Designs, a Texas stone manufacturer, saw production scheduling go from a 24-hour manual assembly job to 5 minutes — not from an APS algorithm, but because orders, inventory, BOMs, and capacity finally lived in one database. "VGTS turned production scheduling from a daily headache into something fast, simple, and visible to the whole team," says CEO Micah Grant. Mr Goli Soda, an Indian beverage franchise, runs 70 franchises and 45 factories on one ERPNext instance, each franchise a separate Company, after moving off Odoo with roughly 80% lower ERP cost and franchise onboarding cut from 60 to 35 days.

Frappe ecosystem stories (not our clients — published case studies you can verify at frappe.io): beadedstream, an Alaska environmental-sensing hardware manufacturer, and Piper Plastics, a plastics manufacturer, both documented running ERPNext in production. We cite them because independent verification beats our word; we label them because claiming them would violate the premise of this article.

How to pressure-test ERPNext against your process before committing

Do not trust this article either — test it:

• Write down your 15-20 core processes — the ones that actually run the business, not a 400-line RFP.
• Check each against the six limits above. Any hard collision is a disqualifier; find it now.
• Demand a scripted demo with your data — your items, your BOMs, your nastiest order. You drive the agenda.
• Run one real order end-to-end in a sandbox — quote to cash, including the accounting entries. An hour in a sandbox beats ten demo hours.
• Call references your size. Ask for one measurable before and after number, not a satisfaction rating.

Where this fits in a proper evaluation

This is week 2-4 of a proper six-week evaluation — the full plan, with checklist, is in our ERP selection guide for $5-20M manufacturers.

Want the shortcut? Get a free expert audit — we will map your processes against ERPNext, including every limit on this page, and give you a straight fit or no-fit answer.