Manufacturing software earns its price by getting three things right: bills of material that hold up under real use, planning tied to actual demand, and product costs that come out true. Most tools demo well and fall down on one of the three, which is why a demo alone will not tell you. What follows is the feature list that decides whether a system holds up on your floor, not the one a sales deck leads with.
The examples below lean on Odoo, the system we know deepest, but the checklist applies no matter which system you are evaluating.
The features that matter
Eight of these show up on nearly every serious evaluation. Skip one and the gap shows up on the floor quickly, no matter how clean the demo looked.
Multi-level bills of material that hold. A BOM is not a static list. It needs to nest through subassemblies, carry versions as products change, and update the cost roll-up the moment a component price moves. A system stuck on a flat, single-level BOM will not survive a real product line, and the workaround people build around that gap is what an audit eventually finds wrong.
Work orders and shop-floor tracking. Planning only matters if the floor can report against it. A work order needs to reserve the components it consumes from the same inventory record everyone else reads, and an operator needs a fast way to start, pause, and close one. If the floor still runs on paper while the office runs the software, there are two systems of record, and they will eventually disagree.
Planning from live demand, not a static schedule. MRP, material requirements planning, is the part of the system that turns confirmed orders, forecasts, and current stock into a schedule of what to build and what to buy, updated as reality changes, not redrawn weekly by hand. Reordering rules should trigger routine purchases on their own, and the master schedule should net demand against capacity before anyone promises a date.
Real routing and work centers. A routing that lists operations without timing them, or a work center whose capacity was typed in as a guess, produces a schedule that looks precise and is not. Capacity should come from staffed shifts and honest efficiency, not nameplate speed, with times pulled from a timed pilot run, not a desk estimate.
Quality checks inside the workflow, not bolted alongside it. A quality process that lives in a separate spreadsheet or a standalone tool gets skipped the first time the floor is busy. A check that sits inside the work order itself, with the failure recorded against the specific operation and lot, is the version that survives a busy Tuesday.
True costing: materials, labor, and overhead. A system that only tracks material cost and estimates the rest is telling you a story, not a number. Labor needs to post from actual operation time, overhead needs a deliberate allocation method, and the result needs to match what your accountant sees at month end.
Inventory that reconciles to the floor. If a manufacturing order consumes a component, the stock number needs to drop at that moment, not at a nightly batch job or a monthly count. Raw materials, work in progress, and finished goods should carry their own values, so a distorted number in one category does not quietly warp the other two.
Integration with accounting and sales, on one data model. The alternative to one data model is a set of separate tools kept in sync by hand. One data model means a confirmed sales order, a manufacturing order, a stock movement, and a journal entry all read the same underlying record, so nobody is reconciling three exports at month end.
Must-have vs. nice-to-have
Not every feature on a vendor’s list earns a place in your decision. The table below separates what a system needs from what is a genuine bonus.
| Must-have | Nice-to-have |
|---|---|
| Multi-level bills of material, versioned | AI-assisted demand sensing |
| Work orders tied to live inventory | Predictive maintenance scheduling |
| MRP planning from real demand | Mobile app for floor supervisors |
| Routing and work centers with honest capacity | Built-in EDI for large retail customers |
| Quality checks inside the work order | Configurator for engineer-to-order variants |
| True costing: materials, labor, overhead | Multi-plant consolidated dashboards |
| One data model with accounting and sales | Barcode or RFID beyond basic scanning |
A vendor that leads a demo with the right column and glosses over the left is worth a second, more skeptical look.
Generic ERP, or a system built for how a job shop runs
The feature list above holds for almost any manufacturing business, but the way those features get configured changes with how a shop runs. A repeat producer building the same catalog week after week leans on reordering rules and a master schedule that nets demand against stock, since the product line rarely changes. A job shop building one-off or engineer-to-order work needs configurable bills of material built per job, and routings that flex without a developer touching the setup each time. The same underlying system can serve both, but only if someone configures it for the shop in front of them, not the one a sales template assumes.
Our own Odoo Manufacturing implementation for job shops and repeat producers is built around that distinction. Our manufacturing industry page covers the wider view: production connected to inventory valuation and the monthly close across a whole plant.
Build, buy, or configure
Every manufacturer evaluating software runs into the same three paths. Build custom software from scratch, and you own every future change, bug, and the knowledge gap a departing developer leaves behind. Buy a rigid, off-the-shelf package, and you inherit its assumptions about how a shop runs, then work around whatever does not match. Configure a platform built to be configured, such as Odoo’s manufacturing app, and an implementation process configures existing tools instead of commissioning a codebase to maintain forever.
Fully custom-built manufacturing software is rarely worth it once a configurable platform covers the workflow. Odoo Manufacturing covers bills of material, work orders, MRP and costing without custom development for most shops, and Dynamics and NetSuite do the same at larger scale. Custom development still earns its place for the one process a configurable system genuinely cannot fit, a proprietary test rig or a regulatory step unique to your product, but that is a narrow exception, not the default the word “custom” implies in most vendor pitches.
Red flags
A few patterns show up in nearly every manufacturing software regret, none visible in a polished demo. Ask about them directly before you sign anything.
Per-seat pricing that punishes shop-floor users. Some vendors price every operator who touches a tablet at the same rate as a back-office analyst, which either prices the floor out of the system or forces everyone onto one shared login instead. Ask what a login for every work center costs before comparing headline prices. Odoo publishes its per-user licensing rather than quoting it privately: Standard runs $31.10 per user per month in the US, billed yearly, a real number to hold other quotes against, detailed on our pricing page.
No real costing depth. If a vendor cannot explain how labor and overhead get allocated to a unit, or if “costing” means material cost with a flat markup on top, the number at month end will not match what your accountant already knows.
A separate sync layer holding the system together. If sales, inventory, and accounting sit in different tools connected by middleware or a nightly export, that layer is a future point of failure. Ask who maintains it and what happens the week they leave.
How to run a real evaluation
A features checklist narrows the field. What decides the purchase is a live evaluation run on your own data, not the vendor’s clean demo dataset. Bring your worst bill of material, the one with the most subassemblies and exceptions, and ask the vendor to build it live. Watch what happens to the cost roll-up when you change one component’s price. Ask an operator, not a manager, to run a work order on the tablet, and check what a purchase order looks like when a component runs short mid-build.
A system that handles your worst case will handle the rest of the catalog without much drama. A system that stalls, or needs a developer on the call to keep working, is telling you what the first six months will look like.
Buy for how your shop runs, not for the feature list
Every vendor’s feature list converges on roughly the same items, because manufacturing workflows are not a secret. What separates a system that earns its price from one that gets replaced within a few years is whether it was configured for how your shop runs, not for the version a template assumes. The feature that matters most is not the longest one on the list. It is the one your current bottleneck needs, whether that is a costing method nobody trusts, a schedule that ignores real capacity, or a floor still running on paper while the office runs software.
Frequently asked questions
What features should manufacturing software have?
At minimum: multi-level bills of material, work orders tied to live inventory, planning that runs from real demand (MRP), routing and work centers with honest capacity, quality checks built into the workflow, true costing across materials, labor, and overhead, and one data model shared with accounting and sales. Anything past that list is a genuine bonus, not a requirement.
Is Odoo good for manufacturing?
Yes, for the workflows above. Odoo’s Manufacturing app covers bills of material, work orders, MRP planning, and costing on the same database as inventory, purchasing, sales, and accounting, so the numbers agree without a separate reconciliation step. Index World is an official Odoo Partner that implements it for job shops and full manufacturing operations alike.
Do I need custom manufacturing software or a configurable ERP?
Most manufacturers need a configurable ERP, not custom-built software. A platform like Odoo already covers bills of material, work orders, MRP, and costing without a codebase to maintain. Custom development still earns its place for the one process a configurable system genuinely cannot fit, but that is the exception, not where an evaluation should start.
What is MRP in manufacturing software?
MRP stands for material requirements planning. It nets confirmed orders, forecasts, current stock, and incoming purchase orders into a schedule, then raises the purchase and manufacturing orders that plan requires. Good MRP updates as real demand changes rather than running from a schedule rebuilt by hand once a week.
How much does manufacturing ERP cost?
Licensing and implementation are two separate numbers. Odoo licensing is published: Standard runs $31.10 per user per month in the US, billed yearly. Implementation is scoped to the shop, not billed by the hour, so the total depends on how many work centers, how much bill-of-material data needs rebuilding, and how many locations are involved.
Can manufacturing software handle job shops and custom orders?
Yes, when it is configured for that pattern rather than for repeat production. Configurable bills of material built per job, engineer-to-order workflows, and routings that flex without a developer are what let a system handle one-off and custom work, alongside the reordering rules and master schedule a repeat producer relies on instead.
Bring your worst bill of material to a free demo and watch it get built live, not just described.