How to Choose an MRP System: Selection Criteria That Start on Your Shop Floor

Most manufacturers do not start looking for MRP software because they read about it. They start because something keeps going wrong.

A line stops because one component ran out. A buyer pays for rush freight again. The system says the parts are on the shelf, and the warehouse cannot find them. The real production plan lives in one planner's spreadsheet, and when that person is on holiday, everyone guesses. Customers get delivery dates that sales hopes are true.

A new MRP system can fix much of this. It can also inherit every one of these problems and repeat them faster, with nicer screens. Which one you get depends less on the software you choose than on what you check before you choose it.

This article covers what an MRP system actually does, a short self-test to run before you shop, the selection criteria that matter most for a manufacturer, how to test a system with your own products, and the reasons MRP projects disappoint.

What an MRP system actually does

MRP stands for material requirements planning. In plain terms, it answers one question every day: what do we need to buy or make, how much, and by when, so we can ship what we promised?

To answer it, the system combines four things:

• What you need to ship: customer orders, and a forecast if you build ahead.
• What goes into each product: the bill of materials, or BOM, which lists every part, sub-assembly and quantity.
• What you already have and what is on the way: inventory on hand, open purchase orders and jobs already on the floor.
• How long things take: supplier lead times and your own production times.

From those, it works backwards from each due date and suggests purchase orders and production jobs, and warns you when something is going to be late.

That is the whole idea, and it shows the key point: MRP is only as good as the four things you feed it. A wrong BOM, a lead time from three years ago or an inventory count nobody trusts will produce a confident, detailed and wrong plan.

If you are not sure how MRP relates to ERP, our plain-English guide to ERP covers it. In short, MRP is the planning part; ERP usually includes MRP along with accounting, sales, purchasing and inventory in one system.

Before you shop: a five-question self-test

Answer these honestly, ideally with your planner, a buyer and someone from the floor in the room.

1. Do your BOMs match what is actually built? If the floor regularly substitutes parts, adds steps or uses different quantities than the BOM says, the system will plan for a product you do not make.
2. Are the lead times in your current system the real ones? Pick five important parts and compare the lead time on file with how long the last few orders actually took.
3. Does your inventory count match the shelf? If people walk out to check before they trust the number, the plan will not be trusted either.
4. Does the floor record what it uses and scraps? If material leaves the shelf without being recorded, inventory drifts away from reality every day.
5. Do you know where planning really happens today? If the answer is "in a spreadsheet only one person understands," that spreadsheet holds rules your new system will need to know.

If you answered no to two or more, that is not a reason to stop. It is a reason to fix those things before or during selection, not after go-live. New MRP software does not clean up data; it acts on it. Our readiness checklist covers the same ground for a full ERP project.

Standalone MRP or MRP inside an ERP?

You will meet both: software that only does manufacturing planning and connects to your accounting system, and full ERP systems where MRP is one part of a single system.

Neither is right for everyone. The questions that decide it are about your business, not the software:

• How much of your trouble is planning, and how much is everything around it? If quoting, purchasing, inventory, costing and accounting are all struggling, fixing only planning moves the problem rather than solving it.
• How well does your accounting system work today? If finance is happy and the pain is on the floor, a standalone system connected to it can be a sensible step.
• How many connections would you be maintaining? Every connection between two systems is something to set up, watch and fix when it breaks. Two systems that share inventory and costs need a reliable link.
• Where will you be in five years? More products, more sites, more customers asking for traceability? Pick for where you are going, not only where you are.

Compare both paths on the full cost over several years, including setup, connections and the work on your side, not only the licence price. Our article on Tier 2 ERP systems covers the options most mid-sized manufacturers end up comparing.

The selection criteria that matter most

Most MRP systems on a shortlist will claim to do all of the following. The difference is in how well they do it for your products and your people. Treat each one as a question to test, not a box to tick.

1. A bill of materials that matches how you build

Can the system hold your BOMs the way your products really are? That might mean several levels of sub-assemblies, parts that change revision on a date, options and sizes, or products that are configured to order. If your engineers design in CAD, ask how changes reach the BOM, and who checks them before they affect what gets bought.

2. Planning logic your planner can understand

The system will suggest what to buy and make based on lead times, minimum order sizes, safety stock and how it groups orders. Your planner needs to be able to see why it suggested something, change the rules for a part, and trust the result. A plan nobody understands gets overridden in a spreadsheet, and then you have paid for two planning systems.

3. Capacity: does it know your machines and people are limited?

Basic MRP assumes you can make whatever it plans, whenever it plans it. Your shop cannot. Ask how the system shows load on each machine or work centre, what happens when a week is overbooked, and whether scheduling is built in or needs another tool. A material plan that ignores capacity produces dates that look fine on screen and slip on the floor.

4. From planned order to purchase order without retyping

When the system says "buy 500 of this part by the 14th," how many steps does it take for a buyer to turn that into a purchase order with the right supplier and price? Can supplier confirmations and changed dates come back in, and does the plan update when they do? Buyers spend their day here, so small amounts of friction add up.

5. Shop floor recording that people will actually do

The plan depends on knowing what was used, scrapped and finished. If recording it means walking to an office computer and filling in five screens, it will not get done reliably. Look at how the floor would record material issues, scrap and completed jobs: scanners, tablets, a simple screen at each station. Then ask the people on the floor whether they would use it.

6. Inventory accuracy and traceability

Ask how the system supports cycle counting (counting a few items often instead of everything once a year) and how it handles locations, lots and serial numbers if your customers or regulators ask for traceability. Accurate inventory is not a feature of the software; it is a habit the software makes easier or harder.

7. Costing and the link to accounting

When material and labour are recorded, does the cost of each job and each product come out right, and does it reach the books without someone re-entering it at month-end? If finance is not part of the evaluation, this gap tends to surface after go-live.

8. The planner's day

Finally, sit with the person who will live in the system and walk through a normal morning: what changed overnight, what is late, what needs to be ordered today. The system that makes that hour clear and quick is often a better fit than the one with the longest feature list.

How to test a system with your own products

A standard demo will run on a tidy sample product with clean data. That shows the software works. It does not show that it works for you. Our article on what a software demo does not tell you goes into this in detail; for MRP, three things matter most.

Give every software company the same real scenarios. Pick a few situations from your own shop, including the awkward ones, and ask each one to walk through them in the same order. For example:

Here is one of our products with three levels of sub-assemblies. A customer orders it for delivery in four weeks. One purchased part has a six-week lead time and we have only half of what we need. Our assembly cell is already full that week. Show us what the system tells the planner and the buyer, and what date it gives the customer.

Use your own BOMs and lead times. Send a sample of real data ahead of time. How the system handles your messy product tells you far more than how it handles theirs.

Put the planner, a buyer and a floor lead in the room. Give each of them a short scoring sheet built from your needs. They will notice things a leadership team will not.

Ask, for anything that matters, whether what you saw works straight away, needs setup, needs custom work or was built for the demo. It is one of the questions software companies hope you do not ask, and one of the most useful.

Why MRP projects disappoint

When an MRP project goes badly, the cause is rarely a missing feature. It is usually one of these:

• The data moved over as it was. Old BOMs and lead times went into the new system, and it planned from them.
• The spreadsheet never went away. The planner did not trust the suggestions, kept the old spreadsheet running, and the new system became one more thing to update.
• The floor stopped recording. Material use and scrap were not captured, inventory drifted, and the plan drifted with it.
• Capacity was left for later. The material plan was right and the dates were still wrong, because nobody planned the machines.
• Training was generic. People learned which buttons to press but not how their own job changes, so they fell back on old habits.
• Too much was expected from forecasting. Smarter forecasting tools, including AI, still need reliable history to learn from. They cannot make up for years of unrecorded scrap and wrong lead times.

Almost all of these are visible before you sign anything, if someone looks. Our article on why starting too fast is the most expensive mistake covers the pattern.

A practical order for choosing an MRP system

1. Write down how planning really works today, including the spreadsheets, the workarounds and who knows what.
2. Run the self-test above and start fixing the worst data, especially BOMs, lead times and inventory accuracy.
3. Write down what you need, separating what you cannot run without from what would be nice to have, and tie each need to a problem you actually have, such as late deliveries or excess stock.
4. Build a short list of two or three systems that fit your size, your type of manufacturing and your budget.
5. Test each one with the same real scenarios and your own data, scored by the people who will use it.
6. Compare the full cost over several years, including setup, connections, training and support, and read what the proposal does not cover.
7. Plan the work on your side before you sign: data clean-up, decisions about how each process should run, testing and training.

Signs you are buying features, not a fix

• You cannot name the three problems the new system is meant to solve.
• Nobody has compared the lead times on file with real ones.
• The planner has not seen the system, or saw it once in a group demo.
• Scenarios were the software company's examples, not yours.
• Capacity and shop floor recording are "phase two" with no date.
• The cost you are comparing is the licence, not the full cost of getting live.

Each of these is fixable. Each is much cheaper to fix now than after go-live.

Where we fit

We work on the buyer's side. In ERP selection we write down how your business plans and builds today, turn your real problems into scenarios, test every shortlisted system the same way and score the results with your team. If you are not sure whether you are ready to start, the ERP readiness assessment looks at your data and processes first. We sell no software and take no referral fees.

Common questions

What is the difference between MRP and ERP?

MRP plans materials and production: what to buy and make, and when. ERP is a broader system that usually includes MRP along with accounting, sales, purchasing, inventory and often more. Some manufacturers run a standalone MRP system connected to separate accounting software; others use the MRP inside an ERP.

Does a small manufacturer need MRP software?

If you build products from parts you buy, and shortages, rush orders or excess stock are costing you time and money, some form of material planning will help. Whether that means a standalone tool, the planning part of an ERP or better use of what you already have depends on how complex your products are and how much the current way is costing you.

Which MRP features affect inventory costs most?

The planning rules: lead times, order quantities, safety stock and how the system groups demand. Set well, they reduce both shortages and excess. Set badly, or fed with wrong lead times, they create both. Close behind is inventory accuracy, because every suggestion starts from what the system thinks is on the shelf.

How long does it take to implement an MRP system?

It depends on how many products and sites you have, how complex your BOMs are and, above all, how clean your data is. The single biggest thing you can do to shorten the timeline is to start cleaning BOMs, lead times and inventory before the project begins.

When should we bring in outside help?

Before you talk to software companies, while you are still deciding what you need. That is when an independent view changes the most, and costs the least.