Nearly every ERP sold to manufacturers claims work orders. Most of them have the noun. Far fewer have the verb. The gap between holding a work order as a record and running production against it is where the manufacturing execution system market comes from, and it is worth understanding before you spend money on either side of it.
What an MES is actually for
An MES exists to run the shift. Sequencing jobs against real constraints, reporting start and finish by operation, posting labor and material as they are consumed, capturing downtime with reason codes, recording scrap, running in-process quality checks, and producing an OEE number from the events that actually happened rather than from a summary typed in afterwards.
None of that is exotic. It is simply the difference between a system that describes a plan and a system that observes reality. If your ERP can produce a work order document and then has nothing further to say until somebody tells it the job is done, you have a genuine execution gap.
The question that identifies your gap
Ask what happens between releasing a work order and closing it. Specifically: during those hours or days, can anyone answer, from the system, how far along the job is, what has been consumed so far, whether it is tracking to yield, and why it stopped for eleven minutes at two in the morning.
If the answer is no, you have an execution gap. The follow-up question is the important one: is that gap best filled by a second system?
What the integration actually costs
An MES bolted onto an ERP has to agree with it about four things, continuously: inventory, cost, lot genealogy, and order commitment. Each of those is a synchronisation surface, and each one fails in its own characteristic way.
Inventory disagrees because consumption is recorded in one system and the balance lives in the other, so there is always a window where both are right and they differ. Cost disagrees because the MES knows what was consumed and the ERP knows what it was worth, and reconciling them is a month-end activity. Lot genealogy disagrees because the transformation happened in the MES and the trace has to cross a system boundary — which is exactly the seam you cannot afford during a recall. Order commitment disagrees because finished goods exist on the floor before they exist in the order book.
None of these are integration bugs. They are the structural consequence of two systems holding the same facts. A better interface makes the window smaller; it does not remove the window.
The case for filling the gap inside the ERP
If the execution capability lives in the same system as scheduling, inventory, costing, and orders, then the four disagreements cannot arise. The work order the scheduler released is the record the operator reports against and the record that posts to costing. Consumption relieves the same inventory the warehouse is picking from. The transformation is a link in the same lot chain as the receipt and the shipment. Finished goods release against the orders that were waiting on them.
That is not an argument that MES products are bad. Many are excellent, and in a plant running specialised machine-level integration — OPC-UA, PLC data collection, machine vision — a dedicated system is the right answer and an ERP will not replace it. It is an argument that the common case, a company doing production alongside distribution, is buying an integration it does not need.
How to tell which case you are
You probably need a dedicated MES if your requirements are machine-level: direct equipment integration, high-frequency process data, sub-second control loops, or regulatory validation of the equipment layer itself.
You probably need a better ERP if your requirements are operational: sequencing jobs, reporting against operations, posting labor and material, capturing downtime and scrap, holding out-of-spec batches, and knowing what a run cost when it closes. Those are the ones that only became a separate purchase because the ERP could not do them — and they are also the ones where the integration hurts most, because every one of them touches inventory, cost, and traceability.
The blunt version: if the reason you are shopping for an MES is that your ERP has no execution layer, consider replacing the reason rather than working around it. More on how that looks in practice on the manufacturing page, and on what breaks first in when a distributor starts manufacturing.