Choosing MES software: the questions a presentation cannot answer
Evaluate an MES on three things a presentation cannot answer. Ask the vendor to reconfigure one of your real production processes live in front of you, ask to see the audit trail for a single changed value, and ask for a three-year total cost of ownership including changes made after go-live.
Every MES demo goes smoothly. That tells you nothing.
Every MES demo runs well.
That is not a sign of dishonesty. It is the nature of a demo.
What you are shown is a pre-configured process, with prepared data, on a rehearsed path.
Your plant will not look like that.
Your processes have exceptions. Your products have variants. Your machines have habits that appear in no documentation anywhere.
So the question “can this system do X” is close to useless.
The answer is almost always yes.
There are only two questions worth asking instead.
“Show me now.”
“What happens when the process changes next month?”
Three requests a presentation cannot satisfy
1. Change one of my processes, right now
Bring one real production process.
Pick the awkward one. A process with a parallel branch, a rework step, or whichever exception occurs regularly.
Then ask the vendor to reconfigure the system in front of you to accommodate it.
What you are watching is not the end result.
What you are watching is who does it and how long it takes.
If the vendor’s own product person can make the change in minutes, your team will probably be able to make it later.
If the change has to be scheduled with a development team, every process change in your plant becomes a small project with a cost and a queue.
This is the single most important question in the whole evaluation.
The reason is simple. Production processes always change.
The differentiator is not whether the system can be configured. It is who is allowed to configure it.
2. Show me the audit trail for one changed value
Ask the vendor to change a value in the system.
Then ask to see the record. Value before, value after, who changed it, when, and why.
Follow it with a second question.
Can this function be switched off, administrators included?
If it can, there is no way to prove the audit trail was running when a given record was created.
For plants bound by CPOB or 21 CFR Part 11, the capability alone is a finding.
The first question invites a marketing answer.
The second can only be answered by demonstration.
3. Break down three-year total cost of ownership
Not the licence price.
A breakdown covering all of this:
| Component | What to ask |
|---|---|
| Licensing | Per user, per plant, or per resource? What happens as you grow? |
| Implementation | How many working days, and what is included |
| Integration | Cost of connecting ERP and equipment, per point or in total |
| Infrastructure | Servers, database, and any third-party licences |
| Change after go-live | Cost of adding a product, changing a routing, updating quality rules |
| Support and upgrades | What is included, and what is billed separately |
The bold row is the most consequential.
It is also the least likely to appear in a proposal.
Plants change constantly. New products, new lines, new rules.
If every change carries a cost, years two and three can exceed the initial outlay.
Fit with your type of production
Many MES products are built for one kind of manufacturing and marketed for all of them.
The tells are clear if you know where to look.
Systems rooted in discrete manufacturing are usually strong on serialisation, assembly and unit tracking.
They weaken in batch production. A tank gets treated as a machine with a fixed duration, when a tank actually holds a volume over a period.
The result is a schedule that looks valid and cannot be run.
Systems rooted in process manufacturing are usually strong on recipes, batches and material genealogy.
They weaken on individual unit tracking and multi-level assembly.
The test is simple.
Bring the most unusual production case in your plant. Variable batch sizes, a rework loop, or material splits and merges.
Then ask to see how the system handles it.
That is the part missing from standard demos.
Integration: shipping versus roadmap
Every vendor publishes a protocol list.
What needs separating is which of those run at customers today and which are planned.
Two questions are enough.
- Which protocols are running in your customers’ production right now? Not supported. Not on the roadmap.
- Can I see a connection to real equipment rather than a simulator?
For ERP, ask about direction.
Many integrations run one way only. Orders come down from ERP, but production results do not go back automatically.
If results are still keyed into ERP by hand, most of the integration benefit is gone.
Early signs an implementation will go badly
Some patterns are usually visible during the sales stage.
- Scope is the entire plant at once. A project that produces nothing usable for a year generally loses support before it finishes.
- Nobody asks about your processes. A vendor who presents without understanding your production flow will discover the mismatches during implementation rather than before it.
- Every answer is yes. A product that does everything usually has not been tested against cases hard enough to find its limits.
- Customer references are not reachable. Not a logo list. An actual person who can be asked how the implementation went.
- Small changes need a vendor ticket. That becomes a recurring cost for as long as the system is in use.
A practical evaluation order
- Write down your five hardest production processes, not the most common ones. These become the test material.
- Shortlist on manufacturing-type fit before looking at anything else.
- Insist demos use your processes, not the vendor’s example.
- Make the three requests above. Change a process, show the audit trail, break down three-year cost.
- Call the references, and ask specifically what did not go to plan.
- Pilot one line before committing the whole plant.
Step three is the one most often skipped.
It is also the one that saves the most time later.
Where VECHR MES fits
VECHR MES was built in response to a complaint that kept recurring on plant floors.
Existing MES software was too rigid, and every process change turned into its own project.
So process configuration is designed to be changed by the plant team rather than by its vendor. Specifications, routing and quality rules are changed through a visual designer, not through a development request.
Discrete and process manufacturing run on the same platform. Scheduling treats tanks and pumps as constraints in their own right, with volume and flow rate, rather than as machines with fixed durations.
IIoT connectivity covers OPC UA, MQTT and Modbus TCP. Those three run today, and anything else is on the roadmap rather than in a support list.
Configuration changes can be promoted from development through testing to production, with a change diff and the ability to roll back.
Deployment can start with a single plant or a few lines and grow, so initial scope does not have to cover the whole factory.
Frequently Asked Questions
Is it better to buy MES from the same vendor as our ERP?
Same-vendor integration is usually easier at first. But MES modules from ERP vendors are often weaker on shop-floor execution, because it is not their centre of gravity. What to test is not who the vendor is but whether the module genuinely handles finite capacity scheduling, genealogy and equipment connectivity, or merely provides forms for recording production results.
What is a reasonable MES implementation timeline?
For a single plant with a few lines, months is reasonable. Projects promised in weeks usually cover installation rather than process configuration. Projects planned to run over a year before anything is usable tend to lose organisational support before they finish.
What cost is most often missing from proposals?
The cost of change after go-live. Processes change, new products arrive, quality rules get updated. If every change requires vendor services, years two and three can exceed the initial licence cost. This is the single largest driver of total cost of ownership and the one least likely to appear on a pricing page.
Do we need an MES that handles both discrete and process manufacturing?
You do if your plant genuinely runs both, or plans to. Many MES products are built for one. The clearest tell is scheduling. A system designed for discrete typically treats a tank like a machine, which produces schedules that cannot be executed in batch production.
How do we assess a vendor integration claims?
Ask for the list of protocols actually running in customer production today, not the ones on the roadmap. Then ask to see one real equipment connection rather than a simulator. The difference between supporting a protocol and running it in production is a difference you only feel during implementation.