TCO for MES: 5-Year Cloud vs. On-Premise Cost Reality

What is Total Cost of Ownership (TCO)?

Total Cost of Ownership (TCO) is the full cost of acquiring, operating, and eventually retiring an asset over its entire lifecycle — not just the purchase price. In manufacturing software, and specifically in an MES decision, TCO is where the cheap system often turns out to be the expensive one, and where the obvious answer — "on-premise is cheaper because we already own servers" — can turn out to be wrong once the full picture is on the table.

This article focuses on TCO in the context where it matters most for SYMESTIC's readers: evaluating an MES or shopfloor-digitalisation platform. The generic definition of TCO has been explained a thousand times. What is rarely explained honestly is what the five-year cost of an MES actually looks like, which costs show up in the vendor quote, and which ones only appear twelve months in.

The five cost buckets every MES TCO must contain

A TCO calculation that only contains "licence plus implementation" is not a TCO calculation — it is a sales quote in disguise. A complete MES TCO spans five buckets:

Two patterns stand out. First, in classical on-premise deployments the implementation bucket is often the largest, larger than the software licence itself, because project work grows with customisation and integration while the sales process anchors on the licence price. Second, cloud-native models shift the weight: software is a predictable subscription, implementation is smaller because connectors and data models are standard, and the internal IT effort is lower.

The TCO iceberg — costs that rarely make it into the first quote

The costs above the waterline (licence, initial implementation) are the ones in the vendor's PowerPoint. Below the waterline sit the costs that determine whether the project succeeds or quietly becomes expensive year after year.

• Customisation that has to be re-done at each upgrade. Every heavy customisation is a technical debt that comes due at the next major release. In classical on-premise MES projects that cost recurs with every major release.
• Integration costs to ERP, PLCs, quality systems. These are almost always underestimated in the quote and true-up during implementation. An ERP interface is rarely a small line item, whatever the statement of work says.
• Internal FTEs for operation. An on-premise MES needs people to keep it running: patching, backups, database administration, user support. Count them at their full loaded cost per FTE; that cost never appears on a vendor invoice, and the worked example below shows how to add it up.
• Downtime during maintenance windows. On-premise systems usually need scheduled downtime for upgrades. Cloud-native platforms are built to update without a maintenance window on the customer side, but ask every vendor what its upgrade process means for your running production.
• Lost productivity during long implementation. Every month between contract and first useful data is a month without the savings the system was bought for. This cost is rarely quantified in TCO models; the worked example below shows how to count it, and for long projects it can be the largest single line.
• Vendor lock-in on exit. If your data model is proprietary and your integrations are custom-coded, replacing the system years later can cost as much as the original implementation did.

A concrete 5-year example — 100 machines across 2 plants

The figures below are a model, not market data. They are assumptions for a scope of 100 machines in two plants over five years, chosen so that the arithmetic can be followed: licence maintenance at 20 % of the licence per year from year 2, internal IT at €120 k per FTE and year, 0.5 FTE per plant on-premise and 0.1 FTE per plant cloud-native, improvement value of €1 m per year once the system is live. Replace each assumption with the figures from your own quotes. The table is meant to show the structure: which cost items exist, how they add up and where the cost of waiting enters.

Three observations from comparing the columns. The cloud-native subscription costs more year on year than on-premise maintenance (assumed €100 k against €60 k) — that difference is real and should not be hidden. In this model it is outweighed by lower implementation, hosting, internal IT and upgrade costs; with other assumptions the result changes, so run the calculation with your own quotes. The last row — the cost of waiting before the system produces results — is the line item that decides many honest TCO calculations and is rarely on a vendor quote. How long a project takes before the first live KPI dashboard differs sharply between products, so ask every vendor for it in writing.

The six most common mistakes in MES TCO calculations

The questions to ask every MES vendor before signing

1. What is the fully-loaded 5-year TCO at our scope, including all items in the five buckets above?
2. What is the expected internal IT effort per year, in FTEs?
3. How long from contract to first live KPI dashboard? (If the answer is measured in many months, you are buying a project, not a product.)
4. What does adding a plant cost in year 3? Adding 50 machines in year 4?
5. What is the upgrade strategy — continuous updates, or paid major versions?
6. What is the exit cost — can we get our data out in a standard format, without paying the vendor to release it?
7. What happens if we customise heavily and then need to upgrade?

How SYMESTIC changes the TCO math

SYMESTIC's cost profile differs from a classical on-premise MES because of how it is delivered: a cloud-native platform with unlimited users, updates included in the subscription and a documented REST API for connecting the ERP. You do not buy or run MES servers or database licences. For machine data you use the Cloud Gateway, free software that runs on your own server or PC, or the IoT Box, a one-time hardware purchase for digital signals. In the Meleghy Automotive rollout across four plants, the scale-up took six months.

This changes the cost structure rather than the price of the software: the buckets in which on-premise projects accumulate cost over five years (servers, upgrades, internal IT) shrink, and the subscription becomes the main recurring item. Whether that adds up to a lower five-year total for your scope is what the five-bucket model above is for.

FAQ

Is cloud MES always cheaper in TCO than on-premise?
No. The outcome depends on scope, contract term and the assumptions behind each cost item. On-premise can still win in heavily regulated environments that require fully on-premise data residency, at very small scopes where a subscription exceeds the modest cost of an existing system, or where strategic on-premise infrastructure is genuinely sunk. Compare both options across all five buckets and over the full contract term.

What is a realistic MES TCO range for a 100-machine, 2-plant scope?
No market-wide range holds up, because quotes differ by scope, number of integrations and contract term. The model above shows how to build your own: an on-premise assumption set that adds up to €2.14 m and a cloud-native one that adds up to €780 k over five years. These are an illustration of the structure, not a market range. Ask each vendor for a quote on your scope across all five buckets.

Why is implementation so much cheaper in cloud-native MES?
Cloud-native platforms are designed as products, not project frameworks. Connectors are standardised (OPC UA, digital signals, a documented REST API), data models are fixed rather than custom-built per customer, and far less server and infrastructure work is needed on the customer side. The effort that an on-premise project spends on environment setup, database tuning, server hardening and bespoke integration is largely absent.

What is the biggest hidden cost buyers miss in MES TCO?
The opportunity cost of slow go-live. As an illustration, if your MES is supposed to unlock €1 m per year in OEE improvements, every month of delayed go-live costs ~€83 k that never appears on a vendor invoice. A 12-month implementation means a €1 m gap that has to be weighed against any upfront licence savings. This is a frequent omission in TCO spreadsheets.

How does SYMESTIC help quantify TCO for a specific case?
The cost side comes from vendor quotes on your own scope. SYMESTIC publishes its prices on the pricing page, so the subscription line of the model can be filled from a public source. For the benefit side, the free 30-minute potential conversation (Potentialgespräch, Microsoft Teams) values the losses at your plant in euros with your own figures; missing values are marked as model values, and the numbers are not passed on. The Meleghy rollout (10 % less downtime and 7 % more output, four plants in six months) shows what the benefit side can look like at one customer; it is not a forecast for another plant.

Related: MES · MES Software Comparison · Cloud MES vs. On-Premise · ROI · SYMESTIC Pricing

About the author

Christian Fieg

Head of Sales at SYMESTIC. 25+ years in manufacturing. Previously iTAC, Dürr, Visteon, Johnson Controls. Six Sigma Black Belt. Led global MES and traceability programmes across 900+ machines, 750+ users and 30+ processes on four continents. Author of "OEE: One Number, Many Lies" (2025). · LinkedIn