Every EPOS vendor we sit across the table from will tell you that cloud is the future. Some of them are right. Some of them are selling you a roadmap that doesn’t yet exist. And almost none of them will be honest about the trade-offs you only start to feel two years in, when the honeymoon is over and you are living with the consequences of the architecture you signed up for.
We work with restaurant and hotel groups across central London who run on every kind of EPOS estate you can imagine - legacy on-prem boxes nursed past their tenth birthday, brand-new cloud-native tills running in a browser, and every hybrid in between. None of those operators is wrong. They have all just made a different bet about what matters most to their business, and the honest answer is that the right architecture depends on details the marketing brochures never get into.
This is the conversation we wish more vendors were having.
What “Cloud EPOS” Actually Means (And What It Often Doesn’t)
The first thing to clear up is that “cloud EPOS” is one of the most overloaded phrases in our industry. Nine times out of ten, when a vendor says cloud, what they actually mean is that the till on the counter is still a local Windows or Android application running on dedicated hardware, and what has moved to the cloud is the back-office: menu management, reporting, central configuration, user administration. The till itself is still very much a local thing with a local database that syncs.
There is nothing wrong with this architecture. In fact, as I will argue below, it is probably the right architecture for most operators. But it is not what most people picture when they hear “cloud EPOS.” The marketing imagery tends to suggest a browser tab on a tablet talking directly to a data centre, with no local state to worry about and nothing to break on site.
True cloud-native EPOS, where the till really is a thin client and the only thing on the counter is a web browser, does exist. It is rarer than the marketing suggests, and the products that deliver it well tend to be either very new, very simple, or both. Once you are running anything more complex than a coffee shop - table management, course firing, kitchen integration, multi-tender split bills - most browser-only tills start to creak.
Knowing which of these your “cloud EPOS” actually is matters enormously for the rest of the conversation.
The Honest Case for Cloud EPOS
When cloud EPOS lives up to its promise, the wins are real and they are not small.
Central reporting is the obvious one. If you run more than two sites, the ability to see live sales, labour and stock across the estate from a laptop in head office is genuinely transformative. No more waiting until Monday for the weekend Z-reads. No more chasing a manager for a polling export.
Remote configuration is the next. Pushing a menu change, a price update or a new staff member to twenty sites in one action - and knowing it has actually applied - is the difference between a head office that can move and one that cannot. The same goes for software updates: when the vendor handles the patching centrally, you stop being in the business of running an EPOS estate and go back to being in the business of running restaurants.
There is no on-site server to fail, no UPS to forget about, no Windows Server licence quietly expiring in a cupboard. New site openings are faster - in some cases dramatically so - because the rollout is mostly a matter of plugging hardware in and pointing it at an account. And the integration ecosystem around cloud-first products tends to be richer, because everything talks over modern APIs rather than ancient COM or flat files dropped on a network share.
For multi-site operators in particular, these advantages compound. They are the reason most new openings we are involved in default to cloud-managed.
The Honest Case for On-Prem (Still Valid for Some)
I am going to say something that is unfashionable: on-prem EPOS is not dead, and for some operators it is still the right answer.
The pitch is simple. It works without internet. Every single time. The performance is deterministic - the till responds in the same number of milliseconds whether your uplink is up, down or being chewed on by a builder two streets away. There is no monthly subscription that creeps up at renewal. You own the licence. You control the upgrade cycle. And if you have a good local support partner, you can keep a well-built on-prem estate running for a decade on hardware that has long since paid for itself.
The operators for whom this still makes sense tend to share a few characteristics: they run a small number of sites, the menu changes infrequently, the connectivity in the building is genuinely poor, and they have a low tolerance for any operational variability at the till. Think wet-led pubs in old buildings, fine dining where service flow cannot be interrupted, members’ clubs where the till is the same one they have used for years and the staff would mutiny if you changed it.
If that is you, do not let a salesperson shame you out of an architecture that works.
The Hybrid Reality (Which Is What Actually Wins)
Here is the part the brochures gloss over. The architecture that wins for most modern hospitality businesses is neither pure cloud nor pure on-prem. It is what we call cloud-managed, locally resilient.
The till is a real application on real hardware on the counter, with a local database. The back-office, reporting, configuration and updates all live in the cloud. When the internet drops - and in central London it will, more often than your ISP will admit - the tills keep running. They take orders, fire to the kitchen, take card payments through a separately-resilient payments path, and queue everything up. When the connection comes back, they sync.
This is the architecture we recommend by default for almost every multi-site operator we work with, and it is the architecture that survives contact with reality. The vendors who do this well are not always the loudest ones in the room.
What This Means For Your Network
Here is the trap. People hear “cloud EPOS” and think it means they can be casual about their connectivity. The opposite is true. The more your EPOS depends on the cloud - even a well-designed locally-resilient one - the more your network has to be treated as a critical system rather than an afterthought.
That means a properly designed managed network with a primary circuit, a meaningful failover (not a 4G stick taped to a window), VLAN separation between EPOS, payments, guest WiFi and back-of-house, and monitoring that tells you when something is wrong before the manager does. Cloud EPOS hates bad connectivity, and hospitality buildings are full of bad connectivity unless someone has deliberately made them not be.
This is the single biggest thing operators underestimate when they migrate from on-prem. The till stops being self-sufficient, and the network becomes the thing the business runs on.
Cost Models: Capex vs Opex Over Five Years
The cost conversation is more nuanced than the headline numbers suggest. On-prem typically looks more expensive on day one - you are buying hardware, licences and installation up front - and cheaper from year two onwards, because you are not paying a per-till monthly subscription. Cloud is the opposite shape: lower up-front, but a steady drumbeat of monthly cost that grows with your estate.
Over five years, in our experience, the totals are usually closer than either side claims. What separates them is what you get for the money. The cloud subscription typically bundles updates, central management, reporting and a level of vendor support. The on-prem licence typically does not, and you end up paying for those things in other lines on the budget - usually in your own time, or in your IT support contract.
The honest comparison is not licence vs subscription. It is total five-year cost of ownership including hardware refresh, support, network upgrades, training, and the cost of the hours your team spends managing the thing.
Data Sovereignty And Exit
Two questions every operator should ask before signing a cloud EPOS contract, and almost none do.
Where is your data held? For most reputable vendors the answer is a UK or EU region of a major hyperscaler, which is fine. For some smaller vendors the answer is murkier, and worth pinning down in writing.
Can you get your data back, in what format, and over what timescale? This is the one that catches people. If you decide in three years that this vendor is not for you, what comes out of the export button - a clean SQL dump, a set of CSVs, or a PDF? The difference matters enormously when you are trying to migrate. We have seen operators effectively locked in by export formats that could not be reasonably re-imported anywhere else.
These are also cyber security questions. Your transaction data, your customer records, your loyalty database - all of it is now sitting in someone else’s environment, and you need to know how it is protected, who has access, and what happens in a breach.
Migration Risk Is Real
If you are moving from on-prem to cloud, do not underestimate the project. Menus rarely come across cleanly - the data model in the new system is almost never an exact match for the old one, and someone has to make decisions about what to map and what to rebuild. Staff have to be retrained, and the muscle memory of experienced servers is hard to overwrite. Integrations to your kitchen displays, your stock system, your accounting, your reservations platform all need to be rebuilt and tested. A migration we are running right now has been six months in planning before a single till changes hands.
It is doable, and it is often worth it, but it is a project. Treat it like one.
The CloudMatters View
We work with both architectures every day, and neither is universally better. The right answer depends on how many sites you have, how often your menu changes, how good your buildings’ connectivity is, how mature your head office is, and how much risk you are willing to carry on the night.
What we do believe, firmly, is that the decision should be made on the actual trade-offs and not on the brochure. Anyone telling you cloud is always the answer is selling. Anyone telling you on-prem is always the answer is also selling. The job of a good hospitality IT partner is to sit on your side of the table and help you work out which set of trade-offs you can live with for the next five years.
If you are weighing up an EPOS decision, or wondering whether the architecture you have is still the right one, we are happy to have that conversation honestly. Start with our managed cloud services and we’ll take it from there.