iPad Restaurant POS System vs Android vs Windows: Which Holds Up in a Real Kitchen?

Walk the till-software aisle at any hospitality trade show and you will hear the same pitch three times in a row. Every stand has the fastest system, the easiest interface, the happiest customers. Ask a harder question — what does it run on, and what happens when that thing meets a real kitchen? — and the smiles get thinner. That is the question this article answers.

Here is the uncomfortable truth the industry dances around: the software gets the demo, but the hardware gets the blame. When the till freezes mid-service, nobody in the dining room thinks ah, the application layer has encountered an exception. They think the till is broken, the staff look helpless, and your TripAdvisor rating takes the hit. The device in your staff's hands matters at least as much as the app on its screen — arguably more, because you will replace software with a phone call, while a dead terminal on a Saturday night is a crisis you cannot phone your way out of.

So let us compare the three platforms properly: iPad, Android, and Windows. Not on paper specs — on fryer heat, spilled Coke, flour dust, drops off the counter, power cuts, and the general chaos of a working restaurant. Prices are UK high-street reality, not US list prices converted at a fantasy exchange rate.

One disclosure before we start, because it matters: we make CubePOS, and CubePOS runs on iPad. We picked that platform deliberately, and we will tell you why at the end. But the comparisons below are written to be useful even if you never buy anything from us — including the bits where the iPad loses.

Which is best? is the wrong question. Every platform on this list runs successful restaurants. Every platform on this list has also ruined someone's evening. The right question is: which failure modes can I live with?

Because they all fail — just differently. iPads crack and get nicked. Cheap Android tablets slow to a crawl and stop getting updates. Windows machines pick the worst possible moment to install updates. Your job is not to find the platform that never fails. It is to pick the one whose failures are cheapest, rarest, and easiest to recover from in your specific venue. A takeaway doing 200 delivery orders a night has different hardware needs from a 30-cover fine dining room, and both are different from a pub doing wet sales at speed.

Keep that in mind as we go through each contender. We will score them at the end on the things that actually happen in restaurants, not the things that happen in brochures.

The iPad is the default choice for modern restaurant EPOS, and there is a reason half the tills you see in new openings are Apple. It is not because restaurateurs are Apple fanboys. It is because the device solves the two most expensive problems in hospitality tech: training time and support calls.

Hand a new starter an iPad till and they already know how to use the device. Pinch, swipe, tap — the gestures are the same ones they use on their phone fifty times a day. You are training them on your menu layout, not on the hardware. In an industry where staff turnover can hit 70 percent a year and a new hire might work their first Friday shift three days after walking in, that matters enormously. Every hour not spent teaching someone which button wakes the screen is an hour spent teaching them your menu, your allergen procedure, your upsells.

Software updates for years. Apple supports iPads with iOS updates for roughly five to seven years. An iPad bought in 2019 still runs current till apps in 2026. That is a support lifespan no budget Android tablet on earth can match, and it changes the maths completely — more on that in the cost section.

Accessories everywhere. Stands, enclosures, rugged cases, receipt printers, cash drawers, barcode scanners, card readers — if it exists for point of sale, someone makes an iPad version. You are never stuck waiting for a niche driver.

It locks down properly. Guided Access and proper MDM (mobile device management) let you pin the iPad to the till app so staff cannot wander onto YouTube or change the Wi-Fi settings. One toggle, done. On consumer Android this varies wildly by manufacturer.

Resale value. A three-year-old iPad still sells for real money on the second-hand market. A three-year-old budget Android tablet is e-waste with a charger. That resale value quietly subsidises your next upgrade cycle.

It looks right front of house. In a dining room, an iPad on a decent stand looks intentional. A chunky Windows all-in-one looks like the back office escaped. For customer-facing tills and tableside ordering, aesthetics are not vanity — they are part of the room.

The price. A base iPad is around £349, an iPad Air is £500-plus, and that is before the rugged case (£40–£80), the stand (£100–£200), and the receipt printer (£200–£300). A two-till setup with printers and drawers lands around £1,500–£2,000 before software. It is not the cheapest way to put a till on a counter.

Repairs are Apple prices. Crack the screen out of warranty and you are looking at £200-plus at Apple, or a third-party repair of varying quality. AppleCare+ for business softens this but adds to the upfront cost. Budget for breakage like you budget for glassware — because it will happen at roughly the same rate.

Dongle life. Modern iPads have one USB-C port. Printer, cash drawer, scanner, card reader, ethernet adapter — they cannot all plug into one port without a hub, and every hub and adapter is another point of failure. Wireless peripherals (Wi-Fi printers, Bluetooth scanners) solve most of this, but wireless adds its own troubleshooting when the router has a bad day.

Heat. iPads do not like hot kitchens. Leave one on the pass above the salamander and it will dim its screen and throttle performance to protect itself — right when you need it brightest. iPads belong on the counter and in the dining room, not in the 40-degree zone next to the fryer. For kitchen display screens, you want something else.

Theft. iPads walk. They walk out of restaurants at a rate that would embarrass a pickpocket convention. Locking enclosures and Kensington cables are not optional extras; they are part of the purchase. A till that fits in a jacket pocket needs to be bolted to something that does not.

Battery wear. An iPad plugged in for fourteen hours a day, every day, will have a tired battery within two to three years. Apple's optimised charging helps, but plan on the device becoming effectively a wired terminal by year three — which is fine, until someone unplugs it to charge their phone and the till dies mid-shift.

Here is where most comparisons go wrong. They treat Android as one thing. It is not. There are two entirely different categories sharing the name, and they have almost nothing in common except the logo on boot.

Category one is the consumer tablet: Samsung Galaxy Tab A, Lenovo Tab, and the legion of sub-£150 no-name tablets. Category two is the purpose-built POS terminal: Sunmi, iMin, Posiflex's Android line, PAX — devices designed from the circuit board up to sit on a shop counter. A £99 supermarket tablet and a Sunmi T2 are as different as a bicycle and a delivery van. Anyone who lumps them together is selling you something.

The pitch is obvious: why pay £349 for an iPad when a perfectly usable Android tablet costs £129? For a single-till café counting every penny, it is a tempting argument. And to be fair, on day one, it works. The till app opens, the printer prints, the card reader reads. Day one is great.

Day four hundred is the problem. Budget Android tablets live and die by their manufacturer's update policy, and for most budget manufacturers that policy is best described as thoughts and prayers. One or two years of security updates if you are lucky; major OS upgrades, forget it. After eighteen months the till app wants a newer Android version than the tablet will ever get, and you are stuck choosing between an outdated app and a new tablet. That £129 bargain starts looking expensive when you are buying it twice.

Performance decay is real. Cheap tablets ship with just enough RAM and storage to feel fine in the shop. A year of app updates, cached data and a full photo library of menu items later, they crawl. Staff start tapping twice because the first tap did not register fast enough, which queues a double order, which the kitchen then has to void. Death by a thousand lags.

Bluetooth roulette. Android's Bluetooth stack varies by manufacturer, and budget tablets are where the weird bugs live. A printer that pairs flawlessly with a Samsung may drop connection daily on a no-name tablet. When your receipt printer is also your cash drawer trigger and your kitchen ticket printer, mostly works is not good enough.

Chargers kill them. The bundled charger on a cheap tablet is often barely adequate, and when it frays — as cables in restaurants do — staff replace it with whatever is in the drawer. Wrong-voltage charging cooks batteries. iPads are not immune to this either, but Apple's stricter accessory ecosystem means fewer truly awful chargers in circulation.

None of this means consumer Android cannot work. Samsung's A-series tablets, bought new and replaced on a strict two-year cycle, are a legitimate budget strategy — you are essentially renting cheap hardware and accepting the replacement treadmill. What does not work is buying the cheapest tablet on Amazon, bolting it to a stand, and expecting five years of service. That is not a till strategy; that is a slow-motion emergency.

Now the other animal. Companies like Sunmi and iMin build terminals specifically for counters: integrated receipt printers, NFC readers, barcode scanners, customer-facing second screens, proper weighted stands, drop testing, spill resistance. Prices run £200–£600 depending on configuration — more than a cheap tablet, far less than a full iPad setup once you add the printer and stand.

Built for the job. An integrated printer means one power cable, one device, no Bluetooth pairing to debug at 6pm on a Friday. The stand is part of the design, not a £120 afterthought. Drop ratings and sealed buttons assume the device will be abused, because it will be.

Sane pricing. A Sunmi T2-class terminal with built-in printer often costs less than an iPad plus a separate Star printer. For takeaways and quick-service counters where the till never moves and looks do not matter, the value is hard to argue with.

The catches. You are tied to the manufacturer's Android build and their update schedule. The app ecosystem is smaller than iPad's — most serious EPOS vendors support Android, but the long tail of niche integrations (that specific loyalty app, that odd delivery aggregator) often launches iOS-first. Support usually comes through the importer or reseller rather than a high-street store, so vet your supplier. And some units are, to put it kindly, not designed to be looked at. Fine for a takeaway counter behind a screen; less fine in a dining room.

Before iPads ate the point-of-sale world, the restaurant till was a Windows machine — usually a chunky 15-inch all-in-one touchscreen, occasionally a small form-factor PC bolted under the counter with a separate touch monitor. Thousands of UK restaurants still run exactly this setup, and if you walk into an established independent that has been open fifteen years, there is a decent chance the till is Windows.

The industry likes to talk about Windows tills the way people talk about fax machines: with condescending amusement. That is unfair. The Windows till has genuine strengths, and for certain venues it is still the right answer. It also has one legendary weakness that everyone has experienced and nobody has forgotten.

It runs everything. Full Windows means full Windows software: your accounts package, Excel, the CCTV viewer, the music system, the old stock program from 2011 that the previous owner swore by. An iPad runs till apps; a Windows till can be the back office as well as the front counter.

Ports, glorious ports. USB-A everywhere, often serial ports for ancient printers, ethernet built in. If your venue has a receipt printer older than some of your staff, Windows will talk to it. No dongles, no hubs, no Bluetooth pairing rituals.

Big screens for small money. A 15.6-inch Windows all-in-one touchscreen costs £400–£700 — less than an iPad plus stand, with nearly twice the screen area. For a busy counter where the till operator needs big buttons and a visible order queue, screen size is a genuine productivity feature.

Cheap second-hand market. Off-lease business all-in-ones flood the refurbished market. A three-year-old commercial-grade unit for £200 is common, and commercial-grade means it was built to survive an office, which is at least adjacent to surviving a restaurant.

It stays put. A 4kg all-in-one does not walk out in anyone's jacket pocket. Theft is a non-issue in a way it simply is not for tablets.

Windows Update. Let us not dance around it. Everyone reading this has a story: the till that decided 7pm on a Saturday was the perfect time for a feature update; the twenty-minute please do not turn off your computer screen with a queue forming. This is manageable — active hours, metered connections, WSUS or MDM policies, scheduled update windows — but manageable means someone has to actually manage it. Most restaurants do not have an IT person. The till that updates itself is a till that will eventually embarrass you.

Fans versus kitchens. This is the killer, literally. Most cheap Windows all-in-ones have cooling fans, and fans inhale whatever is in the air. In a kitchen-adjacent environment that means flour dust and aerosolised grease, which coat the heatsink, which makes the fan work harder, which pulls in more grease, until the machine thermal-throttles into uselessness or dies outright. Fanless industrial units exist and solve this — at roughly double the price. If you are putting a Windows till anywhere near a kitchen, fanless is not a nice-to-have, it is the whole decision.

It needs mains power, always. No battery. A power cut kills a Windows till instantly — mid-transaction, mid-card-payment, everything. Tablets sail through power cuts on battery. If your area has flaky power or your electrics are vintage, a Windows till needs a UPS (another £80–£150, another thing to maintain), or you accept the risk.

When it crashes, it crashes properly. A frozen iPad app gets force-quit in ten seconds. A truly wedged Windows machine needs a hard reboot and a two-minute boot cycle while staff stand around. Blue screens are rarer than they were, but the recovery time from a bad crash is still the longest of the three platforms.

Security is your job. Antivirus, firewall rules, user accounts, patches — Windows expects an administrator. In practice the till runs as admin with no password because that is what the installer left, and it has not been patched since 2023. A till that also browses the web and runs the office email is a PCI compliance headache waiting for an auditor.

Enough theory. Here is how the three platforms compare on the specific abuses a restaurant inflicts on hardware — the things no spec sheet mentions and every owner learns the hard way.

Tills get knocked off counters. It is not a question of if. A busy bartender pivoting with a tray, a delivery driver squeezing past, a Saturday-night crowd leaning where they should not — gravity always wins eventually. The iPad in a decent rugged case (OtterBox, UAG, or a proper POS enclosure) survives most counter-height drops onto tile; without a case, one drop is often the end. Purpose-built Android terminals are drop-tested by design and usually the toughest of the three. Windows all-in-ones are too heavy to knock over easily, which is their defence — but if one does go over, the screen is a large, expensive target, and the internal hard drive (if it still has a spinning one) does not appreciate the shock. Verdict: rugged Android edges it, cased iPad is fine, Windows survives by being immovable.

A kitchen pass in summer sits at 35–40°C. Consumer electronics are generally rated to operate at 35°C, which tells you everything about how much the designers thought about restaurants. iPads handle heat gracefully but visibly: the screen dims, performance throttles, and in extreme cases you get the temperature warning and a dead till until it cools. Cheap Android tablets handle it worse — they throttle harder and their batteries degrade faster in sustained heat. Fanless Windows industrial units, properly specified, handle heat best of all, because they were designed for factories. But the cheap fanned Windows all-in-one in a hot kitchen is a slow suicide. Practical rule: tablets live on the counter and in the dining room, never on the pass above the heat. Kitchen display screens should be commercial-grade panels rated for the environment, whatever platform drives them.

All three platforms use capacitive touchscreens, so all three suffer equally when fingers are greasy: taps register late, swipes stutter, the screen accumulates a film that makes everything worse. This is a screen-technology problem, not a platform problem, and the fix is the same everywhere — a tempered-glass screen protector (sacrificial, replaceable, £10 for a pack), a microfibre cloth on a hook next to the till, and a house rule that the till operator wipes the screen at the start of each shift. Takes thirty seconds. Almost nobody does it until the screen stops responding properly, at which point they blame the software.

None of these devices are waterproof out of the box. An iPad in a waterproof POS enclosure (there are several good ones) becomes genuinely spill-proof — Coke, coffee, sanitiser spray, the works. Purpose-built Android terminals often have better native sealing around buttons and ports. Windows all-in-ones vary: commercial units with IP-rated front panels exist and are worth the premium for bar environments; consumer all-in-ones treat a pint spilled across the keyboard area as a terminal event. Whatever you buy, assume liquid will find the till eventually and plan accordingly.

Standard kitchen gloves do not work on capacitive touchscreens. Full stop. This surprises people every single time, so let us be clear: it does not matter whether the screen says Apple, Samsung or Dell on the back. Your options are touchscreen-compatible gloves (nitrile gloves with capacitive fingertips, a few pence each in bulk), a tethered stylus on a retractable lanyard, or a hygiene procedure where the till operator is not the person handling raw food. Most restaurants end up with a mix of the stylus and the gloves. Buy the gloves before opening, not after the first health inspection.

This one is decisive and underrated. iPads and Android tablets have batteries; when the power goes, the till keeps selling, the card reader keeps reading (most modern readers have their own battery too), and you look like a genius while the pub next door goes dark. A Windows till without a UPS dies instantly — mid-transaction, possibly mid-card-authorisation, which creates exactly the reconciliation mess you do not want at 9pm. With a UPS it rides through short cuts, but the UPS battery needs replacing every few years and nobody remembers until it fails. If your street's power is reliable and your wiring is modern, this barely matters. If you are in an old building with temperamental electrics, it matters more than any feature on any brochure.

The good news: the big peripheral makers (Star, Epson for printers; Dojo, Zettle, Square, Paymentsense for card readers) support all three platforms now. The remaining differences are at the margins. Windows still has the widest support for legacy and obscure peripherals — if you are keeping a fifteen-year-old kitchen printer alive, Windows is your friend. iPad has the smoothest pairing experience for modern wireless peripherals, and the MFi/Wi-Fi printer ecosystem is mature. Android is fine with mainstream peripherals but verify your exact tablet model against your exact printer model before buying — supports Android on a spec sheet and works with this specific £129 tablet are different claims. For cash drawers, all three trigger via the receipt printer, so the drawer question is really a printer question.

When the internet drops — and in restaurants it drops, usually because someone unplugged the router to charge a phone — what happens to trading? This depends more on the software than the platform, but the platform shapes it. iPad and Android till apps are typically offline-first: they queue transactions locally and sync when the connection returns. Card payments are the wrinkle — most card readers need internet for authorisation, though some support store-and-forward for small amounts (check your provider's terms; liability for failed offline authorisations usually sits with you). Browser-based till systems on Windows die completely without internet, which is an argument for native Windows till software or a 4G failover router (£15/month dongle that has saved more services than any software feature). Whatever platform you choose, test the offline behaviour before you need it. Pull the router plug on a quiet Tuesday and see what breaks. Better you than a Friday.

iOS updates are the gold standard here: predictable, overnight, and they very rarely break till apps because Apple's developer ecosystem is disciplined about compatibility. Android updates depend entirely on your vendor — Google Pixel-class support on a Samsung, silence on a no-name. Windows Update is the meme that earned its reputation; it is fully manageable with active hours and maintenance windows, but only if someone sets them. An unmanaged Windows till updating itself is not a matter of if it ruins a service, but when.

How long before the device is unfit for service? A working rule based on what actually happens in UK restaurants: an iPad lasts four to six years (battery tired by year three, still perfectly usable as a wired till to year five or six). A purpose-built Android POS terminal lasts three to five years (hardware is tough; software support is the limit). A consumer Android tablet lasts one to three years (two if you bought Samsung, one if you bought on price alone). A commercial-grade Windows all-in-one lasts five to eight years, often with an SSD and RAM refresh halfway — but only if it is fanless or far from the kitchen, and only if someone manages the updates. The cheapest option per year of service is rarely the cheapest option at checkout.

Let us make it concrete. It is 7:15pm on a Friday. You have forty covers in, the kitchen is eight tickets deep, a table of six just walked in without a booking, and someone has knocked a full Coke over near the till. Here is your evening on each platform.

On the iPad: the enclosure shrugs off the Coke. The new starter you hired Tuesday is already faster on the till than the veteran was on the old system, because the device needs no explaining. At 8pm the card reader's Bluetooth hiccups — you re-pair it in ninety seconds because you have done it before and the process is sane. The iPad on the pass is dimming from the heat, so you move it six feet left where it belongs. Service continues. Your total hardware drama for the evening: minor.

On the budget Android tablet: the Coke finds the charging port because the case is a £12 generic. The tablet is eighteen months old and has started doing the thing where the screen takes a second to wake — staff have learned to tap twice, which means the table of six just got entered twice, and now someone has to void six items while the queue builds. The kitchen printer dropped its Bluetooth connection at 7:40 and nobody noticed for ten minutes, so two tables are asking where their food is. Service continues, but you have aged a year.

On the Sunmi-style purpose-built Android: the integrated printer keeps churning tickets, the Coke wipes off the sealed housing, the stand does not budge. It is not pretty and the second screen is showing the wrong promo, but nothing has stopped working. Quietly the most boring evening of the three, which is the highest praise hardware can earn.

On the Windows all-in-one: the Coke pools around the base and you spend the evening hoping none of it found a vent. At 7:50 a notification appears: restart required to finish installing updates — because nobody set active hours after the last reinstall. You dismiss it and pray. The big screen is genuinely lovely for the complex split-bill the table of six demands. Then at 9:20 the street's power blips for four seconds. The tablets in the story above would not have noticed. Your Windows till reboots, and you spend six minutes reconstructing the open tabs from memory while the card machine — on its own battery, naturally — judges you silently.

Every one of these evenings is preventable with proper setup: enclosures, active hours, a UPS, tested offline mode. The point is not that any platform is unusable. The point is that each platform's failure mode finds you exactly when you are least able to deal with it, so you should choose the failure modes you are best equipped to prevent.

General advice is useless here; the right answer depends on what you run. Here is the honest breakdown by venue type.

Speed, delivery-printer reliability, tight counters, and a till that never moves. The purpose-built Android terminal (Sunmi class) is the value king here: integrated printer, tough housing, sane price, and nobody cares what it looks like behind the counter screen. iPad is a close second if your EPOS software is iOS-first or you want the longer support life. Avoid the Windows all-in-one — it eats counter space you do not have, and the power-cut weakness is real in older takeaway premises.

Small counter, appearance matters, staff turnover is high. iPad, and it is not close. It looks right, new staff need no device training, and one till plus a card reader is the whole setup. A Samsung Android tablet on a two-year replacement cycle is the budget alternative if the numbers are truly tight.

Spills, speed, wet glasses near electronics, tabs and split bills at pace. iPad in a waterproof enclosure is the standard answer for good reason — it survives the environment and handles high-speed tab management. Rugged Android terminals also do well. Windows only if you are already invested in the ecosystem or the till doubles as the office machine. Whatever you choose, budget for the enclosure or the IP-rated front; a pub till without spill protection is a purchase, not an investment.

Front-of-house aesthetics, tableside ordering, wine-list presentation. iPad, no contest. Nothing else looks as good in a dining room, tableside ordering on a big Windows slab is comedy, and your clientele notices details. This is the segment where the Apple premium pays for itself in perceived quality.

Consistency and remote management beat everything. iPad with MDM is the mature choice — push menu updates, lock settings, track devices across sites from one dashboard. A single-vendor Android POS fleet (all Sunmi, all iMin) is the budget equivalent. What kills multi-site operators is a mixed fleet: two iPads here, a Samsung there, a Windows machine at the original site that nobody dares touch. Standardise on one platform, whatever it is. Your area managers will thank you.

Somewhere you need to run the accounts package, the old stock system, and the CCTV viewer. That somewhere is Windows, and nobody is pretending otherwise. Keep it in the office, on a UPS, managed updates, away from the fryer. Let the tablets handle the customers.

Purchase price lies. What matters is the three-year cost including the replacements and repairs you will actually buy. These are illustrative UK figures for a single counter till with receipt printer and cash drawer — your quotes will vary, but the shape of the maths will not.

Two things jump out. First, the gap between the cheapest and most expensive honest option is about £600 over three years — £16 a month. That is less than the cost of one voided table of four. Hardware is not where restaurants lose money; downtime is. Second, the budget Android route is only cheap if your time is free. Two forced migrations in three years, each eating a half-day of setup and a week of staff grumbling about the new laggy till, is a cost — it just does not appear on an invoice.

Answer these before you buy anything, and the right platform usually picks itself:

1. Where will it live? Dining room or customer-facing counter leans iPad. Fixed takeaway counter leans purpose-built Android. Office leans Windows. Near the pass or fryer: nothing consumer-grade, whatever the badge.

2. Who manages updates? If the answer is nobody, rule out unmanaged Windows immediately. iPad's automatic updates are the closest thing to a till that looks after itself.

3. How reliable is your power? Flaky electrics or an old building? Tablets with batteries, or budget for a UPS on day one — not after the first outage.

4. What peripherals must it talk to? List every printer, drawer, scanner and reader, then verify each one against the exact device model — not the platform, the model. This single check prevents more disasters than any other.

5. Does your EPOS software actually support it well? Not technically runs on Android but the vendor's support team knows this device and half their customers use it. Pick the software first, then buy the hardware your software vendor recommends. Doing it the other way round is how you end up with a till the support line has never seen.

6. What is your staff turnover? High turnover makes the iPad's zero-training advantage worth real money. Stable long-term team? They will learn any platform.

7. How many sites? More than one: standardise on a single platform and a single device model, and buy into MDM from the start. Future you, opening site three at midnight before launch day, will be grateful.

There is no single winner, and anyone telling you otherwise is selling the winner. But there are clear answers for clear situations:

Best all-rounder: iPad. Longest software support, zero device training, best accessory ecosystem, happiest front of house. You pay for it upfront and in Apple repair pricing, and you must case it, lock it down, and keep it away from the fryer.

Best value for a fixed counter: purpose-built Android POS. Integrated printer, tough housing, honest pricing. Verify your software vendor supports the exact model.

Best budget play (with eyes open): Samsung A-series tablet on a strict two-year replacement cycle. Cheap, workable, and honest about what it is — disposable.

Best for the office and legacy setups: Windows, fanless, on a UPS, with managed updates. Keep it away from customers and kitchens alike.

Avoid: no-name budget tablets as a long-term strategy, any Windows till with a fan near a kitchen, and mixed fleets across multiple sites.

And now the disclosure we promised. We build CubePOS on iPad because, for the restaurants we serve — independents, cafés, takeaways, small groups — the iPad's combination of long support life, zero training burden and reliable peripherals won the argument. We have watched customers try the alternatives; the ones who left for a £129 tablet usually came back within two years, poorer in time if not in money. That is our experience, not a law of physics. Run the checklist above honestly, and if it points you at a Sunmi terminal or a fanless Windows box, buy that with our blessing. The best till platform is the one that is still working at 9pm on a Friday — everything else is commentary.