Choosing a Restaurant POS System: The Cost Structure Nobody Shows You

On a restaurant running $50,000 a month in card sales, payment processing costs roughly $750 to $1,750 a month depending on the rate you end up with. The POS software subscription underneath it commonly runs $0 to $300. That ratio — processing costing several times the software, every month, for as long as you're open — is the most important fact about buying a point-of-sale system, and it's the one buried under the features grid on every comparison page you'll read.

That's arithmetic, not a statistic: 1.5% to 3.5% of $50,000 against a published subscription price. Run it on your own volume and the shape doesn't change. So the useful question isn't "which POS has the best reporting." It's "what is my all-in processing cost under this vendor, at my ticket size, and what does it cost me to leave." Nobody selling you a system volunteers that number, and the average in an article isn't yours. Here's how to compute it.

No vendor recommendation here. Products get named only where a specific published fact needs a concrete example.

The Five Things a POS Actually Costs

Run this list with every rep on every call. A quote missing one of the five isn't a quote.

Hardware

One-time if you buy, ongoing if you lease.

Source: UpMenu

Ranges, not quotes — screen count and whether you need a printer in three stations move the number fast. The handheld line is worth pausing on: a truck, a coffee counter, or a small fast-casual concept can open on a reader and a tablet for a couple hundred dollars. Whatever you land on belongs in your technology line in the startup budget, not your operating plan.

Software subscription

Charged per terminal or per location, monthly. The real spread runs $0 to $300–$400+.

Two published examples, purely to show the shape of the market — verify current pricing on each vendor's own page, because tiers change:

• Square publishes a free plan with in-person processing around 2.6% + $0.10–$0.15, and paid Plus and Premium tiers around $60/month and roughly $149–$153/month. Source: Square
• Toast publishes a Starter Kit at $0/month software with pay-as-you-go processing around 3.09–3.69% + $0.15, and a Point of Sale plan around $69/month per location with processing around 2.49% + $0.15. Source: UpMenu · Source: Loman

Look at what those two have in common. The $0 software tier is not free. It's the same cost allocated differently — the software is subsidized by a higher processing rate, and on real volume you pay for it many times over. In Toast's own published numbers, the gap between the $0 tier and the $69 tier is roughly 0.6 to 1.2 points of processing. On $50,000 in monthly card sales, 0.6 points is $300 a month to save $69. That's the most common first-year mistake in this category.

Payment processing

The big one. Its own section below, because it's the reason this page exists.

Ancillary and pass-through fees

The line items reps don't lead with, most of them on the processing statement rather than the software invoice:

• PCI compliance fee — a small recurring charge, sometimes with a much larger annual non-compliance fee if you never complete the self-assessment questionnaire. Complete it. That's all you need here.
• Monthly statement, batch, or minimum fees — flat charges, small individually.
• Chargeback fees — per incident, commonly quoted somewhere in the $15–$40 range, and charged whether or not you win the dispute. Ask for your actual figure in writing.
• Gateway fees for card-not-present transactions, priced separately from your card-present rate.
• Instant or next-day funding fees — usually a percentage of the transfer, in exchange for money moving faster than standard settlement.

None of these appear on a pricing page. Individually they run from a couple of dollars to a few tens of dollars a month; collectively they're a fraction of a point on top of your headline rate — small enough to skip in a pitch, large enough to change which quote wins. That's why the math below counts the whole statement, not the advertised percentage.

Implementation and training

One-time onboarding, installation, and menu-build fees. Sometimes real money, sometimes waived as a closing concession — which is the point. Ask what it is, then ask whether it's waivable, because it frequently is and nobody offers unprompted. Also ask who builds the menu: if it's you, that's a block of the week before opening, the week you have the least time.

Why Processing Dominates — and How to Compute Your Own Rate

It's arithmetic. Once you can do it, you can compare any two vendors on equal footing in ten minutes.

The two pricing models

Flat-rate. One quoted percentage plus a fixed per-transaction fee on every card regardless of type. Predictable, easy to forecast, standard on bundled cloud POS. You pay the same on a debit card as on a premium travel rewards card, and the processor keeps the difference.

Interchange-plus. You pay the actual interchange rate the network sets for that specific card — varying by card type and by how it was presented, chip versus tap versus keyed — plus network assessments and a negotiated markup. Less predictable month to month, more transparent, generally cheaper at real volume.

Card-present interchange-plus commonly lands around 1.5–2.5% + $0.10–$0.15 all-in, against a quoted flat rate of 2.6–2.9% plus a fixed fee. Source: Toast · Source: Lightspeed

Flat-rate is simpler; interchange-plus is usually cheaper once volume is meaningful. A concept doing $8,000 a month in cards has almost no leverage and should optimize for simplicity. A concept doing $50,000 has leverage and should use it.

The formula

Effective rate = total processing fees paid in a period ÷ total card sales in that period.

Total fees means everything on the statement — percentage, per-transaction, PCI, statement, batch, gateway. Not the headline rate. A processor quoting "2.49%" whose statement shows $1,580 on $50,000 in card sales is charging you 3.16%, and both numbers are true.

Two variables decide which model wins, and neither is on a marketing page.

Your average ticket. A fixed per-transaction fee is regressive — it hurts small tickets disproportionately.

A coffee shop at a $9 ticket and a steakhouse at a $60 ticket are having completely different conversations about the same quote. Small ticket: push on the fixed fee, treat the percentage as secondary. Large ticket: the reverse.

Your card mix. Rewards cards carry higher interchange. A neighborhood spot taking mostly debit and a business-lunch restaurant taking mostly premium credit see different rates under interchange-plus and identical ones under flat-rate — which is why flat-rate quietly favors the expensive card mix and penalizes the cheap one.

A worked example — illustrative arithmetic, not a vendor comparison

The rates below are the published and commonly quoted ranges cited earlier, run through the formula. They are not quotes anyone gave anyone, and the columns are not stand-ins for named products. One hypothetical restaurant: $50,000 in monthly card sales at a $28 average ticket, about 1,786 transactions, priced three ways.

Check every assumption against your own. The interchange-plus column is a blended estimate that moves with your card mix; the software and ancillary figures are placeholders. This is not a promise that switching saves $510 a month. It shows that the same restaurant, same volume, same cards, can pay a full point of spread depending on structure — and that the $0-software column is the most expensive on the sheet.

What a $510 gap buys, if you found one: at a fully loaded $16 an hour, about 32 labor hours a month, one part-time closing shift a week. Annualized it's a little over $6,100 — a used walk-in cooler, or a month added to your working capital reserve. That's the real comparison. Not "which system has better reports."

The sentence to say out loud

"What is my effective rate, all-in, at my ticket size and my monthly volume — not your advertised rate?"

Ask it exactly that way, to every rep, and get the answer in writing with the assumptions stated. You get numbers you can line up side by side, and you learn fast which reps do the math with you and which deflect to features. If a rep won't give you one all-in number against your stated volume, that's your answer about what the statement looks like in month three.

Bundled Processing vs. Bring Your Own Processor

The difference determines whether your rate is negotiable later or fixed forever.

Bundled / proprietary. The POS vendor is also your processor. Software pricing is subsidized by processing revenue, processing is not optional, and hardware is generally locked to that processor once bought. This describes most of the cloud POS market aimed at independent restaurants; Toast and Square are the commonly cited examples of vendors requiring their own processing with no bring-your-own option.

Open / BYOP-capable. Hardware and software pair with an independent processor you negotiate separately. This is where interchange-plus rate-shopping actually pays off, at the cost of one more vendor relationship and one more support number to call when something breaks at 7pm. Clover is the commonly cited example of hardware that can, in some configurations and depending on the reselling processor, pair with an independent processor. Source: Tech.co · Source: Restaurant Launchpad

Treat that as a pattern to verify with the specific vendor and reseller in front of you, not a fixed rule. Reseller agreements vary and change.

The practical consequence is what most comparison articles get wrong. A bundled system's software price is not comparable to an open system's in isolation. The processing subsidy has to be added back first — which is what the effective-rate calculation does, and why it's the only comparison that means anything.

The other thing to price before you buy hardware: proprietary terminals commonly cannot be repointed at a different processor. Sign a bundled deal, find a better rate two years in, and the switching cost isn't just the contract — it's the hardware, from zero. That sunk cost sits on the other side of every rate negotiation you might want to have.

Contract Length and the Cost of Getting Out

Legacy merchant-services agreements commonly run one to three years, with early termination fees either flat — often quoted in the $100–$600 range — or calculated as liquidated damages against the remaining term, which can run into the thousands on a mid-contract exit. Auto-renewal clauses that extend the term another full year absent written notice inside a 30-to-90-day window are common enough to assume they're in there until you've read otherwise. Source: Swipesum · Source: KORONA POS

Several current cloud POS vendors instead offer month-to-month software. Do not read that as month-to-month everything. The software agreement and the processing agreement are frequently separate documents with separate terms and separate end dates, and month-to-month software on top of a three-year processing contract is a three-year commitment wearing a friendly hat.

Three things to get answered before signing:

1. What is the term on the software, and what is the term on the processing?
2. Is there an early termination fee, and is it flat or a liquidated-damages calculation? Get the actual number for a hypothetical exit twelve months in.
3. Does either agreement auto-renew, and what is the notice window?

Put the notice window on a calendar the day you sign, with a reminder two weeks before it opens. The renewal date always arrives during a busy season, and the fee for forgetting is an extra year.

Then pull the lever: ask for month-to-month. First-time operators assume the paper is fixed. It frequently isn't, particularly if you'll trade a slightly higher rate for flexibility — which for an unproven concept is often the right trade.

Passing the Fee to the Guest

Surcharging and cash-discount or dual pricing get pitched hard as the answer to processing cost. Two sets of rules govern them, and you need both before you turn anything on.

Card network rules set a ceiling. Visa publishes a 3% cap on merchant credit card surcharges, requires the merchant to notify its acquiring bank at least 30 days before it starts, and requires disclosure to the customer. Source: Visa Caps and notice rules are set per network and have been revised more than once — confirm the current rule for every card brand you accept with your processor rather than assuming one number covers all of them. And a ceiling existing is not permission.

State law decides whether you can do it at all. Some states prohibit surcharging outright, some restrict it to no more than your actual cost of accepting the card, and at least one has moved against it through broader consumer-pricing legislation with a contested enforcement history. Any list published today may be wrong by the time you read it. Ask your processor whether surcharging or dual pricing is compliant where you operate, get the answer in writing, and remember the network cap applies on top of whatever your state allows.

Two adjacent setup items, neither a vendor-selection criterion. Sales tax and any local meals or beverage tax has to be configured correctly in the POS at go-live — a question for your bookkeeper and your state revenue department. And if you serve alcohol, any age-verification or ID-scan requirement at the point of sale attaches to your liquor license, not your POS vendor; see the permits and licenses checklist.

Integrations That Actually Matter

Your POS is the record of every sale. Anything downstream that needs sale-level data either integrates cleanly or becomes a spreadsheet somebody re-keys at 11pm, badly, on the days they remember. Four matter, and for each, one question separates a real integration from a logo on a partner page.

Online ordering and delivery aggregation are different problems, and vendors blur them deliberately. Native online ordering — your own site taking your own orders — is a separate decision with its own economics, covered in commission math and platform criteria for online ordering.

Aggregation is the POS question. Each marketplace ships you a tablet and each tablet needs a human watching it — the "tablet wall," and how orders get missed on a Friday. The fix is third-party order-aggregation middleware, consolidating every marketplace feed into one POS or KDS routing point. Otter, Chowly, ItsaCheckmate, and Deliverect are examples of the category, not a ranking. Source: Tolodora · Source: BusinessWire

A KDS is only worth what the physical pass can absorb. Routing tickets to a screen doesn't help if bagging and plating fight over the same eighteen inches of counter, so a dedicated expo and packing station is the other half of that integration, and the cheaper half.

Offline Mode Is a Spectrum, Not a Checkbox

A POS that goes dark stops taking money with a line of people in front of it. Every vendor says they work offline; what they mean varies enormously. Order entry, ticket printing, and KDS routing generally continue. Card authorization and anything that reaches a server — loyalty balances, gift cards, house accounts — depend on connectivity.

Two vendor-documented examples, illustrating how wide the spectrum is rather than comparing the two:

• Square's offline payments mode gives a merchant 24 hours to reconnect and sync card payments taken while offline, within a 72-hour outer window. A payment can still decline on reconnection, and the merchant bears that loss. Source: Square
• Toast devices enter offline mode roughly 40 seconds after losing connectivity. Order taking, card and cash payments, and ticket and KDS routing continue; gift cards, loyalty redemption, text-to-pay, comp cards, and house accounts don't work until reconnection. A local sync mode lets Toast devices on the same local network keep talking to each other through an internet or cloud outage. Source: Toast · Source: Toast platform guide

Two respectable approaches, materially different failure modes. One holds the charge and puts the decline risk on you inside a defined window; the other degrades to a named feature list and keeps terminals talking locally. Ask every rep this, verbatim, and write down the answer:

"What exactly still works if the internet drops for an hour, and what happens to a card charge taken during that window if it later declines?"

For a restaurant with dependable wired internet and a backup, that's a good question. For a food truck, it's the whole decision.

Buying Hardware vs. Leasing It

Buy outright. Higher upfront cost, no ongoing payment, some resale value — though per the lock-in point above, resale really means "sell it to another user of the same system," a thin market.

Lease, or bundle hardware into the subscription. Lower upfront cost, refreshes and failures are the vendor's problem, and you preserve cash in the months you have the least of it. The tradeoff: you pay indefinitely, and total cost over three-plus years commonly exceeds the purchase price.

The decision variable isn't money — it's confidence. How sure are you this is the last POS you'll choose for a while?

Neither answer is wrong. Buying hardware you'll abandon in fourteen months is.

How the Calculus Changes by Track

The cost structure and the evaluation questions are identical across all three. What changes is which feature you're willing to overpay for.

Brick-and-mortar restaurant

Everything above applies unmodified. Cost and complexity concentrate in multi-terminal builds — bar station, host stand, patio handhelds, two KDS screens — and each added terminal may carry its own software fee depending on how the vendor licenses. Get per-terminal pricing spelled out before you count stations, and confirm how tickets route between bar and kitchen. That routing logic is where these builds go wrong.

Food truck: offline mode and cell signal are the whole decision

Weight connectivity above price. Not alongside it — above it.

A truck has no fixed internet. It runs on cellular, and cellular is what fails at the festivals, remote lots, and dense event crowds where you make your best days — ten thousand phones on one tower is a dead zone regardless of your carrier's coverage map. Source: Food Truck Operator Ask the offline question first and let it eliminate vendors before you compare rates.

• A cellular connection built into the POS hardware beats tethering to your personal phone. That phone is also your dispatch line, your maps, and your ordering phone — don't make it your payment infrastructure too.
• A standalone mobile hotspot, commonly about $30–$80/month, is the backup rather than the primary. Source: Addmi
• Budget connectivity as an operating cost from day one. It's small, recurring, and routinely forgotten — as is the fact that your POS draws power from the same system as everything else on board. Food truck equipment and power planning covers the generator and load side.

Keep one distinction clean: getting a signal at a location and being permitted to sell there are separate problems. Coverage is infrastructure. Vending zones, event permits, and site approvals are licensing, with jurisdiction-specific answers. Don't let a rep who solves one imply he's solved the other.

Ghost kitchen: delivery integration is the whole decision

Most of the hardware section doesn't apply. No dine-in terminals, no bar, usually no host stand. One question replaces all of it: how do orders from four marketplaces get into one screen without a human babysitting four tablets?

Aggregation isn't a nice-to-have here — it's close to the entire POS decision. A system with deep dine-in coverage and weak aggregation is the wrong system at any price, and it's the system most reps are trained to demo. Push to aggregation in the first five minutes: native or middleware, what it costs, which marketplaces are supported today.

The physical companion is staging. Aggregated orders sit somewhere between bag-up and courier arrival, and a shelf by the door is how couriers grab the wrong bag — pickup shelving and heated holding makes the routing you just bought work. Ghost kitchen setup and delivery economics covers licensing structure and commission math; the equipment side of a ghost kitchen build covers what to buy.

A Framework, Not a Recommendation

Three broad categories you'll encounter while shopping:

• Cloud POS bundled with proprietary processing — most of the market aimed at independent restaurants. Simplest to buy, single support number, processing not negotiable.
• BYOP-capable systems that separate hardware and software from processing. More work to assemble, and the only place interchange-plus rate-shopping actually pays.
• Legacy and on-premise systems, still common in some regions and segments. Least flexible on modern integrations, which matters most if delivery or online ordering is part of your model.

Which is right depends on your volume, ticket, track, and tolerance for an extra vendor. So instead of a ranked list, here's the sheet. Run every vendor down it and keep the answers.

That last row is the one operators skip and regret. Ask whether you can export your own historical sales data, in what format, and whether it's self-serve. Two years of item-level sales history is the raw material for every menu and pricing decision you'll make later. A system that won't hand it back is charging you a cost that never appears on the quote.

Frequently Asked Questions

How much does a POS system really cost for a restaurant?

Budget in five parts, not one. Hardware runs roughly $700–$2,000 for a single counter terminal and $2,000–$4,000+ for a multi-terminal build with kitchen display screens; handheld readers go as low as $10–$100 each. Software runs $0 to $300–$400+ a month by tier and terminal count. Processing is usually the largest line by a wide margin — on $50,000 in monthly card sales, a 1.5% to 3.5% effective rate is $750 to $1,750 every month. Then ancillary fees (PCI, statement, chargeback, gateway) and a one-time implementation or menu-build fee that is often waivable if you ask.

What is an effective payment processing rate and how do I calculate mine?

Effective rate is total processing fees paid in a period divided by total card sales in that period. Total fees means everything on the statement — percentage, per-transaction fee, PCI, batch, statement, gateway — not the headline rate you were quoted. A processor advertising 2.49% can easily deliver an effective rate above 3% once fixed fees are counted, especially on small tickets. Ask every vendor for one all-in effective rate against your own expected ticket size and monthly volume, in writing, with the assumptions listed.

Is interchange-plus or flat-rate credit card processing cheaper for a restaurant?

Flat-rate is simpler and easier to forecast; interchange-plus is usually cheaper once card volume is meaningful. Card-present interchange-plus commonly lands around 1.5–2.5% plus $0.10–$0.15 per transaction, against flat rates commonly quoted at 2.6–2.9% plus a fixed fee. Average ticket matters as much as the percentage — a $0.15 fixed fee is 1.25% of a $12 ticket but 0.25% of a $60 one — so a coffee counter and a steakhouse should negotiate different parts of the same quote. Run your own numbers rather than trusting either model's reputation.

Do ghost kitchens need a different kind of POS system?

They need one evaluated on different criteria. With no dine-in terminals, no bar, and no host stand, most multi-terminal hardware considerations disappear and delivery-marketplace integration becomes close to the entire decision. The risk to solve is the tablet wall — a separate tablet per platform, each needing a person to watch it, which is how orders get missed during a rush. Ask whether the POS aggregates marketplace orders natively into one ticket feed and KDS or requires third-party middleware, and what that middleware costs monthly. A system with excellent dine-in features and weak aggregation is the wrong choice here at any price.

What POS system works best for a food truck with bad cell service?

Weight offline capability above price, and evaluate the specifics rather than the checkbox. Ask what continues to work when connectivity drops and what happens to a card charge taken during the outage if it later declines — vendors differ sharply. Square's offline mode gives you 24 hours to reconnect within a 72-hour outer window, and a held payment can still decline with the merchant bearing the loss. Toast devices enter offline mode about 40 seconds after losing connection, keep taking cards and routing tickets, and lose gift cards, loyalty, and text-to-pay until reconnection. A cellular connection built into the POS hardware beats tethering your phone; a hotspot at roughly $30–$80 a month is the backup.

Should I buy or lease my POS hardware?

It depends less on money than on confidence. Buying costs more upfront and nothing ongoing, and makes sense when you're reasonably sure this is the system you'll run for several years — established concept, known volume, a multi-terminal build you expect to keep. Leasing or bundling hardware into a subscription preserves cash, hands refreshes and failures to the vendor, and caps your loss if you switch in year two, a real possibility for a first location or a truck. The tradeoff: you pay indefinitely, and total cost over three-plus years commonly exceeds the purchase price. Weigh resale honestly — proprietary hardware usually can't be repointed at a new processor, so its resale market is other users of the same system.