Nobody has to do the math.
Your guests scan the card on the table, land in a shared ten-minute lobby, and split the check however they like, evenly or line by line. Everyone pays their own way with Apple Pay. The server never walks a card, and you get the table back sooner.
Sits on top of Toast and Square. Your POS does not change.
Ready to scan the code on the table.
Point your camera at the little card on the table.
8m 24s
faster table turns when the guest pays, not the server.
Somebody already measured this properly.
In 2016 two Cornell researchers watched what happened when diners settled up from the table instead of waiting for a server. They observed one restaurant by hand, then checked themselves against 243,301 transactions across 45 more. Turns got shorter, servers got time back, and the effect held at scale.
Read the study, and everything else we cite We lead with somebody else’s peer-reviewed study because we do not have our own numbers yet. When we do, they will be measured against each restaurant’s own baseline and published the same way.- 15.5%
- shorter table turn, measured across 46 restaurants
- 36%
- less server time spent on every single check
- $375,000
- of demand a $2.5M dining room never got to serve
Minutes are the only thing you can’t buy more of.
A restaurant has a fixed number of tables and a fixed number of hours. The only way to serve more people is to give each table back sooner. Move the sliders to your own room. The arithmetic is shown, not asserted.
Your room
Seat to seat, including settling up.
Cornell measured 8m 24s. We default to 5.
A freed table only earns anything if there is someone to seat at it. Slide to 100% if there is always a queue, and to 0% if you rarely have a wait.
Projected
More seating capacity
+5.88%
Pure geometry of your turn time. It holds whether or not anyone is waiting for the table.
Revenue assumes you can seat 70% of the tables you free up. How this is calculated
Where the minutes come from. Cornell’s researchers measured a 8m 24s drop in table-turn time across 243,301 transactions when guests paid at the table themselves. We default this calculator to five.
What we measured against
Cornell’s researchers observed 8m 24s faster table turns and 36% less server time per check when guests paid at the table themselves. 243,301 transactions, 45 restaurants, peer-reviewed.
How the calculator works
Turns per table are service minutes divided by turn time. We take the difference before and after, multiply by your table count, then scale by the share of freed tables you say you can actually fill. Months use 52 ÷ 12 weeks, not a flat four.
Why the numbers stay modest
We default to five minutes saved when the measured figure is eight and a half, and to a 90-minute turn when the study observed 54. Both choices push the result down. Cornell’s own worked example puts $375,000 of unmet demand on a $2.5M restaurant.
The case against us.
Every company in this category publishes a page of glowing numbers. Here are the arguments that actually get made against a product like this, and what we honestly think of each.
It’s a real one, and we won’t pretend otherwise. Toast ships Mobile Order & Pay and reports that restaurants adding it see a 10–12% lift in processing volume. If all you want is a guest paying their own check, Toast can already do that, and it’s bundled.
What Toast doesn’t do is the part that’s actually hard: a table of five arguing about who had the second bottle. Their flow is built around one person paying one check. Ours is built around the group: a shared lobby, a headcount the bill respects, and itemised claiming that settles to the penny.
So the honest framing is: if your tables are mostly deuces, use Toast’s and save your money. If you’re doing six-tops on a Friday and your servers are splitting checks four ways by hand, that’s the problem we exist for.
And Toast isn’t the only one here. DoorDash already ships a Toast-integrated tableside order-and-pay product . That is the old Bbot, still actively sold, with DoorDash’s distribution behind it.
The risk we’re carrying: Toast could build group splitting properly tomorrow, and we’d have to be meaningfully better at it than a company with a thousand engineers. That’s the bet, and it’s ours to lose, not yours.
Then the check splits two ways, and those two people pay more than their share. This is the sharpest attack on the lobby idea and it deserves a straight answer.
Three things blunt it. The lobby shows a live headcount, so you can see it says “2 at the table” when there are obviously four of you, so the error is visible before anyone pays, not after. Second, closing the lobby is deliberate; nobody pays until someone taps the button. Third, and most importantly, the server can still close the check the normal way at any point. billy is a faster path, never the only path.
What we can’t fix is somebody who deliberately doesn’t scan so their friends absorb their share. That’s a friendship problem. It happens with cash too.
Agreed. Payment is one segment of a turn, and if your tickets are slow, or you’re short a busser, or guests linger over coffee regardless, saving four minutes at the end changes very little.
More fundamentally: freed capacity is only worth money if somebody is waiting to use it. The Cornell researchers who measured this were explicit about it: the gain assumes “a demand for the additional seats”.
Which is why our calculator has a slider for exactly that, why it doesn’t default to 100%, and why we only sell to high-volume rooms with a wait. If you have open tables at eight on a Saturday, we are not your problem worth solving.
No. There is no app and no account. The card on the table opens a web page in whatever browser is already on the phone, and it closes when they are done. Nothing to install, nothing to remember a password for, nothing left on the phone afterwards.
That is not us being generous, it is the only version that works. A table of five is not going to install anything to pay for dinner, and every extra step between sitting down and settling up is a table that takes longer to turn.
They are, and the data backs them up: only 31% of consumers view QR-code menus positively, and one restaurant group measured a 10% drop in check averages after switching to them. Restaurants were right to pull them.
But look closely at what people actually hate. Reading a menu on a phone is worse than reading it on paper: you pinch, you scroll, you lose your place, and you can’t look at it and talk at the same time. It replaces something pleasant with something worse.
Paying is the opposite: a task with an ending. And the numbers separate cleanly: 60–80% of Gen Z, millennial and Gen X diners say they’re willing to pay tableside by phone. So we deliberately don’t touch the menu, the ordering, or anything a guest enjoys about being waited on. We only take the part everyone already dislikes.
Worth saying plainly: for older guests that willingness drops to roughly a quarter. Which is exactly why nobody is ever required to scan.
They could. It’s a known attack, and both the FBI and FTC have warned about tampered codes in public places. Any product in this space that tells you otherwise is lying to you.
What we do about it: the code on your table is bound to that table and signed, the session it opens is short-lived, and it always resolves to our own domain, so a substituted sticker fails a check rather than quietly taking a payment. The printed cards are also designed to be hard to overlay cleanly, and we tell you what to look for.
It reduces the risk. It doesn’t eliminate it. Nothing does.
We don’t know yet, and we’re not going to quote you a number we can’t stand behind. Competitors claim tips go up by around 10%. Plenty of servers will tell you that taking them out of the payment moment does the opposite.
What we can say: the tip screen defaults to your existing suggested percentages, and the current US full-service average is 19.3%, so we know what “no worse” has to look like. Measuring this honestly in our first pilots, against each restaurant’s own baseline, is the single most important thing on our roadmap. If tips drop, the servers will know before we do, and they’ll kill it. Fairly.
It has, repeatedly, and we’d rather you heard this from us than found it out later.
Presto Automation shipped over 277,000 pay-at-table tablets to Applebee’s, Chili’s and Red Lobster, went public, and ran a negative gross margin on $14.2M of revenue. It accumulated a $266M deficit, wound the pay-at-table business down entirely, and was delisted. The SEC later charged it over claiming its drive-thru “AI” was automated when the orders were largely being typed by people in the Philippines.
sunday, the best-funded company in this category, went from a $100M Series A to a $21M Series B four years later, cut headcount from around 400 to just over 100, exited four of seven markets, and after roughly $146M raised still describes itself as only “very near” profitability. Mr Yum and me&u raised about $165M between them and had to merge to survive. Tabbedout raised $29M, told the Wall Street Journal an IPO was next, and now exists as a product for pop-up bars.
Our read: nobody has proven this as a standalone business in the US. The successful outcomes were all sales to strategics: Bbot to DoorDash, Rooam to American Express, MyCheck to Shiji. That is the honest shape of this category, and it’s the risk we’re taking on knowingly.
Then we have a serious problem, and it’s worth being precise about how exposed we are. Toast’s self-serve API access is read-only. Writing to checks and payments, which is the entire product, requires approved Partner status: compliance, privacy, security and legal review, a negotiated agreement, certification, a single-restaurant alpha, then a beta. Toast states plainly that it “is not able to integrate with all interested integration partners.”
Toast also publishes a list of deprecated API functions, which is a polite way of saying the ground we build on moves.
What we do about it: build for Square as well as Toast so no single platform is the whole company, and keep the guest-facing experience ours rather than borrowed. What we can’t do is pretend the dependency isn’t real. If you’re evaluating us, that risk is partly yours too, and you should price it in.
From a peer-reviewed Cornell Hospitality study, mostly, and every figure on this site links to its source.
Worth knowing what we didn’t use. The statistic everyone in this category repeats, that guests wait ten to twelve minutes for the check, has no primary source we could find. We went looking for the study behind it and came up empty, so it appears nowhere on this page.
One competitor publishes five different “minutes saved” figures across its own website : 7, 10, 12, 13 and 15. When a headline metric moves by a factor of two depending on which page you land on, it’s marketing, not measurement.
What changes, and what deliberately doesn’t.
Stays the same
- Cash still works, exactly as it does now.
- A server can close any check the old way, mid-session.
- Your POS stays your POS. billy doesn't replace anything.
- Tip percentages stay whatever you already have set.
Goes away
- No walking a card to the terminal and back.
- No splitting a check four ways by hand at the station.
- No table waiting on you to notice they're ready.
- No "can we get separate checks?" after the fact.
We’re picking our first few rooms.
We’re looking for busy full-service restaurants with a real wait on weekends and tables that actually split. That’s where the arithmetic works. Pilots are free, and we’ll measure against your own baseline so the result is yours to keep either way.