Food Hall Splits: When Everyone Orders from Different Vendors

Four friends. Four vendors. Four receipts. One person bought a round of drinks from the bar. Now everyone's staring at a pile of paper, trying to figure out who owes what. This is the hardest split in casual dining.

The multi-vendor problem

Food halls have exploded. Cushman & Wakefield tracked over 450 food halls operating in the United States as of 2023, up from just 70 in 2015. The format appeals to groups because everyone can get exactly what they want: tacos, ramen, pizza, poke. No more compromising on a single restaurant.

But this flexibility creates a splitting nightmare. At a traditional restaurant, there’s one receipt. One total. One tip line. At a food hall, there are as many transactions as there are vendors visited—plus whatever was purchased at the central bar or coffee stand. The bill isn’t a bill. It’s a collection of independent transactions with no obvious way to reconcile them.

 450+food halls now operating in the United States. Up from 70 in 2015.
A 543% increase in under a decade. 

The National Restaurant Association’s 2024 industry report identified food halls as one of the fastest-growing segments in dining, with 72% of consumers saying they’d visited a food hall in the past year. The appeal is obvious: variety, speed, and the ability to accommodate different dietary preferences. The math problem that follows? Nobody’s favorite part of the experience.

Sources: Cushman & Wakefield, “Food Halls of America” (2023); National Restaurant Association, “State of the Restaurant Industry” (2024).

A typical scenario

Here’s the setup. Four friends—Alex, Jordan, Sam, and Taylor—meet at a food hall on a Saturday afternoon.

AlexTaco stand$16.50

JordanRamen counter$19.00

SamPizza window$14.00

TaylorPoke bowl$18.50

Each person paid their own vendor. Easy so far. But then Jordan grabbed a round of beers from the bar: $36 for four drinks. And Sam bought a basket of loaded fries from the appetizer window that everyone shared: $12. Now the math gets complicated.

The communal purchases

Round of beers (Jordan paid)$36.00

Shared fries (Sam paid)$12.00

Communal total$48.00

Per person share$12.00

Jordan is owed $27 ($36 minus their $9 share). Sam is owed $9 ($12 minus their $3 share). Alex and Taylor each owe $12 to the pool. But who pays whom? And what about the tips at each vendor—did everyone tip the same amount? What about tax, which varies by vendor depending on what was purchased?

This is where the mental math starts to break down.

Why your brain gives up

In 1956, cognitive psychologist George A. Miller published one of the most cited papers in psychology: “The Magical Number Seven, Plus or Minus Two.” His finding was simple but profound: human working memory can hold approximately 7 items (give or take 2) at once. Try to juggle more, and information starts falling out.

Now count the variables in a four-person food hall split:

4Individual orders

4Vendor-specific taxes

4Tips (if any)

2Shared items

4Per-person shares of shared items

18+Total variables

That’s 18 or more variables—far beyond Miller’s limit. Your brain isn’t built to track this. So what happens? You simplify. You round. You say “let’s just call it even.” And that’s where unfairness creeps in.

“There is no magic here—memory span is simply the maximum number of items that can be held in short-term memory at once. Beyond that limit, errors increase exponentially.”
George A. Miller, Harvard University, 1956

John Sweller formalized this phenomenon in 1988 with cognitive load theory. When the cognitive demands of a task exceed working memory capacity, performance degrades. People make errors. They give up. Or they take shortcuts that sacrifice accuracy for simplicity. In bill splitting, that shortcut is usually “let’s just split it evenly”—the solution that feels fair but often isn’t.

Sources: Miller, “The Magical Number Seven, Plus or Minus Two,” Psychological Review (1956); Sweller, “Cognitive Load During Problem Solving,” Cognitive Science (1988).

Mental accounting at the food hall

Richard Thaler won the Nobel Prize in Economics partly for his work on mental accounting—the way people categorize money into different “accounts” based on subjective criteria. In a 1999 paper, Thaler showed that people don’t treat all dollars equally. A dollar spent on “food” feels different from a dollar spent on “drinks.”

At a food hall, this creates problems. Each vendor represents a separate mental account. The tacos feel like one purchase. The beer feels like another. The shared fries occupy some fuzzy middle ground. When it’s time to reconcile, people struggle to combine these accounts into a single coherent picture.

The mental accounting trap: You remember what you spent at “your” vendor, but you’ve already mentally closed that account. The communal purchases? Those feel like separate events. Combining them requires reopening accounts that your brain considers settled.

Baba Shiv and Alexander Fedorikhin demonstrated in 1999 that when cognitive load is high, people default to more automatic, emotionally-driven decisions. In food hall splitting, this manifests as: “I don’t want to think about this anymore, let’s just split it.” The person who ordered less absorbs the cost of the person who ordered more—because doing the actual math is too exhausting.

Uri Gneezy’s landmark 2004 study showed that equal splitting causes people to order 37% more than when they know they’ll pay individually. The food hall reverses this dynamic: everyone ordered individually, but now they’re being asked to act as if they split equally for the communal items. The modest orderer still gets penalized.

Sources: Thaler, “Mental Accounting Matters,” Journal of Behavioral Decision Making (1999); Shiv & Fedorikhin, “Heart and Mind in Conflict,” Journal of Consumer Research (1999); Gneezy et al., “The Inefficiency of Splitting the Bill,” The Economic Journal (2004).

How groups actually handle it

Without a system, groups default to one of four approaches—each with predictable problems.

Common

”Everyone pays their own”

Each person paid their vendor. Done. But what about the communal purchases? Someone’s out $36 for beers.

Simple for individual orders

Ignores shared items entirely

Common

”Split the communal evenly”

Add up shared items, divide by 4. Better, but doesn’t account for who paid what or who consumed more.

Handles shared items

The beer-drinker and the non-drinker pay the same

Rare

”One person covers, settle later”

Someone puts everything on their card. The group promises to Venmo. 44% of those promises never resolve.

Fast in the moment

Creates debt that often goes unpaid

Accurate

Track everything, reconcile

Log every purchase, note who paid, calculate net positions. Accurate but requires a system.

Fair outcome for everyone

Nobody wants to do this manually

The fourth approach is the only one that produces a fair outcome. It’s also the one nobody wants to do—because it requires tracking 18+ variables and performing multi-step arithmetic while your friends are ready to leave.

The practical solution

Here’s how to actually reconcile a food hall outing without mental math or awkward approximations.

Collect all the receipts

Before anyone leaves, gather every receipt—individual orders and communal purchases. If someone paid cash and has no receipt, note the amount.

Identify who paid what

Mark each receipt with the payer's name. Jordan paid for the beers. Sam paid for the fries. Everyone else paid their own vendor.

Assign shared items

For each communal purchase, note who participated. Did everyone drink the beers? Or just three people? The fries everyone shared equally? Mark it.

Calculate net positions

Each person's total = their individual order + their share of communal items. Compare to what they actually paid. The difference is what they owe or are owed.

Settle up

Transfer money to balance the books. Minimize transactions by having people who owe pay people who are owed directly.

This process is accurate but tedious. The more vendors involved, the more receipts to track, the more variables to juggle. A group of 6 at a food hall with 2 shared rounds of drinks? You’re looking at 30+ variables. That’s where technology becomes not just convenient but necessary.

The worked example

Let’s finish the math for our four friends. Here’s what each person actually spent versus what they should have paid:

What each person consumed

Alex: Tacos ($16.50) + share of communal ($12)$28.50

Jordan: Ramen ($19) + share of communal ($12)$31.00

Sam: Pizza ($14) + share of communal ($12)$26.00

Taylor: Poke ($18.50) + share of communal ($12)$30.50

Total consumed$116.00

PersonPaidConsumedOwes/(Owed)

Alex$16.50$28.50Owes $12.00

Jordan$55.00$31.00Owed $24.00

Sam$26.00$26.00Even

Taylor$18.50$30.50Owes $12.00

The settlement: Alex sends Jordan $12. Taylor sends Jordan $12. Sam’s square. Jordan gets their $24 back. Everyone paid exactly what they consumed.

The non-drinker scenario: What if Taylor doesn’t drink? Now the beer splits 3 ways instead of 4. Taylor’s share of communal drops from $12 to $3 (just the fries). Their total consumed becomes $21.50. They’re now owed $3.00 instead of owing $12.00. One variable change ripples through the entire calculation.

The tip and tax complication

We haven’t even touched tips and taxes yet. At a traditional restaurant, there’s one tax rate and one tip calculation. At a food hall, each vendor may have different policies.

8.875%NYC sales tax on prepared food

0%Tax on grocery items (some vendors)

VariesTip expectations by vendor type

Did the taco stand have a tip jar? Did the ramen counter have suggested tip percentages on the POS screen? Did the bar expect 20% while the coffee stand expected a dollar? These variations add more variables to an already overloaded calculation.

The honest answer: most groups don’t reconcile tips and taxes accurately. They round. They estimate. They say “close enough.” And that’s fine for casual outings where the differences are small. But for larger groups or more expensive outings, those small differences compound into real money.

How research shaped the design

Every finding about cognitive load and mental accounting maps to a specific design decision in splitty.

Working memory maxes out at 7 items→Scan receipts instead of typing—offload the data entry entirely

Multiple receipts multiply cognitive load→Scan each receipt separately—settle one bill at a time to keep the math simple

Shared items require explicit allocation→Tap to mark which items were shared and by whom

Net position math is error-prone→Calculate who owes whom automatically with one-tap settlement

High cognitive load triggers shortcut decisions→Show everyone’s total in under 30 seconds—before fatigue sets in

Four vendors. One tap.

Scan every receipt. Assign the shared items. Know exactly who owes whom—before anyone leaves the food hall.