Dinner with friends is bigger before the bill even lands

Put a few friends around a table and the meal grows. In the diary study that measured it, meals eaten in large groups were over 75% larger than meals the same people ate alone. Same person, bigger dinner.

We've already written about how splitting the bill evenly pushes people to order about a third more. That effect needs a split rule to exist before it can do anything. The studies below don't. They're about what the table does to the bill before anyone has said a word about dividing it.

A week of food diaries

In 1992, John de Castro and a co-author paid 153 adults to write down everything they ate for seven days. No lab setup, just over 3,800 ordinary meals recorded as they happened. The pattern was clean enough to become the paper's title: the amount eaten in meals is a power function of the number of people present. More company meant bigger meals, and in large groups they came out over 75% larger than the ones the same diarists ate alone.

Company didn't just add food. Meals with other people ran longer, and the diaries caught people eating faster too.

You mirror your table

Part of the mechanism is plain mirroring. A 2015 meta-analysis in Social Influence pooled 38 studies of what researchers call modeling: sit someone next to a companion who eats a lot and they eat more, next to a light eater and they eat less. The pooled effect was r = .39, large by social psychology standards. The pull is actually stronger in the shrinking direction, so a restrained table can trim your meal. At a birthday dinner where everyone's ordering big, though, you can guess which way the current runs.

The part that gets me is what this does to the receipt. If nobody's order was fully independent, the bill is something the table produced together, a group decision that no single person made.

Saying your order out loud changes it

Dan Ariely and Jonathan Levav tested a subtler force at the Carolina Brewery, where they posed as waiters and offered free beer samples to 100 tables. Half the tables ordered aloud, one person after another. The other half wrote their picks down privately.

The aloud tables chose more variety and enjoyed it less. They rated their beer 6.80 out of 10 against 7.75 for the private tables, and 38% of them regretted their choice versus 12%. Only one seat escaped: whoever ordered first was as satisfied as the private orderers, and everyone after them was, on average, less so. The authors' read is that once your order is public, you're partly picking to seem distinctive instead of just picking what you want. Multiply that around a full table and the receipt drifts away from what anyone would have chosen alone.

The drinks line keeps count

Drinks scale with company too. A 2015 study in Addiction had 183 Swiss young adults answer hourly surveys on their phones during real nights out, Thursday through Saturday between 8pm and midnight. Across 1,441 evenings and 7,205 check-ins, the more friends present in a given hour, the more drinks people had in that hour, a small but statistically reliable bump per additional friend that ran stronger for men. This is associational data rather than a controlled experiment, but it was measured live instead of through next-day recall, which is where drinking studies usually go soft.

Even the tip feels the headcount

This one is older. In 1975, four social psychologists, Bibb Latané of bystander-effect fame among them, recorded tips from 396 real restaurant parties. Tip percentage fell as a power function of party size, from around 18% for a solo diner to about 12% for a party of six. Their explanation was diffusion of responsibility: with more people at the table, the tip becomes everybody's job and therefore nobody's. The decline itself is solid field data. The explanation has been argued over ever since, so treat it as the authors' interpretation rather than settled science.

The pattern got a modern echo. A 2025 paper in the Journal of Behavioral and Experimental Economics went through more than 800,000 card transactions from a Norwegian pizza chain and found that whether a table tips at all rises and then falls with party size, an inverse U shaped by the bill and by how much alcohol was ordered. The curve is a different shape and it tracks whether a table tips rather than how much, so it echoes the old result more than it confirms it. Fifty years on, though, headcount still reaches the tip line.

Where dvup comes in

Add it up and it's a little humbling. The headcount grows the meal, your neighbors' orders tug at yours, more friends show up as more drinks, and the crowd thins out the tip. All of it happens before anyone picks a split rule, which means the split can't prevent any of it. It's simply the only part you control. And I wouldn't want to prevent it anyway; meals with friends are bigger because meals with friends are better.

So dvup makes the controllable part fit the table in front of you. For the regular crew with similar orders, split evenly in two taps and let repetition keep it fair. For the big or lopsided table, scan the receipt and have everyone claim what they actually had, so the rounds land on the people who drank them and the tip gets divided into named per-person shares instead of sitting on the bill as a blob nobody owns. Splitting, balances and settling up are free with no cap on expenses. Receipt scanning gives you five free scans, then a short ad earns another, about one a day, and dvup+ removes the ads and the waiting.

Questions