Framing and tipping behavior: A field experiment

We are conducting a field experiment with a partner firm handling electronic payments for several German restaurants. Customers are guided through an electronic restaurant bill with an interactive tipping feature at the bottom (displaying default percentage options for tipping). Restaurant customers are randomly assigned to one of five experimental conditions (between-subject design), which display differences in the display of the tipping options intended to encourage higher tipping percentages. Details on the experimental conditions are contained in the hidden part of this section.

There are five experimental conditions (1 baseline, 4 treatment). The 4 treatment conditions each alter one aspect of the baseline condition but are otherwise identical.

Baseline condition: When customers change the tipping percentage, smileys are displayed on the screen. The higher the chosen tipping percentage, the more enthusiastic the displayed smiley. This is the baseline condition, as the company had already implemented it before the field experiment. Displayed tipping percentages are 5%, 10%, 15%, and 20%. However, customers can also give zero tips or set their own amount. 10% is preselected as the default.
Treatment condition I: Same as baseline, only now smileys are always shown adjacent to the tipping amounts.
Treatment condition II: Same as baseline, only without the smileys. In the evaluation phase, we will likely treat this as the control group as there are no visual / emotional framing elements in this condition.
Treatment condition III: Same as baseline, only now the presentation of the smileys is accompanied by a vibration.
Treatment condition IV: Same as baseline, but no smileys appear. Instead, the color of the tip amount frames becomes friendlier (i.e., deeper green) the higher the tipping amount.

2025-03-31

2025-07-31

Binary variable indicating whether a tip has been given; tip in percent; absolute tip

Subject to availability: customer satisfaction; retention; tipping behavior upon retention

Based on historical data, our partner firm handles approximately 1,000 transactions per month. To provide us with sufficient statistical power for our main analyses, as well as heterogeneity analyses to uncover potential mechanisms, our data collection will last from April through July, 2025. Each customer is randomized into one of the five treatment conditions. Repeat customers will thus be in the same experimental group for all transactions. Aside from the framing interventions to the default tipping options, the payment flow each customer follows is identical across groups.

Note that our partner firm has notified us that they might have to stop data collection for our experiment early due to operational reasons.

Randomization done in automated manner

Individual

No

-

Several thousand observations (all handled transactions from April through July 2025). During the same timeframe in 2024, the partner firm handled around 5,200 transactions via their app. We expect similar numbers during our investigation period.

Around 1,000 observations per treatment group.