Why We Order the Most Popular Dish
What a restaurant field trial, a vendor website experiment and seven consumer studies say about popularity labels, and the diners they help most.
Answer to the question
Do "most popular" labels change what we order?
Yes, for some diners. In a randomised restaurant trial, showing the ranking of the five most popular dishes raised demand for those dishes by 13 to 20 per cent. Other studies find that popularity signs help most where they carry information and can add complexity for consumers focused on avoiding a poor choice.
Key takeaways
- In a randomised natural field experiment in a restaurant, giving diners the ranking of the five most popular dishes raised demand for those dishes by 13 to 20 per cent (Cai, Chen and Fang, 2009).
- The same trial found no significant saliency effect, so the lift is better explained by the information in the ranking than by prominence alone.
- In a field experiment on a vendor website, narrow-appeal vendors received more visits than equally popular broad-appeal vendors once popularity information appeared (Tucker and Zhang, 2011).
- Across seven studies, popularity signs made decisions more complex for consumers with a prevention focus, widening the options considered and lowering choice commitment (Ghiassaleh, Kocher and Czellar, 2020).
- Robert Cialdini's Influence holds that people lean on the behaviour of others most when they are unsure, which describes the first-time diner.
A menu of forty dishes asks a first-time guest to choose with almost nothing to go on, and a most popular tag fills the gap with other people's experience. The tag says that many diners have already tried the dish and kept ordering it. Whether that changes what reaches the table has more than one honest answer, and the research gives each of them a hearing.
How most popular labels steer a diner
Robert Cialdini's Influence argues that people read the behaviour of others as evidence of what is right to do, and that they lean on it hardest when they are unsure and when the others resemble them. A new diner is unsure by definition. The guests who ordered before them are alike in the way that matters, since they too came for dinner. A most popular label turns that crowd into a single line of text.
The shortcut is cheap. Reading forty descriptions takes effort, and a most popular tag lets a guest hand part of the search to everyone who ordered before. Most of the time that is a sensible trade, because a dish that keeps being reordered is often reordered for good reasons. The shortcut fits less well when the signal is thin, when the guest wants something other than what the crowd wants, or when the dish is chosen for an occasion the crowd was not thinking of.
Catherine Tucker and Juanjuan Zhang tested a version of this signal in a field experiment on a website that lists wedding service vendors. Popularity information is usually expected to push customers toward mainstream choices with broad appeal. Their hypothesis ran the other way. The same level of popularity implies higher quality for a vendor with narrow appeal than for one with broad appeal, so the signal should help the niche vendor more, as a most popular tag might help a dish with a small following. After popularity information was introduced, narrow-appeal vendors received more visits than equally popular broad-appeal vendors.
For a menu, that points to a lift that depends on what the number is read against. A dish that is popular and also unusual may gain more from a most popular tag than a dish that everyone expects to be popular. That is a reading of the experiment, not something it measured on dishes. The wider field evidence on social proof, from hotel towel cards to music downloads, sits in What is social proof in marketing, and it too turns on who the others seem to be. Where the crowd's choices are visible, outcomes can also split: Why the Best Song Does Not Win describes how identical songs met different fates in parallel music markets.
When does a label stop helping?
Arezou Ghiassaleh, Bruno Kocher and Sandor Czellar studied popularity signs such as best seller and top rated across seven studies. The signs did not work equally for everyone. For consumers with a prevention focus, meaning a concern with avoiding a poor outcome and not with reaching a good one, the signs made the decision more complex. They widened the set of options the person considered, raised uncertainty about the choice and lowered commitment to it. The authors draw the implication that marketers could target such signs more carefully.
The two sets of findings describe different diners. A label settles the doubt of someone who wants a guide, and it adds one more thing to weigh for someone already counting the risks of each dish. Three questions follow from the papers and give a practical way to judge how far a most popular label is likely to move an order:
- Is the guest new to the room, so that other people's choices are real information?
- Does the tag say something about quality for this particular dish, as popularity does for a narrow-appeal vendor?
- Is the guest mainly trying to avoid a bad choice, which can turn the tag into another option to compare?
These questions are our synthesis of the two papers and were not tested as a set. They fit a wider pattern in which small changes to a menu shift choice without changing a dish. The Compromise Effect on the Wine List shows the same thing from the price side: a dear bottle added at the top changes what sells beside it.
What did the restaurant trial measure?
Giving diners the ranking of the five most popular dishes raised demand for those dishes by 13 to 20 per cent. Hongbin Cai, Yuyu Chen and Hanming Fang ran this as a randomised natural field experiment in a restaurant, and they designed it to separate two explanations. In observational learning, the guest infers quality from what others chose. In a saliency effect, any dish placed in front of the guest sells better because it is easy to see.
The trial found no significant saliency effect. That leaves the information in the ranking, and not mere prominence, as the likelier source of the lift. The researchers also found modest evidence that the effect was stronger among infrequent customers, which fits the idea that a label helps most the people with the least to go on. This is the same pattern the first-time guest in the opening paragraph would predict.
For a kitchen, the practical reading is modest. A most popular tag that rests on what guests have actually ordered carries real information, and the trial's design used exactly that. A tag that is applied to every dish, or to none that guests chose, says nothing to a newcomer and is unlikely to do the same work. The 13 to 20 per cent figure belongs to one restaurant and one ranking, and it describes demand for the listed dishes, not the whole bill.
Both sides of the question meet in this trial. The label lifted orders where it carried information, and the experiment tested a ranking of real choices and not a bare tag. Dining satisfaction was higher when guests were shown the five most popular dishes, and it was not higher when they were shown only the names of some sample dishes.
Common questions
Do "most popular" labels change what we order?
In the one randomised restaurant trial reviewed here, yes. Showing diners the ranking of the five most popular dishes raised demand for those dishes by 13 to 20 per cent.
Do popularity labels work for everyone?
No. Across seven studies, popularity signs made decisions more complex for consumers with a prevention focus, widening the options they considered and lowering commitment to their choice.
Does a label work only because it draws the eye?
In the restaurant trial, no significant saliency effect was found, so the lift is better explained by the information in the ranking than by prominence alone.
Can a popularity label help an unusual dish?
A field experiment on a website of wedding service vendors found narrow-appeal vendors received more visits than equally popular broad-appeal vendors once popularity information appeared. That experiment did not involve dishes.
Sources
Cialdini, R. Influence. Seed book; vault hub Books/Influence.
Cai, H., Chen, Y., & Fang, H. (2009). Observational learning: Evidence from a randomized natural field experiment. American Economic Review, 99(3), 864-882. doi:10.1257/aer.99.3.864 https://doi.org/10.1257/aer.99.3.864
Tucker, C., & Zhang, J. (2011). How does popularity information affect choices? A field experiment. Management Science, 57(5), 828-842. doi:10.1287/mnsc.1110.1312 https://doi.org/10.1287/mnsc.1110.1312
Ghiassaleh, A., Kocher, B., & Czellar, S. (2020). Best seller!? Unintended negative consequences of popularity signs on consumer choice behavior. International Journal of Research in Marketing, 37(4), 805-820. doi:10.1016/j.ijresmar.2020.04.003 https://doi.org/10.1016/j.ijresmar.2020.04.003