Menu Psychology: What the Evidence Says About Dollar Signs, Decoys, Anchors and Layout

Menu advice travels fast: guests look at the upper right first, dollar signs make people spend less, a decoy dish sells the one next to it, an expensive item makes the rest look cheap. Before paying for a redesign, it helps to ask what was actually tested — with real orders or hypothetical choices, in one restaurant or several, and whether anyone repeated it. This review sorts the common tactics by that standard, using the original studies.

One caution about the literature itself. Several widely cited food-choice studies came from Brian Wansink's lab at Cornell, and in 2018 a university committee "found that Professor Wansink committed academic misconduct in his research and scholarship, including misreporting of research data, problematic statistical techniques, failure to properly document and preserve research results, and inappropriate authorship" (Cornell University, 2018). Treat any single study as a lead to test, not a rule.

The evidence at a glance

Dollar signs and price format

The source of the dollar-sign advice is a field experiment at St. Andrew's Cafe, a lunch restaurant at the Culinary Institute of America. Parties were randomly given menus with prices as "$20.00", "20." or "twenty dollars". The authors report that "guests given the numeral-only menu spent significantly more than those who received a menu with prices showing a dollar sign or those whose menus had prices written out in words" (Yang, Kimes & Sessarego, 2009). The overall test across the three formats was not significant, the difference appeared after controlling for party size, meal length, tip amount and table, and the authors cautioned that it may apply only to lunch at that restaurant. No replication was found. Dropping the symbol costs nothing; just do not budget for a lift.

Where guests look, and what they order

The "sweet spot" — the idea that eyes go first to the upper right of a two-page menu — did not hold up in the eye-tracking study that tested it on a two-page menu. Twenty-five people reading a two-page dinner menu looked first at the upper left and moved through it much like reading a book (Yang, 2012); the study did not measure sales, and the author notes that where people look may not decide what they order. A later eye-tracking study of 54 people reading single-page menus found areas that drew more attention on three of four layouts, but its abstract does not place them in the upper right, and no sales were measured (Kuo, Bavik & Ngan, 2020).

Position within a list has somewhat better support. In a Swedish university cafeteria randomized over 50 days and 7,968 lunches, moving the vegetarian dish to the top of a three-item board and the meat dish to the bottom cut the meat dish's share by about 5.6 points, from roughly 50% to 44% — while the vegetarian dish's own share did not rise significantly (Andersson & Nelander, 2021). An earlier café study found items at the start or end of a category were chosen more often than those in the middle, though only one category was significant on its own (Dayan & Bar-Hillel, 2011). If a dish earns well, keeping it near the top of its section is a free, reasonable bet; moving a dish to the bottom may cost it orders.

Decoys

A decoy is an option nobody should pick, placed to make a neighbor look better. The original 1982 study, using hypothetical choices by business students, found a decoy raised the target option's share by 9.2 points on average (Huber, Payne & Puto, 1982). A 2014 paper reporting 38 new experiments found the effect in most abstract comparisons of numbers but in none of the 27 where the options were experienced or shown as pictures, including a failed replication of a restaurant example, and concluded that "its practical validity should be questioned" (Frederick, Lee & Baskin, 2014). No test with real menu orders was found. Size tiers and add-on bundles should each make sense on their own price.

Prices ending in .99 or .95

The strongest field evidence for just-below prices comes from mail-order clothing catalogs, where a $9 ending raised demand in three experiments — in the third, only for items not also flagged "Sale" (Anderson & Simester, 2003). A preregistered meta-analysis of 69 studies and 40,541 participants found just-below prices increase purchase decisions slightly (g = 0.13), with effects that are "small and highly heterogenous" and bias corrections that suggest "smaller and, at times, nonsignificant true effects" (Troll et al., 2023). No restaurant sales study was found. Choose an ending style for consistency with the room, not for revenue.

Dish names

Descriptions have the best recent evidence, though for choice rather than revenue. Across five university dining halls, 185 days and 137,842 diner decisions, taste-focused vegetable labels "increased vegetable selection by 29% compared with health-focused labels and by 14% compared with basic labels," with larger effects where the recipes themselves were tastier (Turnwald et al., 2019). An earlier and widely quoted cafeteria study led by Brian Wansink, then at the University of Illinois, reported that descriptive names raised unit sales 27% over six weeks among about 140 customers, flagged as significant but without describing the test or the underlying sales counts (Wansink, Painter & van Ittersum, 2001). The practical reading: describe what the dish tastes like, accurately. It will not rescue a dish that is not good.

Expensive anchors

The idea that a high-priced item makes everything else look reasonable has no peer-reviewed test with real restaurant orders that this review found. The closest evidence is correlational: across 270 restaurants' wine lists, wine sales were "the same no matter how expensive the most expensive wine" on the list (Yang & Lynn, 2009). Price each dish to earn its place; a dish that exists only to make others look cheap takes a line of the menu and some of the kitchen's attention.

What moves menu profit more reliably

Compared with these effects, the economics of the menu are large and certain: what each dish costs to make now, what it earns in dollars, and how often it sells. Menu engineering puts those on one page. A common rule classifies a dish as popular when its share of its section's sales exceeds 70% of an equal share (LeBruto, Ashley & Quain, 1997); the cutoff is a convention, not a law.

Test a change on your own menu

1. 1.Change one thing: a position, a description, a price format. Changing three at once tells you nothing about any of them.
2. 2.Compare equal periods — at least four weeks each, same days of the week — and note anything else that changed: prices, season, weather, events.
3. 3.Compare the dish's share of its section, not its raw count, so a busy month does not look like a successful redesign.
4. 4.Judge the result in contribution dollars. A dish that gains orders from a higher-earning neighbor can lower profit while its own count rises.
5. 5.Treat the answer as an observation. Without randomization, a before-and-after difference suggests an effect; it does not prove one.

Marji does not import POS sales, so item counts for a test come from your POS item report. For the cost side, every extracted invoice line needs your review before confirmation records an ingredient price for recipe recosting. Check the recipe's quantities, units, yields and completeness before using its plate cost in the contribution calculation. Invoice extraction alone does not establish a reviewed dish cost.