Dish photography and descriptions: the numbers that decide the click, and what to do with each one

What the algorithm measures first is not the food, it is the complete listing?

Before we argue about framing, accept something uncomfortable: Google does not score appetite, it scores signal.

A Google Business Profile listing with photos gets 42% more direction requests and 35% more website clicks than one without them (Google Business Profile), and that gap opens with the first image, not with number eighty. The Chapinero restaurant I mention above had 180 photos and not one of the dish carrying a 68% margin, so the listing was full and empty at the same time. When you upload the storefront, the team and somebody's birthday dessert, you are telling the local algorithm that you sell generic atmosphere. The decision those two figures trigger together is simple, and we apply it in that order: audit which dishes carry the sales, then decide what gets photographed, never the reverse. Dropping the same square photo into Maps, delivery and the table menu is the cheapest way to lose three times.

Three storefronts, three criteria, and one mistake that costs in all three

Maps rewards variety and freshness of material; delivery platforms rank the item with an image and a complete listing above the one that shows up bare; and the printed menu moves average check through the written description, without touching price. Here is a figure almost nobody cross-references: Shake Shack's kiosk check runs higher than other in-store channels by a «high double-digit percentage» (Shake Shack, 2024 earnings call), and a kiosk is nothing more than a menu with a photo and a description that nobody interrupts. The operating takeaway: produce three versions of the same dish — vertical for delivery, horizontal for Maps, dense copy for the table — and measure each channel separately. The difference is not the gear. A recent phone with window light at eleven in the morning beats a badly handled DSLR under green fluorescents, and fixing white balance at midnight is work nobody should pay for twice.

The camera matters less than the selection: 20% of the menu pays for the party

The difference is WHAT gets photographed. Menu engineering has shown for decades that 12 to 15 dishes concentrate 70% to 80% of sales on an average menu, and inside that group two or three carry high contribution margin and deserve double the visual production. Diego F. Parra and the Masterestaurant team build that list before anyone picks up a camera: first the item-level sales report, then unit contribution margin, and only then the shoot. If the budget covers fifteen good photos, make it those fifteen instead of a sweep across the whole menu. There is a real tension here and I would rather name it before resolving it. A photogenic cheesecake running 41% food cost pulls attention across the listing and ends up cannibalizing the dish that leaves 68%, so you raise traffic and drop profitability in a single move. The answer is not hiding the dessert: it is ordering visual hierarchy so the high-margin dish takes first position in the gallery while the dessert comes in as reinforcement, third or fourth.

The cheesecake paradox: the photo that sells most is not always the one that earns most

What happens if you invert that order for a quarter? The mix shifts toward the pretty item, average food cost climbs two or three points, and on a 30% margin those points eat close to a tenth of operating result. The call: sort the gallery by margin, sort the printed menu the same way, and park the dessert where it does not compete with the main. A dish description performs when it names the attribute the guest says they value, and the industry has those statements measured. Some 72% of U.S. diners pay more at restaurants with sustainable practices, and 18% would accept 6% to 10% on top (Toast, Restaurant Sustainability Survey 2025); roughly 44% are motivated by locally sourced ingredients (Toast, 2025); 38% would pay more for protein-rich dishes (Nation's Restaurant News, 2025); and down at the nutrition detail, 36% pay more for low cholesterol and 30% for low sodium (Nation's Restaurant News, 2024).

Which words get paid: the guest already told you what earns a premium?

Four figures, one reading: the empty adjective earns nothing, the verifiable attribute does.

Rewrite the descriptions of your fifteen leading dishes citing origin, technique and protein weight, and delete «delicious», «exquisite» and the rest, which will not lift your check by a single peso. The worst-photographed category on almost any menu is the bar, and that is where the margin lives. Wine consumption fell 5.8% in 2024 in the United States (Circana, 2024), while beer already accounts for more than 4 out of every 5 drinks sold on-premise (Technomic via Nation's Restaurant News, 2024), and 32% of casual dining guests say they are interested in ordering a mocktail (Circana, 2025). That 32% is demand with no visual supply in most of the venues I review. A mocktail costs between 1,200 and 2,000 pesos to produce and sells at cocktail price, which makes it the best margin-per-photo item in the whole operation.

Beverage photography: the margin that gets photographed least is the one that earns most

Takeaway: if your gallery does not hold a single photographed drink, spend the next session on three — a poured beer, a mocktail and the house cocktail — and push them to Maps and delivery the same day. That images move the mix is not opinion, it has been measured at large chains. Orders including McDonald's $5 Meal Deal showed a check 12% higher than orders without it, and roughly 25% of McDonald's customers ordered it against 10% at Burger King (M Science via Restaurant Business, 2024). The gap between 25% and 10% is not explained by the recipe: it is explained by exposure, by the photo and the item listing repeated on every screen. Add that 62% of customers would lose loyalty without a personalized experience (Nation's Restaurant News, 2024) and that 42% of younger guests share a main course more often (Acosta Group, 2025). The joint decision: photograph at least one shareable build and one combo with visible pricing, because both formats lift check and both sit image-less on nearly every listing.

The 3 numbers you should tattoo on yourself

Three numbers and one action for each. First: 42% more direction requests and 35% more clicks with photos on the listing (Google Business Profile) — action, this week upload ten new images led by your highest-margin dish, not by the storefront. Second: 70% to 80% of sales come from 12 to 15 dishes — action, pull the item-level sales report, mark those fifteen and ban spending a single peso of production outside that list until they are covered. Third: 72% pay more for sustainability and 18% would accept 6% to 10% on top (Toast, 2025) — action, rewrite the descriptions of those fifteen dishes naming origin and technique, and test a 6% increase on the three with the best margin for four weeks. If the mix holds by week five, the photo session has already paid for itself with room to spare. The gap is not the camera. A recent phone shooting by a window at eleven in the morning beats a mishandled DSLR in a kitchen lit by green fluorescents, as anyone who has ever fought a white balance at midnight knows.

Where the comparison actually breaks?

The gap is WHAT gets photographed:

the traditional method covers the whole menu with equal effort, while menu engineering says 12 to 15 items drive 70-80% of sales, and that inside that group two or three carry a high contribution margin and deserve twice the production. There is a genuine tension here worth naming before resolving it: the photogenic dish is rarely the profitable one. A cheesecake at 41% food cost pulls the diner's eye, while the fish of the day at 26% cost and a plain plating never gets a shop window. The bridge is classic Kasavana-Smith menu engineering: the photo goes to the high-popularity, high-margin quadrant, and the high-margin item with weak popularity gets the most carefully written description, because words can create demand where an image cannot. On description the gap runs wider than on photography. A University of Illinois study led by Brian Wansink measured that descriptive, evocative menu names lifted item sales by 27% and raised perceived food quality, with the same dish at the same price.

Where the comparison actually breaks — in practice?

Twenty-seven percent for writing better. And still, most menus I review carry three words per line, because somebody decided back in 2014 that a menu had to look minimal.

The third break point is the delivery algorithm, which does not see photos: it reads text. Uber Eats, Rappi and DiDi Food rank internal results off the product listing, and an item with no description is invisible to a customer typing «ceviche» into the app's search bar. A 24-word description naming ingredients is not literature here, it is indexing. That single line does two different jobs: it seduces on the printed menu and it classifies on the platform. Then the part nobody budgets for. A 62-dish outside shoot runs USD 1,100 to USD 2,200 in Bogotá or Mexico City, and a month later 30% of those items are off the menu.

Where the comparison actually breaks — key points?

The Masterestaurant protocol produces 15 dishes across two mornings with a phone, window light, a neutral backdrop and a framing guide, and frees budget for what actually scales:

geotargeted ads inside a 3 km radius, which convert nothing without decent photography, not even at double the spend.