Item-Level Promotions: The Deals Hidden Inside Restaurant Menus

Item-Level Promotions: The Deals Hidden Inside Restaurant Menus

Item-level promotions can be hard to judge from a restaurant's offer banner. “Buy 1, get 1” does not identify the qualifying dish. “20% off select items” does not identify the items or establish the saving on your order. The useful details may require a closer look at the menu and promotion terms.

DealMeal Labs analyzed 253,204 USD offer-days across Uber Eats, DoorDash, and Grubhub, plus a separate snapshot of 353,625 menu items. The evidence shows a difference between naming an item in a promotion title and placing items in a promotional menu section. It does not establish how many menu items were actually discounted or whether restaurants deliberately concealed their deals.

Quick answer

Open the menu before treating a restaurant's headline promotion as an offer on the food you want. DealMeal's title patterns detected specific item names in 4,523 offer-days, or 1.79% of the offer dataset: 953 distinct offers at 852 stores, using 380 title strings. All detected matches were on Grubhub, although the patterns missed additional item-naming phrasings and DoorDash separately named two fixed-price meal bundles.

In the separate Northern Virginia menu analysis, promotional sections were more common on Uber Eats. Their presence does not prove a discount: the menu data contains list prices, without an item-level before-and-after price. Check the eligible item, the reward, the minimum spend, and the final cart before deciding whether the promotion improves the order you intended to buy.

Key findings

• Item naming was rare in the observed titles. The 4,523 detected offer-days represented 953 distinct offers, 852 Grubhub stores, and 380 titles. They were 16.04% of Grubhub's offer-days in the multi-metro US sample.

• Vague wording was common. A narrow definition identified 58,513 offer-days, or 23.11%, as explicitly vague about the item. A wider definition increased that to 83,005, or 32.78%.

• “Select items” leaves an important calculation unfinished. Of 10,637 such offer-days, 70.42% recorded neither a minimum spend nor a savings cap. The available data could not establish their dollar value.

• A named BOGO may reward a different item. Automated comparison flagged 536 of 1,398 buy-one/get-one pairs, or 38.3%, as potentially different-item rewards. This is an upper-bound estimate pending manual review.

• Promotional-section dishes had modestly lower list prices in one specification. With the broader section classifier and a dish-only comparison, the median within-restaurant gap was −$1.00 across 408 platform-restaurants. This is not a measured discount.

What are item-level promotions?

An item-level promotion depends on particular menu items rather than applying automatically to everything in the basket. Examples include buying one named dish to receive another, getting a specific free item after reaching a spending threshold, or receiving a percentage discount on selected menu items.

Three pieces of information can be separate: the restaurant's main promotion, the menu heading where participating items appear, and the eligibility rules shown before checkout. A generic banner may announce the deal without identifying the dish. A menu section may help locate candidates without establishing the final price or proving that every item in that section receives a discount.

The distinction matters because the same intended basket can contain eligible and ineligible food. Before calculating a percentage saving, identify the food it covers. Before treating a BOGO as two meals for the price of one, identify both the purchased item and the reward.

Which promotion titles actually name a menu item?

The offer analysis covered July 25 through September 2, 2026: 39 observation days across a 40-day span, with August 29 missing. The 253,204 offer-days resolved to 51,572 distinct offers. An offer-day means an offer observed on a particular day; it is not a separate promotion every time it appears.

Three non-overlapping title patterns detected item naming: “Buy One X, Get One Y Free,” “Free X,” and “BOGO X,” with generic or selected-item expressions excluded. Together they identified 4,523 offer-days from 953 distinct offers at 852 stores, using 380 titles.

The Grubhub matches occurred at 852 of its 5,677 observed stores, or 15.01%. Named examples included McNuggets, Royal Crispy Chicken, Crunchwrap Supreme, and Bean Burrito. These titles clustered on chains, so the observations should not be treated as a random sample of restaurants.

Zero detected matches is not proof that a platform never names a product. DoorDash recorded 3,197 offer-days titled “$12 Meal Box” and 22 titled “$35 Family Meal.” The supplied analysis identifies these as DoorDash fixed-price bundles rather than named restaurant dishes. They are genuine product-name exceptions and fall outside the three detection patterns.

The patterns also missed 136 BOGO offer-days, including phrasings such as “Buy Latte; Get free Croissant.” Those records were all on Grubhub, but they demonstrate that 1.79% is a detected share under a defined method, not an exhaustive count of every item-naming promotion.

Generic banners compress very different deals into the same wording

Three generic title strings accounted for 45,695 offer-days, or 18.05% of the entire corpus: “Buy 1, get 1 free,” “Buy 1, get 1,” and a leading-space variant. A further 2,181 offer-days said “Buy 1, get a free item.” These titles describe a mechanism without identifying the food.

Using the narrow definition, 58,513 offer-days, or 23.11%, were explicitly vague about the item. The wider definition identified 83,005, or 32.78%. The difference reflects the classification rule; neither figure means that the remaining titles necessarily provided complete eligibility information.

Observed title diversity also differed. Grubhub had 649 distinct titles, or 23.01 per 1,000 offer-days, compared with 575 and 7.87 on Uber Eats and 405 and 2.67 on DoorDash. Grubhub's rate was 8.6 times DoorDash's. That describes wording diversity per unit of observation, not the number of unique bargains, their quality, or who authored them.

A food-word check provided another view of the same title dataset. None of the combined 225,003 DoorDash and Uber Eats offer-days contained any of 33 specified food words. Grubhub had 2,871 such offer-days, or 10.18%. The result depends on the word list: adding “meal” identifies the 3,219 DoorDash bundle offer-days described above.

BOGO deals: check what you buy and what you receive

Among 51,437 BOGO offer-days, the patterns identified 2,930, or 5.70%, as naming the items. Another 48,371, or 94.04%, were explicitly vague, while 136, or 0.26%, fell into neither category under the detection rules.

Within Grubhub, 2,930 of 3,541 BOGO offer-days, or 82.74%, named an item under those rules. All 1,493 Grubhub free-item offer-days did so. These results concern descriptions within the observed dataset; they do not show that Grubhub had more item-level deals than another platform.

The most explicit format still requires careful reading. In the 1,398 “Buy One X, Get One Y Free” offer-days, string comparison flagged 536, or 38.3%, as potentially rewarding something different from the purchased item, across 109 titles and 105 stores. That is an upper bound pending manual review: parsing errors and legitimate variants can make names look different without establishing a meaningful cross-sell.

Examples illustrate the consumer issue. “Buy One Cookies-N-Cream Milkshake, Get One Mocha Latte Free” pairs different drinks. “Buy One General Tso's Chicken, Get One 1 Piece Vegetable Spring Roll Free” pairs a main with a side. Neither description promises a duplicate of the purchased item. A generic BOGO banner alone would not communicate that distinction.

Ask whether the reward serves the purpose of your order. A side, drink, or dessert can have value, but it may not replace the second main you planned to buy. For context on promotion formats, see DealMeal's analysis of which food delivery app has the most BOGO deals.

“Select items” can advertise a rate without revealing its dollar value

The analysis found 10,637 offer-days containing “select items,” representing 4.20% of the corpus, 148 distinct titles, and 2,385 stores. They included 7,491 percentage-off, 2,651 dollar-off, and 495 BOGO offer-days.

For 70.42% of this population, neither a minimum spend nor a savings cap was recorded. Without the eligible item names and prices, the available dataset cannot turn the promotion into a reliable dollar saving. The missing information may be available to a diner inspecting the live menu; its absence from the dataset does not mean that nobody can calculate it.

The item-scoped percentage offers advertised a median 25% rate, compared with 20% for offers that could be valued. That is a difference in advertised rates, not evidence of greater realized savings. The eligible portion of your intended basket, its prices, and any applicable limits determine the outcome.

None of the 79,532 percentage-off offer-days matched the item-naming patterns, while 7,491, or 9.42%, explicitly referred to selected items. A large percentage can therefore coexist with an unidentified eligible subset. The analysis does not establish that these offers were worse; it establishes that their value could not be verified from the available records.

Read the scope before comparing the rate. DealMeal's report on free-item-with-purchase delivery deals explores a related problem: an offer can announce a reward without clearly identifying it.

DealMeal data insight: item names and minimum spend tell different stories

Within Grubhub's multi-metro offer sample, the 4,523 detected item-naming offer-days—953 distinct offers at 852 stores, using 380 titles—recorded a minimum spend on 99.07% of rows, or 4,481 observations. Their median minimum was $15. Grubhub's other 23,678 offer-days had a $50 median minimum, with 98.15% recording a threshold.

Threshold coverage was similar, but threshold size differed. This is a within-Grubhub comparison, not a comparison of Grubhub against the other platforms. A lower qualifying basket may make an item-specific offer relevant to a smaller order, but the reward and final total still need checking.

Promotional menu sections reveal a separate part of the picture

The menu analysis used a September 21, 2026 snapshot of 353,625 items across 3,763 platform-restaurants and 2,772 branches. Its scope was Northern Virginia, unlike the multi-metro offer-title dataset. A platform-restaurant is a restaurant listing on one platform; the same physical branch can have multiple listings.

A tight menu-heading classifier identified 1,997 items, or 0.565%, in promotional sections at 375 platform-restaurants, or 9.97%. A broader classifier added headings associated with limited-time offers, bundles, coupons, happy hour, and similar promotions. It identified 3,001 items, or 0.849%, at 561 platform-restaurants, or 14.91%.

Uber Eats had the highest observed prevalence of promotional menu sections in this Northern Virginia snapshot. Separately, Grubhub accounted for all pattern-detected item-naming titles in the multi-metro offer analysis. These are two scoped findings, not matched evidence that one platform exposes more discounts or that every platform consistently follows one presentation strategy.

Menu headings require interpretation. “Buy 1, get 1 free” is directly promotional. “Most Ordered” is a ranking heading, and “Chef's Special” may be ordinary menu naming. Both classifiers excluded ranking and specialty labels. Even the broader classifier may miss relevant headings, so section counts depend on a stated rule rather than complete promotion identification.

Are dishes in a deals section actually cheaper?

There was no item-level discount field in the menu data: no former price, reduced price, or recorded percentage reduction. The comparisons below concern list prices. An item listed under a promotional heading is not established to be discounted.

Item mix also changes the answer. Combos made up 28.71% of promotional-section items versus 5.03% elsewhere, a 5.7-times difference. Their median list price was $20.90, compared with $13.99 for dishes. Comparing all menu items can therefore compare more combos on one side with more individual dishes on the other.

DealMeal compared median prices within each platform-restaurant and ran five specifications. The first table reports direction and price differences; the second gives the supplied statistical results. “Cheaper,” “dearer,” and “tied” refer to the restaurant's promotional-section median relative to its other-menu median under that specification.

The broader, dish-only specification found lower promotional-section medians at 216 platform-restaurants, higher medians at 172, and ties at 20. Its median gap was −$1.00, with p = 0.029. The interval for the probability of a cheaper median, 0.506–0.607, is close to an even split; the narrower dish-only specification had p = 0.093. The result is modest and depends on the specification.

This is not a comparison of identical dishes sold with and without a promotion. Different dishes can have different ingredients, portions, and prices even within one restaurant. The evidence supports slightly lower dish list-price medians under the broader classifier, not a causal saving or a guaranteed $1 discount.

BOGO sections produced a separate null result under the tight classifier: 105 platform-restaurants had cheaper medians, 95 had dearer medians, and seven were tied, for 207 in total. The median gap was −$0.30, with z = 0.64, p = 0.525, and a 95% interval of 0.453–0.596. A first item's list price alone does not measure the value of receiving a second item.

Do BOGO menu sections and banners appear together?

Only 1,295 stores appeared in both datasets: 34.4% of the menu dataset's restaurants and 2.90% of the offer dataset's stores. Coordinates were available for 763 of those joined stores; 98.7% of that located subset was in Northern Virginia. The offer window ended 19 days before the menu snapshot.

Among the joined stores, 138 had a BOGO menu section, 464 had a BOGO banner observation, and 123 had both. A BOGO section occurred at 26.5% of stores with a BOGO banner, compared with a 10.7% baseline across the joined sample. The supplied odds ratio was 19.6, with Fisher exact p = 2.75 × 10−43.

This is a strong association in a small, geographically concentrated overlap across different dates. It does not mean banners caused sections or that both were active together. More importantly, the join includes stores with an observed offer by construction. It cannot answer whether promotional sections generally go unadvertised. Fifteen joined stores had a BOGO section without a BOGO banner observation, but that is not proof of a currently hidden BOGO.

How to evaluate item-level promotions before ordering

1. Choose the food first. Start with the dishes and quantities you actually want, then see whether a promotion fits.

2. Open the eligible menu section and terms. Confirm the named item, size, variation, and any qualifying selection instead of relying on the banner.

3. Read both sides of a BOGO. Identify what you buy and what you receive. Check whether the reward replaces something you planned to order.

4. Check thresholds and limits. Look for the minimum spend, any maximum saving, and the eligible items. Do not apply a selected-item percentage to the whole basket.

5. Confirm pickup or delivery eligibility. Use the ordering method you intend to choose; this analysis does not establish that an item-level promotion works through both.

6. Compare the final intended cart. Review the eligible saving and displayed charges before placing the order. Recheck availability because these observations are historical.

DealMeal's guide to finding and comparing food deals with DealMeal explains the discovery workflow. Use discovery to find candidates, then confirm item eligibility and checkout terms on the ordering platform.

Methodology and limitations

The offer dataset contains 253,204 distinct USD offer-days and 51,572 distinct offers across Uber Eats, DoorDash, and Grubhub, observed from July 25 through September 2, 2026. Records were filtered to USD, non-offer placeholders excluded, and daily offers deduplicated by date, platform, offer identifier, and title. The observations span 107 geographic cells, with 36.29% associated with Northern Virginia. They are not a nationally representative restaurant sample.

Item naming was detected from three title patterns: buy-one/get-one with named items, titles beginning with a specific free item, and titles beginning with a specific BOGO item. They contributed 1,398, 1,593, and 1,532 offer-days respectively, without overlap. The method excluded 513 generic “BOGO Select …” offer-days. Boundary choices were imperfect—for example, broader category names could qualify in one pattern while a selected-side expression was excluded—so these results are method-dependent.

The food-word check used this fixed list: burger, pizza, taco, burrito, sandwich, wings, nuggets, fries, shake, latte, roll, bowl, salad, sushi, pasta, noodle, chicken, steak, cake, brownie, cookie, donut, coffee, meatball, salmon, hoagie, sub, wrap, quesadilla, pancake, waffle, bagel, smoothie. It checks the same title dataset rather than independently replicating the analysis. “Meal” was tested separately and identifies the DoorDash bundle exception.

The Northern Virginia menu snapshot, dated September 21, contains 353,625 items at 3,763 platform-restaurants and 2,772 branches. Both tight and medium promotional-heading classifiers were reported. The 13,657 zero-priced records, including 22 in promotional sections, were excluded from price comparisons. Nearest-rank medians were calculated without interpolation. No discount was inferred from a list price, no prices were imputed, and no outliers were trimmed.

Price tests compared each platform-restaurant's median promotional-section list price with its other-menu median using two-sided sign tests with continuity correction. The tight, all-kind result was also checked at branch level: 352 pairs, with 172 cheaper, 164 dearer, and 16 tied, gave p = 0.703. That supports the same all-kind null in that specification, not the broader dish-only finding.

Named offer items were not matched to individual menu-item prices in the 1,295-store overlap. The price analysis therefore remains a section-based comparison, not direct valuation of named promotions. The menu findings come from one snapshot without independent replication. Neither instrument supports city rankings, current availability claims, or realized checkout savings.

Item naming also varied over time: Grubhub's daily count ranged from 48 to 219, with a median of 102; its daily share ranged from 10.76% to 23.91%, with a median of 15.57%. The first 27 days averaged 96.5 detected offer-days per day, while the last 12 averaged 159.8. The dataset does not explain the shift after August 20, 2026.

For more on interpreting the evidence, read how DealMeal measures food delivery deals. The DealMeal app works across the United States, while DealMeal.co currently showcases selected regional examples on the web; the narrower menu research scope does not limit the app's availability.

Frequently asked questions

What does “select items” mean on a food delivery promotion?

The offer applies to an eligible subset rather than automatically to the whole menu. Check the participating items and terms before applying the advertised rate to your basket.

Does BOGO always mean getting a second identical item?

The purchased item and reward can differ. Read both item names and check eligible variations. The dataset's automated different-item estimate is an upper bound pending manual review, rather than a verified cross-sell prevalence.

Are all items in a restaurant's deals section discounted?

A promotional heading alone does not establish that. This menu dataset contained list prices without former prices or item-level reductions, so it cannot identify how many items actually received a discount.

Did Grubhub have more item-level promotions than Uber Eats?

The analysis does not establish that. Grubhub accounted for all detected item-naming titles in the multi-metro offer dataset. Uber Eats had the most promotional menu sections under the tight classifier in the separate Northern Virginia snapshot. The measures and scopes differ.

Can a 25% selected-item offer save less than a 20% offer?

The final saving depends on the eligible food and applicable limits. The observed median rates cannot settle that comparison without the item prices, terms, and intended basket.

Does this study prove restaurant deals are hidden from the banner?

It shows that many recorded titles did not identify participating items. It cannot establish how often promotional menu sections were entirely unadvertised, because the joined sample included stores with observed offers by construction.

Choose the eligible meal, then judge the deal

Item-level promotions deserve a closer look because a short banner can omit the details that determine whether an offer fits your meal. Naming a dish, placing it in a promotional section, and reducing its price are different facts. The evidence supports checking each one separately.

For a BOGO, read both item names. For a selected-item percentage, identify the eligible portion of the basket. For a deals section, inspect the dishes and terms rather than assuming its heading guarantees a saving. The most useful promotion is one that improves the value of food you already want.

Find candidate deals, then check the details

DealMeal helps you discover and compare available food deals across supported platforms and nearby restaurants. Use it to find relevant offers, then confirm eligible items and the final cart on the ordering platform.

Get DealMeal and compare available food deals