Rapid Evidence Review: Implementation of Portion Size and Portion-Related Energy Reductions in Out of Home Settings

Database searching identified 211 records (Web of Science n = 147; PubMed n = 64). After deduplication and removal of non-English records, 162 records were screened at title/abstract and 34 full-text reports were retrieved and assessed for eligibility; 8 were excluded (7 reviews and 1 outside the 2016 onwards date range). One additional grey literature report was identified via Google searching and included (Table 1), giving 27 included records in total (26 peer-reviewed and 1 grey). The evidence base was concentrated in the United States (9/27, 33%) and the United Kingdom (9/27, 33%), with additional studies from the Netherlands (3/27, 11%) and other countries (6/27, 22%). Across the peer-reviewed studies, publication years ranged from 2016 to 2024, with most studies published between 2018 and 2023.

In terms of setting, 13 of the 26 peer-reviewed studies were conducted in real-world settings, such as restaurants, cafés, cafeterias/canteens, or takeaways. Eight studies were conducted in laboratory settings and five in online or simulated environments. This mix is useful for understanding mechanisms and feasibility, but it also means that effects observed in laboratory or hypothetical ordering studies may not translate directly to fast-paced OOH purchasing contexts.

Table 1: Number of articles identified, screened and analysed

Studies clustered around four broad and sometimes overlapping intervention types: (1) direct physical portion reduction and portion standardisation, where the amount served or sold was reduced or standardised; (2) portion composition or meal redesign, where the energy content of the portion offered was reduced by changing component balance or replacing higher-energy elements; (3) behavioural nudges, portion norms and menu architecture, where interventions aimed to shift what feels normal or encourage selection of smaller portion options; and (4) information and labelling approaches, where calorie, portion or related messaging was used to support smaller choices. Because many studies combined more than one intervention element, some studies are discussed in more than one subsection where relevant to different mechanisms of action.

Using the study categorisation completed during extraction, direct physical portion reduction and portion standardisation were the most commonly identified intervention components (15/26 peer-reviewed articles). Information, labelling or communication features appeared in 8/26 studies. Format manipulation approaches (for example, plateware or packaging related) were present in 5/26 studies. Elements relating to social norms and pricing appeared in 4/26 studies each. Waste management features were present in 3/26 studies, and supply-chain level interventions were identified in 2/26 studies. Approaches were often combined rather than tested in isolation: 13 studies used a single main approach, 11 used two approaches, and 2 combined three approaches.

The included studies varied considerably in their policy relevance and evidential strength. Some were real-world evaluations conducted in restaurants, cafeterias or other OOH settings over multiple days or weeks and using purchasing or sales data, while others were short laboratory or simulated ordering studies designed to test specific mechanisms. This distinction is important when interpreting the findings: controlled studies are useful for understanding how portion-related interventions may work, but real-world studies are more informative for implementation, feasibility, and likely impact in OOH settings.

Direct physical portion reduction and portion standardisation were the most commonly identified intervention components in the included evidence. These interventions reduced the amount served or sold of a target item, introduced smaller portions as standard, or used operational changes such as standardised serving tools, containers or portion specifications to deliver smaller portions more consistently. A smaller subset of studies reduced the energy content of the portion offered through changes to meal composition or by combining replacement of higher-energy elements with modest portion reduction; these are considered separately below because they are not equivalent to reducing physical portion size alone.

In a natural field experiment conducted in a restaurant over nine days, Kee et al. (2022) examined the role of selecting a standard portion or a smaller (50% fewer calories) portion size across all menu items at lunchtime. The smaller portion was accompanied with a donation to a food charity and/or messaging about food waste. The donation increased selection of smaller portions by 22.2%, increasing to 23.6% when combined with the food waste messaging. The change in purchasing was more commonly seen in those living with overweight. 

Reimann et al. (2017) examined a somewhat different context, testing children’s meal choices in a school-based setting using food options sourced from McDonald’s. In that study, a toy was paired with a smaller children’s meal, reducing energy by 160 kcal, and children preferentially selected the smaller meal (43% chose a smaller meal when it was paired with a toy, versus 7% when a regular-sized meal was paired with a toy). This suggests smaller portions can be made more attractive in children’s meal settings, although the relevance of this finding to adult OOH meal choices is limited.

In a stepped-wedge pilot trial in six worksite cafeterias, Hollands et al. (2018) reduced the portion sizes in cafeterias by 10% (by volume), but only for a subset of their products (up to half). With this more modest change, they did not find any significant change in the total number of calories sold each day. However, 44% of customers were in favour of the smaller meals, and this appeared to be linked to the accompanying reduction in price. This latter point was a noted concern from some of the nine businesses involved.

In a real-world intervention in a worksite cafeteria and a restaurant, Berkowitz et al. (2016) offered smaller versions of meals in both a worksite cafeteria (a half sandwich or small salad or soup) and a restaurant (smaller bar versions of full dishes). Both the original meals and reduced-size versions priced at about half the cost were available in both settings. Like Hollands et al. (2018), only a fraction of meals was altered and offered alongside the standard meals. In both settings, average calories consumed were reduced (11.7% and 43.5%, respectively) compared with periods when only the standard portion sizes were available. Reduced-size entrées accounted for 5.3% to 12.8% of selections in the worksite cafeteria and 18.8% to 31.3% in the restaurant setting, indicating that at least part of the reduction reflected uptake of the smaller meal option.

In an uncontrolled before-and-after study in fish and chip shops, Goffe et al. (2019) evaluated a multi-component intervention to promote smaller portions, including supplier-led engagement with shop owners and managers, provision of box packaging for smaller portions, promotional posters, and business incentives. Servings of chips decreased by 60 g for standard portions and 26 g for smaller portions, although fish size increased by 24 g for standard portions and 2 g for smaller portions. Sales of smaller portion sizes increased from 14% pre-intervention to 21% post-intervention. 

In a mixed-methods field experiment across three restaurants running from autumn 2022 to spring 2023, Cohen et al. (2024) also investigated the acceptability of changing to smaller standardised meals. They developed new menus using language around “balanced portions” (lunches/dinners ≤ 700 kcal and breakfasts ≤ 500 kcal) using standardised portions (e.g., a ¾ cup for rice). Customer footfall increased with the new menus, and the study suggested that the portion-controlled balanced menus were acceptable to customers. There was also no evidence that customers consumed additional calories on subsequent days. However, of the three participating vendors, one struggled with adherence because of additional preparation time. 

In a laboratory self-service study, Haynes et al. (2020) examined how smaller visually presented portions influenced self-service intake by changing bowl size. Portion sizes were determined by asking a separate set of participants to visually compare images of meals that represented different fractions of the manufacturer-recommended serving size, ranging from 40% to 300%. Three sizes were selected for comparison: a ‘normal large’ portion (120% of the manufacturer-suggested serving size), a ‘normal small’ portion (90% of a serving), and a ‘smaller than normal’ portion (60% of a serving). Participants given the smaller bowl did tend to help themselves to some additional food beyond the initial serving, but typically still consumed fewer calories overall than those given the larger portions.

Over consecutive days, Raghoebar et al. (2019) found that being served a smaller portion (60% of the reference meal made to the manufacturer’s recommended serving size) led to participants picking a smaller portion the subsequent day. They conclude that participants’ perceived descriptive and injunctive social norms were changed. They noted that personal norms were not changed. By contrast Vermote et al. (2018) reduced the portion size of French fries by 20%, but the majority of people recognised the reduction (even over-estimating the size of the reduction). Despite this, they were still satisfied by the quantity provided.

Direct reduction approaches were evaluated in a mix of settings. In cafeteria and canteen contexts, studies reduced portions of main meals (e.g., lasagna or curry and rice), sandwiches, or snacks. Several studies used purchasing or transactional data and assessed impacts over multiple days or weeks, allowing estimation of changes in energy purchased and feasibility of implementation at scale. In restaurant settings, direct reductions were more commonly tested through field experiments or service-based trials. Laboratory studies in this category tended to focus on measured intake during a single eating occasion. Some studies were designed around a whole meal, others focused on individual food items (e.g., discretionary snacks). These studies help to clarify the likely direction of effect, but provide limited insight into longer-term consumer adaptation or compensatory eating behaviours. Serving sizes tended to either be reduced as a proportion of what was previously used (e.g., in a restaurant setting) or relative to retailer recommended serving sizes. 

A smaller group of studies reduced the energy content of the portion offered without reducing the size of the meal, but by altering the proportion of meal components. This was done for example by reducing higher-energy components and increasing lower-energy components, or by combining the replacement of higher-energy items with modest reductions in portion size. These approaches are relevant to OOH practice because they may preserve perceived meal size or value, but they should be distinguished from direct physical portion reduction.

In a real-world stepped-wedge randomised trial across 19 worksite cafeterias conducted over 25 weeks, Reynolds et al. (2021) tested two ways of reducing calories purchased: replacing higher-energy foods with lower-energy versions and reducing portion size. They replaced 10% of high-energy foods with a lower-energy equivalent (by volume) and also reduced portion sizes (e.g., depending on the food, reducing the size of the serving-ware or reducing the count of meatballs served). Calories purchased reduced by 4% when the availability was changed, and 11% when both availability and portion size were changed simultaneously. Crucially, foods that were not actively changed also showed a spill-over reduction of 10% calories purchased when the high-energy foods were not available. Hollands et al. (2019) took a different approach and kept the appearance of a standard meal by increasing the serving of vegetables (increasing the vegetable volume between 31 to 237% of their original quantity) while they simultaneously decreased the amount of meat in the meal (by between 12% and 34% of the original meat volume). Across four different settings, including restaurants and self-service buffets, customers ate more vegetables. Reinders et al. (2017) similarly found that replacing a fraction of the meat with more vegetables increased vegetable consumption. The important result was that people were satisfied with the reduced meat content, albeit less satisfied than during the control phase when meals were not reduced. This is relevant because it suggests that portion-related interventions can change not only the energy content of meals, but also the overall balance of meal components, for example by increasing vegetables while reducing meat.

Rather than physically manipulating portions, several studies aimed to reduce energy ordered or consumed by changing behavioural cues and portion norms. These approaches included prompts and defaults, changes to naming and framing of portion options, social norm messaging, and “choice architecture” changes that make smaller portions more salient or easier to select. Within the extracted evidence, explicit social norm components were identified in 4 studies and format manipulation approaches in 5 studies, often alongside other intervention elements.

In a short experimental laboratory study with two sessions 24 hours apart, Robinson et al. (2019) tested whether participants could be trained to normalise smaller portions by showing participants pictures of smaller portion sizes. First, a study was carried out to identify how many biscuit-sticks were thought to be a portion size (from 2 to 24 were shown, while the manufacturer recommended portion was 11). Participants were then given either a small (6 biscuit sticks, (14g)) or large (16 biscuit sticks, (37g)) serving. Although only over a short 24-hour period, participants exposed to the new normative messaging on the first session tended to eat fewer biscuits on their second session than those originally exposed to the larger portion sizes and also estimated a ‘normal’ portion size as a smaller helping (8 biscuits).

Rather than show what a normal meal looked like, Kee et al. (2022) compared motivations around food choice, using two messages around health that might address personal reasons for food choice or a charitable donation that might impact others. The donation to a food charity was included in the price of a smaller meal (i.e., the portion was smaller, but the relative difference in price compared to a full meal was treated as a donation). The healthy message increased sales of small plates by 15.8% whereas the charitable messaging increased sales of the smaller plates by 22.2%. The two actions together increased sales of smaller portions by 23.6%. Participants living with overweight were more susceptible to all messaging and for this group, the combined messages were substantially more impactful than either alone. In a later study, Kee et al. (2023) used labelling to describe portions either in neutral size terms or with anthropomorphic labels. In a natural experiment at a State Fair café, the “Regular/Plus Size” labelling increased selection of the smaller portion by 14.3 percentage points when objective portion size information was also provided. Among individuals living with overweight or obesity, participants were 23.4 percentage points more likely to choose the smaller portion under the “Regular/Plus Size” labelling than under the “Slim/Regular” labelling. However, prices and portion sizes were also changed at the same time, with the larger portion priced at double the smaller portion, so the effect of labelling alone cannot be isolated.

Werkman et al. (2022) examined the effects of unit size and serving size on snack intake, waste and satisfaction. In the first part of the study, portion size was manipulated by offering participants either: i) a small serving of small cookies (6 × 3.6 g), ii) a small serving of large cookies (2 × 11 g), iii) a large serving of small cookies (12 × 3.6 g), or iv) a large serving of large cookies (4 × 11 g). When participants were served smaller cookies (3.6 g), they consumed less than when they were served larger cookies (11 g), without reporting lower satisfaction. This was observed for both smaller and larger served portions. Participants also wasted more uneaten food when provided with a larger serving than when provided with a smaller serving. Similarly, in a second part of the study, where participants served themselves, they took less food when gingerbread was presented in smaller units (6.25 g) rather than larger units (25 g). Across the experimental designs, smaller unit sizes consistently reduced intake. Although this evidence was limited to the snack foods provided in the study, calorie intake was reduced by up to 50%. Participants who served themselves smaller units were no less satisfied with the amount eaten than those who had access to larger units. Smaller unit sizes also increased the time participants took to choose their food, suggesting that they considered their options more carefully.

This category includes interventions intended to shift what feels like an appropriate amount, for example by exposure to smaller served portions, redesigning menus so that smaller portions appear as the standard or as the most prominent option and prompts that reframe smaller portions as normal or desirable. Some studies also tested pricing signals as part of the nudge, for example altering relative value cues that can encourage upsizing, recognising that perceived value for money is a major driver of OOH portion choice.

Ayaz et al. (2016) found that changing the size of plates, but not the portion size, had no effect on total energy intake. The logic was that how food appears on a plate ought to change how the portion size is perceived. The difficulty is that people tended to eat what they have been provided, a result shown both by Zuraikat et al. (2018) and Reinders et al. (2017). In a crossover laboratory experiment conducted over four weekly meal sessions, Zuraikat et al. (2018) automatically gave a take-home container for any leftover food to test whether people might eat less there and then. They tested with larger (rather than smaller) portions and found that people ate substantially more with larger portion sizes. ‘Doggy bags’ did reduce how much extra people ate with the larger portions, but people still ate more the more they were provided. 

In a series of controlled experimental studies, Schumacher et al. (2020) tested whether describing everyday snack products, such as chocolates and crisps, with a “surprise” label increased their appeal. The intervention was the label itself rather than an actual surprise item or reward. This labelling increased preference for larger snack portions, which the authors interpreted as reflecting positive anticipation.

Ge et al. (2018) asked how customers would respond to smaller portion sizes (up to 40% smaller). People were, on average, willing to pay a small premium for larger portions compared to smaller portions. Unlike Ayaz et al. (2016), this study found a negative relationship between how much of the plate was covered with food and the price premium (the emptier the plate, the lower the acceptable premium), but this relationship was dependent on people’s prior beliefs and their intention to buy food (whatever the price).

Overall, norm and nudge approaches provide evidence that portion choice can be influenced without removing autonomy. However, much of this evidence comes from short-duration laboratory or simulated ordering studies, which limits certainty about whether effects would be maintained over time and generalisability to high-throughput takeaway or food-to-go contexts.

Information and labelling interventions aimed to support smaller portion choices through calorie information, portion descriptors, prompts, or messaging linked to over-ordering and waste. Within the peer-reviewed evidence base, information, labelling or communication elements were identified in 8 studies. The review also included one grey literature report from NESTA about UK online takeaway ordering, testing multiple interface and messaging interventions in a large, simulated ordering study.

Information approaches were commonly evaluated as menu-based interventions (including online menus), and outcomes were typically calories ordered or selected rather than objectively measured intake. These interventions were often framed as relatively low-burden to implement but also depend on consumers noticing and acting on information at the point of purchase.

Information alone had limited effect on portion size purchases. However, it was a low-effort addition to other interventions that increased effect on portion-size purchases, for example, Haws et al. (2016) and Kee et al. (2022). Similarly, Hua et al. (2023) found that using different language to describe portion sizes, changing standard sizes to “just right” and large to “hearty” didn’t substantially matter on its own, but labelling smaller portions in general did increase their sale. However, like other studies, the prices were also decreased, and lower prices for the smaller dishes (either 50 or 70% the price of larger meals) promoted smaller meal choices. 

In a simulated online ordering study, Finlay et al. (2023) used a virtual delivery app to look at both calorie labelling and price. Participants in the randomised control trial were able to order from a sandwich shop, coffee shop and a fast-food outlet using an app. The app presented paired combinations of no labelling or labelling the calories, and standard pricing or pricing proportional to calories (between portion sizes). The calorie labelling didn’t change the portion size selected but did reduce the total energy of foods selected. When prices were proportional to calories, fewer large options were selected for the virtual coffee shop, but in the fast-food venue, people compensated for smaller portions by buying more smaller side dishes. An increased spend in all three settings demonstrated that a substantial number of people were still buying the more expensive large options. In a large online randomised experiment involving more than 2,000 participants, Hua et al. (2023) also looked at the sale of smaller portion sizes and found that changing the type of label of the larger portions had no impact on the sale of smaller portions. 

In a randomised online hypothetical ordering study, Haws et al. (2016) examined whether participants were willing to select smaller portions when price reductions were proportional. They compared two pricing strategies for calorie-controlled portions, either a proportional (linear) price change or a disproportional price change. Consumers were asked to pick items from an online menu of main dishes in full or half portion sizes. They selected lower calories when the pricing was proportional (halving the price with half-sized portions). The non-linear price reduction (25% lower price for 50% lower calories) led to no change in the calories of foods selected. However, the reduction in calories was not attributed to selecting smaller portions so much as healthier full-sized meals (based on calorie labelling).

A different way of informing consumers about the size of portions was used by Sharma et al. (2023). Here, augmented reality was used to visualise portion size. Compared to a printed picture, the augmented reality reduced over-ordering for home deliveries, but did not change order sizes in a restaurant setting. Lorenz-Walther et al. (2019) also focused on food waste as a motivator, but used posters to highlight the issue. In their study, reducing the portion size did reduce waste, but the posters had negligible effects. They found that customer behaviour split into either making a more conscious effort to clear their plate or ordered something different altogether.

Across studies, effects of information and labelling were variable and context dependent. Effects appeared more plausible when information was paired with changes to the ordering environment (for example prompts, defaults, or clearer value cues), rather than as a standalone strategy.

Across studies, outcomes were captured at different points in the purchasing and consumption pathway, which limited comparability and precluded meta-analysis. Many studies measured which portion option was selected (for example smaller vs larger, half vs full, or acceptance of default options), often using observed choice, ordering tasks, or transactional measures. Because studies differed in whether they measured selected portion size, calories ordered, calories purchased, immediate intake, or later compensation, effect sizes are not directly comparable across all studies.

A common outcome in field studies and online ordering experiments was calories ordered or purchased. This was frequently derived from menus or product information. In some cases it was taken from sales and transaction data and in one case it was summed for the outlet rather than the individual consumer. These outcomes are directly relevant to population exposure to energy from OOH foods, but they do not capture plate waste or later compensatory eating unless paired with additional measurement.

Laboratory, and some field studies, measured consumption during the eating occasion. These studies provide clearer evidence on intake compared to calories ordered/purchased but often occur under controlled conditions and may not reflect real-world OOH purchasing constraints and habits. They are also specific to the eating occasion rather than representative of a diet.

Very few studies assessed whether reduced portions were followed by increased intake, either at that moment (e.g., ordering additional dishes) or at subsequent eating occasions. When measured, follow-up was usually short and methods varied, limiting firm conclusions about longer-term compensation.

The focus of several studies was food waste rather than food intake. Food waste (food left on the plate) was estimated from the quantity left or, in one case, binned. Several studies examined food waste alongside intake, recognising that portion-related interventions may change plate waste as well as energy consumed. Definitions and measurement approaches varied, including objective plate waste measures and self-reported behaviours (for example intentions to take leftovers home, which would itself only reduce energy intake if the leftovers displaced other foods that might have been consumed).

A smaller subset of studies measured acceptability, satisfaction, perceived value for money, and operational feasibility. Very few studies reported business-relevant outcomes such as impacts on sales patterns or implementation rates, which is a key evidence gap for OOH policy translation. One study, which recruited businesses, discussed the operational challenges and adherence to the portion-size protocol.