Nesta and the Behavioural Insights Team (BIT) have been exploring how changing the way food options are presented can help encourage consumers to eat more healthily when ordering takeaways online. In our first round of experiments we show how providing default options, eliciting social norms and making smaller portion sizes available can help nudge consumers towards ordering fewer calories, which could play an important part in our mission to reduce obesity.

Online platforms such as Deliveroo, JustEat and UberEats have made ordering takeaway meals easier than ever. Accelerated by the pandemic, the use of takeaway delivery platforms has grown rapidly over the last five years — a recent survey found that two-thirds of adults had used the most popular delivery platform, JustEat, in the previous year.

We all know that ordering a takeaway is rarely the healthiest choice, and research supports this: the Food Standards Agency (FSA) estimates that ‘out-of-home’ meals are 21% more calorie dense than those cooked at home and academic research has found that exposure to takeaways is associated with higher obesity prevalence.

Nesta has been working with the BIT to explore what fresh opportunities online delivery platforms may present for us to help people live healthier lives. We are particularly interested in understanding how small changes to the way food options are presented can “nudge” consumers to choose healthier food options without restricting the actual food options they can choose from.

Last year, with support from Nesta, BIT developed Take a BITe, a simulated online takeaway experiment platform. It allows us to test how changes to the way restaurant, food and portion size options are presented can affect consumers' food choices. Participants recruited to take part in experiments on the platform can choose between five types of cuisines, 15 restaurants and 381 food or drink items, each in at least three different portion sizes (you can try the platform out for yourself).

While those participating in experiments on Take a BITe do not use their own money or actually receive the foods they order, past research has indicated that shopping behaviour in simulated stores is comparable to that in real-life stores. This suggests that experiments on hypothetical platforms such as Take a BITe could help us understand which out of the many possible interventions we could implement are likely to be most impactful when implemented in a real-world setting; and if an intervention does not have an impact in the controlled environment of our simulated platform, it might not be worth testing it in the real world.

The image below shows what participants see as they interact with Take a BITe.

In our first round of experiments on Take a BITe, we tested whether the following behavioural interventions could nudge participants towards selecting smaller portion sizes:

• Default – When people are presented with a pre-selected option (a “default”), they are typically more likely to select that option. In our experiment the smaller portion was pre-selected as the default.
• Default + social norm – People tend to conform to social norms, meaning that they prefer to do what they think other people do. On top of the pre-selected default, the small portion was labelled as “regular”, the medium as “large”, and the large as “extra large”. This signals to participants that the small (or “regular”) portion is the normal choice and other options are deviations from the norm.
• Default + social norm + availability – When people are presented with a list of options they have a tendency to select central options. On top of the preselected default and social norm, we added a new “extra small” portion (labelled as small and made the default). This puts the originally “small” portion (now labelled as regular) in the centre of the list of portion size options.

We randomly allocated 6,000 participants to one of the three interventions above, or to a control group that used the business as usual version of the Take a BITe platform (ie, three portion sizes were available labelled small, medium, or large, and none were pre-selected as defaults). Participants were largely representative of general internet users with respect to age, gender, socioeconomic status, location etc. All participants were asked to place a hypothetical order for themselves using Take a BITe.

All three interventions significantly reduced the calorie content of participants’ baskets compared to the control. The best performing intervention – default + social norm + availability – on average led to a 177 kcal reduction per order, as can be seen in the graph below. The interventions worked as intended, by reducing the portion sizes selected, without affecting other behaviours such as the restaurant or food type chosen, or the number of items selected.

At Nesta we are very keen to make sure that interventions do not increase the health gap that exist between the most and least affluent people in the UK. In this study we found some preliminary evidence that the interventions might work better for people from lower socioeconomic backgrounds and therefore could help reduce health inequalities.

Alongside decreasing calories purchased, the interventions tested in this study reduced the average basket price by £1.18-2.79 per order. This is an important limitation from the point of view of takeaway platforms and restaurants, and will likely limit uptake of such interventions.

Taken together, this study suggests that behavioural interventions involving pre-selecting small portions as defaults, relabelling small portions as regular, and making additional smaller portions sizes available could significantly reduce the number of calories that people order from online food delivery platforms.

Public opinion is important to the government and the food industry, and so public acceptability is likely to affect the delivery and uptake of potential interventions designed to encourage healthy eating. We surveyed the participants in the takeaway platform experiment, which were largely representative of general internet users, to understand their attitudes towards a range of potential changes that could be made to takeaway delivery platforms which would aim to encourage healthier choices.

The majority of platform users support health interventions in delivery platforms: 67% of delivery platform users would like delivery platforms to be modified to increase the selection of healthy and reduce the selection of unhealthy foods, and 57% would welcome government interventions to promote these changes.

Public support for the three interventions tested in this study was mixed: setting defaults to pre-select lower-calorie / healthier portions (ie, defaults) was supported by 46%, changing portion size labels to elicit social norms (eg, changing the names from small to regular) was supported by 50%, and adding a smaller portion was supported by 56% of participants. Importantly, very few people (seven percent or less) strongly opposed any of the health interventions we asked about.

Promoting healthy choices is more popular than discouraging unhealthy ones: The most acceptable intervention in food delivery platforms was adding promotions on low-calorie / healthy foods (supported by 68%). The least acceptable intervention in food delivery platforms was removing promotions from high-calorie / unhealthy foods, (supported by 36%).

Health interventions on delivery platforms could have mixed effects on businesses: 71% of delivery platform users said they would be more likely to use food delivery platforms if they were modified to increase the selection of healthy and reduce the selection of unhealthy foods. However, this increased usage may be offset if interventions result in consumers ordering fewer, cheaper, or lower profit margin items. Further work needs to be done to identify interventions that decrease calories purchased without negatively impacting takeaway platforms and restaurants profits.

In summary, our survey suggests that there is already some public support for health interventions in delivery platforms; there is still room to increase this support.

In the long term we hope to test the impact of these types of interventions on real platforms. However, more immediately we plan on running further experiments on the Take BITe platform to test the relative impact that other interventions could have on consumers' food choices.

The next set of interventions we want to test focus on repositioning unhealthy food products to be less prominent at point of purchase. Repositioning interventions have shown some promise as a method of promoting healthier choices, but remain a relatively under-explored strategy. Furthermore, positioning interventions can be designed to benefit both businesses and individual’s health, eg, they can be designed to nudge people towards healthier products with similar profit margins , potentially making them a more feasible intervention for restaurants and takeaway platforms to implement.

We are interested in working with policymakers and retailers of all sizes to explore ways of making online services better for customers' health and for business. If you have any thoughts or questions, or would like to work with us, please get in touch with [email protected]