Choice Architecture for Food Service and CPG Brands: How to Nudge Sustainable and Healthy Choices

Last updated October 30, 2026

In a survey cited by Katherine White, David Hardisty and Rishad Habib in Harvard Business Review's analysis of the green consumer gap, about 2 in 3 consumers said they wanted to buy from brands that advocate sustainability, and only about 1 in 4 actually did.

Customers choose reusable cups and healthier menu items when the ordering screen or counter makes the better option the default and puts a small, visible cost on the wasteful one. Appeals to environmental or health values on their own move far fewer people. The Decision Lab is a behavioral science consultancy that redesigns these choice environments for food service, quick-service restaurant (QSR), beverage and consumer packaged goods (CPG) brands. Our clients have included one of the world's largest coffee chains, one of America's largest fast-food chains, a South African beverage company and a major European airline, and the five principles below come from that work and from the published field experiments we draw on to design it.

1. Make the better option the default. People tend to keep whatever comes pre-selected, whether that is cutlery in a delivery order or fries in a kids' meal.
2. Charge for the wasteful option instead of rewarding the better one. A small fee on a disposable item changes behavior far more than a discount of the same size on the reusable one.
3. Give the better option more of the menu. Adding healthier or plant-based options, and placing them where people decide, shifts sales without removing anything.
4. Fix the systems and scripts staff use. Point-of-sale screens and counter scripts often steer customers away from options the company already offers.
5. Make the better choice rewarding at the moment it is made. Points or playful feedback at checkout give people a reason to pick it on this visit.

The Decision Lab helps QSR, beverage, food delivery, retail and CPG teams redesign the moments where customers choose: ordering flows, menus, packaging, loyalty programs, store layouts and the scripts frontline staff follow. We are most useful when a brand needs to:

• Increase reusable cup and refill adoption without slowing service or annoying customers
• Make healthier menu choices easier while keeping every option available
• Test a packaging or product change before a costly rollout
• Run behavioral experiments that go beyond surveys and concept tests
• Turn findings into changes to point-of-sale screens, apps, stores and staff training

A typical engagement combines behavioral diagnosis, experiment design, simulated shopping environments, pilots and implementation support. Examples are collected in our behavioral science practice for consumer brands.

For each use case, we start with one intervention that has evidence behind it and set guardrail metrics before launch, so a change that hurts sales or service shows up early.

Most CPG teams already run surveys and concept tests, and many run in-store trials. The Decision Lab adds four methods that observe or simulate real decisions before a product or pack reaches market:

1. High-fidelity simulated shopping environments
2. Simulation-first validation before production investment
3. AI-simulated respondents for early idea screening, with human studies for final validation
4. Research with frontline retail staff and category buyers

These methods measure what consumers pick when they face the choice, and treat what they say they prefer as a hypothesis to test.

Evidence: TDL client project, published case study.

For Pinterest, we built high-fidelity working replicas of Pinterest and two competing platforms from a pool of about 7,000 images, personalized to each participant's shopping category and style. Across 2,400 simulated shopping sessions, we ran five behavioral experiments that tracked measures such as save rates and time to purchase. The experiments produced 29 measured strengths, each based on observed behavior, that Pinterest now uses with its advertisers. The same method can test a new pack or product page before it goes live. See our Pinterest shopper decision journey case study.

What it adds over a standard concept test: shoppers make choices in a working replica of the real platform, so the measure is what they save and buy, where a concept test measures a rating.

Evidence: TDL client project, published case study.

With a global consumer electronics company, we tested every target behavior for a new feature through experiments before any production code was written. Design time fell 75%, even with research steps added, and the team's self-reported confidence in its design decisions doubled from the start of the engagement to the end. The client's team now owns the process, as described in our simulation-first validation case study.

What it adds over post-launch testing: design failures surface before development money is spent, when changing course is still cheap.

Evidence: published working paper.

James Brand, Ayelet Israeli and Donald Ngwe compared willingness-to-pay estimates from large language models with estimates from human studies. The model estimates were often inaccurate, and in some cases wrong in direction, though they improved when the model was tuned on earlier survey data from the same category. Our Artificial Populations platform runs AI focus groups in minutes, and we use simulated respondents the way that research suggests: to screen out weak ideas early, before spending on human studies and field tests.

What it adds over a survey panel: ideas can be screened in minutes, before a panel is fielded, though the panel still makes the call.

Evidence: TDL client project, published case study.

Consumer research usually stops at the consumer. In our fast-food kids' meal project (covered below), the decisive insight came from restaurant crew, and the client's director of consumer insights told us collecting data from crew would never have occurred to them. For CPG brands, the equivalent people are store staff, merchandisers and the category buyers who decide what gets shelf space.

What it adds over in-store research: staff watch the moment of choice every shift, and they know which prompts and screens push customers one way.

In our work with one of the world's largest international coffee chains, information was rarely the missing piece, and the reasons people stuck with paper cups were mostly practical and social: carrying a cup is a burden for people on the move, and being the only person at the counter with one can feel odd.

Evidence: published field experiment.

The strongest published evidence points to price framing. Wouter Poortinga and Louise Whitaker ran a field experiment on reusable coffee cups across 12 university and business cafés. At the café that combined environmental messaging, free reusable cups, reusable cups for sale and a 25p charge on disposables, the share of hot drinks served in reusable cups rose from about 1 in 20 before the changes to about 1 in 6 after. Across the sites, a discount for bringing a reusable cup made no detectable difference, while a charge on disposable cups raised reusable use.

Evidence: published natural experiment with comparison stores.

Tatiana Homonoff found the same pattern with shopping bags. After Montgomery County, Maryland, introduced a 5-cent tax on disposable bags, the share of shoppers using at least one disposable bag fell by 42 percentage points compared with nearby stores where no tax applied. A 5-cent bonus for bringing a reusable bag, offered by stores before the tax, had virtually no effect. People react more strongly to losing a small amount than to gaining the same amount, and we suspect deposit and cup-return schemes work best when the deposit is framed as money the customer loses by not returning, rather than as a refund they might collect.

1. Check the operations first. Reusable cups need somewhere to be washed or returned, and staff need a fast way to fill a customer's own cup without slowing the queue.
2. Make reusable cups easy to get at the counter, for sale or on loan, and default to a reusable cup for dine-in orders wherever the store can wash it.
3. Add a small, visible charge on disposable cups instead of a discount for bringing a reusable one, and show it on the menu board.
4. Run it in pilot stores against comparison stores, tracking the share of drinks served in reusable cups alongside queue time and complaints.

Evidence: TDL client project; outcome figures are client-reported.

We reviewed more than 100 studies on sustainable purchasing and habit change. We then ran three ideation workshops with the client team and built an idea book of more than 115 behavioral levers for its cafés and its mobile ordering app. We ranked those by expected impact against cost and narrowed them to the 39 highest-potential interventions in the final playbook, alongside an interactive blueprint of what drives and blocks reusable cup use that anyone in the company can explore. The client has since piloted some of the playbook interventions and reports hundreds of thousands of cups diverted from landfill. Those are the client's own figures from pilots without a published comparison group, so we treat them as a sign of scale rather than a measured effect. The full project is in our reusable cup behavior change case study.

Evidence: published study comparing changed and unchanged cities.

Delivery apps are full of defaults, and most were probably set to speed up ordering with little thought for waste. Guojun He, Yuhang Pan, Albert Park, Yasuyuki Sawada and Elaine Tan studied Alibaba's Eleme platform after it switched the cutlery default to "no cutlery" and began awarding green points for cutlery-free orders in three of China's largest cities. Compared with seven cities where the app did not change, the share of orders placed without cutlery rose from about 1 in 30 to about 1 in 4, and the effect lasted. The researchers found no drop in how much customers used the platform. Because the default and the points launched together, the study cannot separate how much each contributed.

When we look at a delivery or ordering app, we start by listing everything a customer receives unless they act, because each one is a candidate default. We also set guardrails before launch, so a change that annoys customers shows up early. A green default should take one tap to undo. If customers cannot easily get what they want, the change erodes trust in the brand and invites a backlash.

• Cutlery and napkins: default to none, with a one-tap opt-in
• Condiment packets: default to none, or to one for dishes that need it
• Bags: ask before adding one to pickup orders
• Item ranking: test where plant-based or lower-waste dishes appear in menus and search results
• Green rewards: points or a running tally of what the customer has saved, shown at checkout
• Guardrails: order completion and support contacts, tracked from the first day

Only the cutlery default has strong published evidence behind it. The other items are candidates we would test the same way, against a comparison group of customers or cities.

For QSR and beverage brands, we work on the points where an order gets made: menu boards, kiosks, apps, drive-thru scripts and point-of-sale screens. Our fast-food work started with an audit of every ordering channel and ended with changes to screens, staff training and a reward aimed at children. Our coffee chain work covered its cafés and its mobile ordering app. We pair the published evidence on defaults and menu design with tests in the client's own channels, and hand the method to the client's team so it can run the next round. Beverage brands face the same mechanics on cup size and refills.

Evidence: meta-analysis of field experiments.

Romain Cadario and Pierre Chandon pooled 299 results from 96 field experiments on healthy eating nudges. Nudges that change what is easy to do, such as making the healthy option more convenient or changing portion sizes, cut daily intake by about 209 calories on average, roughly three times the 64-calorie reduction from nudges that only give information, such as calorie labels. Effects were also larger in restaurants and cafeterias than in grocery stores.

Evidence: published observational study and field experiment.

How much of the menu the better option occupies matters too. Emma Garnett and colleagues analyzed 94,644 meals sold in Cambridge college cafeterias. Doubling the vegetarian share of the menu, from 1 in 4 options to 2 in 4, raised the vegetarian share of sales by about 15 percentage points in two cafeterias and by about 8 points in an experiment in a third. Total sales stayed about the same, and the largest shifts came from diners who usually ate the most meat.

Evidence: industry-funded observational study.

Defaults work on menus as they do in apps. When Walt Disney World made fruit or vegetable sides and low-fat milk, water or juice the default in kids' meals across its 145 restaurants, about half of buyers kept the healthy side and about 2 in 3 kept the healthier drink, with fries and soda still available on request. Disney funded the study and withheld some baseline sales data, so it shows how many customers accepted the default rather than a measured change from before.

Evidence: proprietary test results.

• Client: one of America's largest fast-food chains
• Problem: healthier kids' meal sides were on the menu, and few families chose them. Our survey of hundreds of US parents showed fast food was treated as a treat, with children usually steering the order.
• What we found: in interviews with dozens of restaurant crew members, we learned that fruit was officially the default side, but the point-of-sale screen forced crew to ask customers to choose between fries and fruit. Many crew members did not know fruit was supposed to be the default.
• What we changed: the point-of-sale screen, the staff training protocol and some customer-facing kiosk screens, plus a big red button children could press after choosing a healthier meal, that played a congratulatory message
• Result: up to 40% more healthy kids' meal orders with the updated systems, and 20% more customers choosing the healthier meal with the button, in testing
• Caveat: the published case study does not report sample sizes or comparison groups, so these figures should not be read as an estimate for every restaurant

The full project is in our healthy kids' meal choice architecture case study.

1. Audit what the point-of-sale screen and kiosk ask, and remove prompts that push customers away from the default.
2. Make the healthier side and drink the default in bundles, with swaps on request at the same price.
3. Give healthier and plant-based dishes a larger share of the menu, placed where customers decide.
4. Train crew on what the default is, since many crew members in our fast-food work did not know.
5. Give children a reason to want the healthier option at the moment they order.
6. Track total sales and meal mix against comparison stores.

Packaging changes behavior mostly through what it makes easy at the moment of purchase and use, so we design and test it with the same choice architecture methods we use on menus and apps. These are the six levers we work through with CPG teams:

1. Default quantity and replenishment cadence. The amount that arrives unless the customer changes it, in a subscription, bundle or multipack, tends to stick the way defaults do elsewhere.
2. Portion size and consumption pacing. Changes to portion and serving size were among the most effective nudge types in Cadario and Chandon's meta-analysis, and single-serve or resealable formats can change how fast a product is used.
3. Refill, return and deposit design. Following the loss-framing evidence on bags and cups, we frame deposits as money lost by not returning, and put the return point somewhere as easy to reach as the shelf.
4. Disposal instructions at the moment of use. Sorting or return instructions belong on the part of the pack that gets thrown away, where people read them as they empty it.
5. Visual salience and shelf comparison. How a sustainable or healthier variant looks next to its neighbors on the shelf or product page affects whether shoppers consider it at all.
6. Friction in opening, storing or returning. Every extra step, such as rinsing a container or keeping it until the next trip, is a cost the customer pays, so a good redesign often removes steps before it adds a message.

On-pack sustainability claims run straight into the gap from the opening statistic. Shoppers often say a recyclable or refillable format matters to them, and then buy the familiar one. We test packaging changes against what people pick or use in a realistic setting, including the simulated shopping environments described above, and treat stated preference as a hypothesis to check.

Evidence: proprietary pilot.

Sometimes the product that needs redesigning is a service around the product. With Distell, a South African beverage company, we built a digital program to reduce binge drinking, starting from fieldwork with more than 600 people in Soweto. The social side of drinking came up again and again in that fieldwork, and the program was designed with that in mind. Participants in the initial pilot cut their drinking by about two thirds by the end of testing. That figure comes from a proprietary pilot without a published control group, and we report it as early evidence rather than a general effect. The details are in our Distell binge drinking case study.

Choice environments for sustainability in retail are designed by in-house insights teams, public-sector behavioral units, academic researchers and specialist behavioral science consultancies. The Decision Lab is a behavioral science consultancy that helps retail, QSR, beverage and CPG brands redesign menus, product defaults, packaging, digital ordering and frontline processes to make sustainable and healthy choices easier.

We also run climate and sustainability projects with public and nonprofit partners, and our more than 60 published case studies range from field experiments with the World Bank to the retail projects above. A typical retail or food service project with us runs in four stages:

1. Diagnose where the decision happens. We audit the ordering or shopping journey and talk to customers and frontline staff.
2. Build a long list from evidence and narrow it with the client. For the coffee chain, that meant more than 115 levers ranked down to 39.
3. Test before rollout, using simulated shopping environments or pilot stores measured against comparison stores.
4. Hand over the method along with the recommendations, so the client's team can design the next round without us.

We are probably not the right fit for a brand that wants a messaging campaign and is not willing to change defaults or prices. The evidence in this guide says those changes are where most of the effect comes from.

If your team wants to increase reusable or refill adoption, reduce single-use waste, make healthy menu choices easier or test a packaging change before rollout, we can help find the decision points with the most potential and validate interventions before you scale them. Browse our behavioral science case studies for consumer brands or talk to our team about a behavioral design and experimentation project.

What is the most effective way to get customers to use reusable cups?

A small charge on disposable cups, combined with reusable cups that are easy to get on the spot, has the strongest field evidence. In one café experiment, that combination plus messaging raised reusable cup use from about 1 in 20 drinks to about 1 in 6. Discounts for bringing your own cup showed no detectable effect.

Do sustainability and health nudges hurt sales?

In the studies in this guide, they did not. Doubling vegetarian options in Cambridge cafeterias left total meal sales roughly unchanged, and Eleme's no-cutlery default did not reduce how much customers used the platform. Brands should still track order rates and complaints during any pilot, because those results come from specific settings and may not carry over.

What is choice architecture in retail?

Choice architecture is the design of how options are presented at the moment a customer decides: what is pre-selected, what appears first, what costs extra and how much space each option gets. In retail and food service, it covers menus, shelves, packaging, apps and the scripts staff follow. Our choice architecture reference guide covers the concept in depth.

Is it ethical to use defaults to change what customers buy?

We consider a default ethical when customers can switch back easily at no extra cost and the change serves their own stated goals or a clear public good. Defaults that hide options or make opting out tedious cross into manipulation, and they tend to damage trust in the brand once customers notice.

Can AI-simulated customers replace field experiments for CPG brands?

Not yet. Published comparisons show language model estimates of willingness to pay are sometimes close to human results and sometimes wrong in direction. We suspect they are weakest on sustainability questions in particular, because they learn from what people write about green products, the same stated intentions the opening survey showed buyers often fail to follow.