Good Earth

augmented reality app

Teammate: Aaron Faucher

My role: secondary research | experience prototyping | user testing


Food production is a major cause of environmental problems, and some food products are dramatically better for the environment than others. How might emerging technologies promote eco-friendly consumer behavior at the grocery store?


Good Earth is an AR app that empowers people to make eco-friendly purchasing decisions at the grocery store. It leads you to the items on your shopping list, and it provides you with the information you need to make informed decisions on the go.


Concern about environmental issues doesn't automatically translate into eco-friendly behavior.

We chose to examine the effectiveness of a number of psychological theories in persuading people to engage in environment-friendly behavior. Examples include the Theory of Planned Behavior and Cialdini's Norm of Social Proof.

We also drew inspiration from a number of precedents that are successful at persuading people to change their behavior in favor of environmental and social causes:

Imperfect Produce is a startup that sells ugly, misshapen vegetables by creating an aura of desirability around them.

Buy Black is a Chrome extension that promotes black-owned businesses by offering alternative suggestions to online shoppers.

Seafood Watch is a program that makes consumer guides for sustainable seafood.


People are limited in their motivation and ability to engage with complex issues. There is a need to simplify.

Our ideation process began with rapid paper prototyping. We utilized the Crazy Eights method, popularized by Google Ventures, to rapidly sketch 18 intervention concepts for consideration. These concepts ranged from immersive walkthroughs of a grocery store led by 3D AI agents, to behavioral dashboards showing behavioral trends over time. Then, we constructed a decision grid to evaluate these ideas based on the following factors:

  • Is the solution generalizable to other contexts, or is it overfit to this scenario?
  • Does it incorporate an innovative use of an emerging technology?
  • Do our observations so far give us a gut feel that this intervention may succeed?

We eventually decided to move ahead with an augmented-reality-mediated intervention. Here's a visualization of the idea:

A just-in-time intervention in which contextual triggers are overlaid on top on different products.

Why Augmented Reality?

Contextual. Non-invasive. Handsfree.

Our chosen setting, the grocery store, presented a number of unique challenges. Our solution needed to be mobile so that it could accompany shoppers as they moved about the store. Our guerrilla research revealed shoppers at the grocery store were in a perpetual hurry, so retrieving and using a mobile phone would have been too laborious. Our solution needed to be hands-free so it would integrate seamlessly with the shopping experience. This is why we chose to go with augmented reality on a head-mounted display, a device we expect will be more prevalent in the near future.


We explored the influence of information, convenience, friends' behavior, explicit nudges, and appeals to morality.

We decided to move forward with the concept of an augmented reality system that would overlay persuasive UI elements in the user’s field of view in the context of their decision. Here are three examples of our early prototypes. Each shows a progression of information that a user might see in augmented reality:

User Testing

We got thrown out of Whole Foods for user-testing our ideas in the frozen foods aisle.

We conducted a brief first round of user tests with three graduate students at Carnegie Mellon. Then, we traveled onsite to a local high-end grocery store in Pittsburgh’s East Liberty neighborhood. We conducted user tests in the contexts of users making a purchase decision, showing them the prototypes on an iPhone screen. Here are some of the key insights from this round of user testing.

  • Grocery shopping is often done in a hurry.
  • Environmental concerns are not always top-of-mind.
  • Pictures are more effective than words.
  • People are strongly influenced by what their friends are doing.


In our final iterations, we compared the influence of eco-friendly information, friends' purchasing behavior, and explicit nudges.

We incorporated the above insights into a second round of paper prototypes. We narrowed the scope of our prototypes by employing data about friends' preferences, data about envionmental impact, and a menu of options (best/ok/avoid). Our prototypes tested combinations of these elements under different situations, such as price sensitivity and preference, in order to better ascertain which elements were most influential (or least effective) in nudging user buying choices. Here are four examples of our prototypes:

Final Design

People like information, soft nudges, and the freedom to make their own choices.

The final design included the following features:

  • Just-in-Time AR Display. Our final prototype was built on the Microsoft HoloLens. Grocery store ambient noise was played through the headset to increase the user’s immersion. AR overlays depicting UI elements were position-locked to specific products, which fully simulated the experience of viewing digital information overlayed onto a physical product.
  • Best / Ok / Avoid Tags. Based on our research of just-in-time persuasion strategies, we augmented the users’ view with an array of choices, rather than just a single ‘Recommended’ choice. Ultimately we imitated the language of the Monterey Bay Aquarium’s Seafood Watch.
  • Only-Positive Social Influence. Due to the complex effects of social influence in our final user study, we decided that Good Earth would selectively display friends’ purchase behaviors only when these purchases were on the ‘positive’ end of the spectrum.

Here are some visuals to illustrate the final design:

At the grocery store, Good Earth guides users to the aisle where the items on their grocery list are located. For each item, the app uses AR to present alternative options as well as eco-friendly recommendations.

Users are able to interact with the data in order to find out more about Good Earth's recommendations.

If I had more time...

...I would test our designs in context, i.e., with real customers in an actual grocery store. Given the scope of our project, we were able to conduct initial research at the grocery store, but we were not able to take our designs back to our users. If we were to take this project further, that would be a crucial next step.