How might we help people sustain and improve a diet healthy for them?
Instacart Health is a hypothetical new feature for Instacart, which helps users shop and eat healthier by analyzing their grocery consumption data.
I was just one when my father, a newly minted physician, passed away from a mysterious cancer. My brother was two, and my sister seven. My mother, for months, was inconsolable. For decades, I sought for ways to prevent the same fate from happening to me and my own future family. When I heard about diet’s connection to a litany of chronic diseases, I refrained from red meat, cut out dairy, boycotted family trips to fast food joints, replaced dessert with fruit, and adopted a weekday vegetarian diet. Yet even with all these adjustments, I still flout them in the name of competing priorities.
Perfect compliance to rigid rules, I realized, is not a sustainable or even necessary approach to diet control. Imperfect yet sufficient compliance, however, is difficult to measure and thus difficult to respect. Good intentions can be slippery slopes. From conversations with others (and almost daily news reports on the state of our nation's health), I have realized that this is a common and troublesome predicament. What if there were a way we can set more flexible boundaries, which we could still measure and hold ourselves accountable to?
The Original Idea
The products currently available in tackling this problem seem to fall into two categories: calorie counting and food journaling. Both types of products require frequent, painstaking maintenance that is also unsustainable. I came up with an idea that revolved around a much more feasible task of scanning grocery receipts.
A mobile app, called Nutriceipt, will parse the data on grocery receipts into all sorts of ratio-based food reports. Some will be related to food group (grains vs. vegetables vs. meats vs. fruits), others will be related to attributes of the food (organic vs. non-organic, local vs. non-local, seasonal vs. non-seasonal, whole vs. processed). By quantifying our diet in this way, we are able to measure our nutrition through the relative consumption of its constituent parts, a much more forgiving and flexible standard.
I spent months meeting several times with a mentor to refine the idea, while researching nutrition in Mark Bittman's VB6, T. Colin Campbell's The China Study, Michael Pollan's In Defense of Food, and Tom Rath's Eat Move Sleep. I studied the environmental impact of food in Dan Barber's The Third Plate and Michael Pollan's The Omnivore's Dilemma. I partnered with a progressive nutritionist. I also gathered great feedback in-person from Michael Pollan and Edmond Sanctis, the founder of health food company Sahale Snacks.
I originally planned to create a simple, low-tech, manual prototype, sidestepping at this stage any time-intensive software programming. I had already recruited over twenty beta testers, for whom I would manually perform all the features of the app for approximately two months, a length of time potentially sufficient to capture holistic dietary habits. I planned to physically collect the receipts myself, analyze the data individually, and send back digital reports weekly.
I then would have conducted interviews throughout the experiment to verify the following major assumption: the value of the information in the dietary report is worth the hassle of submitting the receipt.* For the interview, we would observe beta testers reviewing one of their dietary reports, as they think out loud. We would then ask multiple layers of questions to understand any holes in our concept and to validate the many other assumptions that hold this idea together. Afterwards, we would visit the drawing board to think of as many solutions as possible to any uncovered problems, then choose the most viable solutions, and then iterate onward.
After some deep contemplation, and a few delightful experiences with Instacart, I realized that I cared more that this idea was realized than that I was the one to make it all happen. Having considered long and hard about two of the largest assumptions in the idea of Nutriceipt (the finicky scanning of many very long grocery receipts and how it would become a sustainable enterprise), I realized that both problems would be completely eliminated in the world of Instacart.
Instacart had already digitized grocery receipts for many major supermarkets in the country, with information-rich databases of nutrition information and other relevant data. The nutrition analytics would appear automatically and hassle-free on Instacart, with no work from the user, just like Mint and the automatic pull of credit card data. By adding the concept of Nutriceipt to Instacart, it could level up the experience of anyone who experienced the data analytics of Nutriciept and also the experience of every customer of Instacart.
The Business Case for Instacart
Instacart plans on building “the best way for people anywhere in the world to shop for groceries.” With 18 metropolitan areas under its belt, Instacart has already in many regions delivered on this vision, but it has been working sedulously to expand their target market to include the less affluent. With more budget-friendly initiatives like the discount-oriented Instacart Plus program, Instacart hopes to transform the way all levels of people shop for groceries. With Plus only available in San Francisco though, Instacart’s vision of “people anywhere” is still a work in progress.
As a budget-oriented father of a young family (and its primary grocery shopper and chef), I tried out Instacart several times before realizing that, after tipping what I felt was an appropriate amount, it wasn’t a viable financial option for me. I’ve spoken with several people in the Seattle metropolitan area who experienced the same results. It was certainly delightful, but it felt like an unnecessary luxury, an expense that young families (saving for mortgages and children) would have a hard time justifying.
Instacart’s mission is to “give…customers back their invaluable time so they can spend it doing the things they love with the people they love.” What if Instacart was more than just a grocery shopping and delivery service? What if it also provided comprehensive health analytics that could help customers shop better, for themselves and for the environment? What if Instacart not only gave their customers back their invaluable time, but also their invaluable health?
So I spent two weeks designing and repackaging the product as a new feature for Instacart.
Say hello to Instacart Health.
This could flip the switch for many customers, from viewing Instacart as a luxury, time-saving service to an indispensable service that could help improve their health. Anyone who’s had an unfortunate year of poor health can attest (often vehemently) to the egregiously high expense of health care. Investing in health is one investment that is likely to pay off financially in the long run, a value proposition any budget-oriented individual can get behind. And with metrics on environmental impact, Instacart Health solidifies itself as an aspirational program that aligns with customer’s deeply held values on supply chain transparency, carbon footprint reduction, and voting with dollars.
As an admittedly under-qualified outsider, I contend that Instacart is in a prime business position to tackle this strategy. It already has the relevant data. It has already scored dozens of partnerships with the largest supermarkets across the country, which means it has the greatest variety of products out of any of its grocery-delivery competitors, and thus more data on everything. It has much of the technology already built to support the infrastructure for this new feature, such as the digitized receipts and old records of customer purchases. Instacart can become the Mint for your “dietary budget” and the Fitbit for your “diet regimen,” at a time when people are taking back the responsibility to eat healthier.
One of the greatest business justifications for implementing Instacart Health is that in order to output the most relevant data, it encourages (or requires that) users shop through Instacart for all their groceries in order to track all their purchases. It inspires and necessitates pure customer loyalty, and long-term customer loyalty. It satisfies one of the primary variables in Nir Eyal’s theory on how to hook customers, by allowing users to develop a richer and richer nutritional profile through their regular investments in shopping through Instacart. The resultant gradual accumulation of data over time solidifies valuable loyalty to the Instacart service.
Designing the Prototype
Framing the Design Challenge
To begin the two-week design sprint, I first engaged in an exercise of empathy by storyboarding five proto-personas (an approach I learned from the Lean UX series), fictional characters who I imagine would love a service like Instacart Health. These stories served as a canvas of inspiration to design the features throughout the service.
Idea Impact Analysis
I then analyzed the viability-desirability-feasibility as well as the revolutionary-evolutionary-incremental features of the app. These methods were gleaned from the Field Guide to Human-Centered Design (IDEO). I analyzed how Instacart would solve the problems intrinsic to Nutriceipt, while Nutriceipt would add a revolutionary feature to Instacart, bringing new loyal customers.
One of the important insights I had early on in this overall project is realizing that the Nutriceipt app idea wasn’t necessarily business viable. How will it become a sustainable business like Mint, without the incredibly high affiliate fees that Mint gathered from credit card advertising? Instacart, on the other hand, already has a working business model, and so, yes, tacking on a new feature that would utilize already existing technology would not take a massive amount of new resources. It would thus certainly be viable.
I wondered if a standalone app’s data analysis was desirable enough that people would save and scan their often crumpled and sometimes extremely long receipts weekly or biweekly. It’s a clunky endeavor. I wasn’t sure if the experience would be seamless enough to activate desirability. This is important if you want to change the weekly habits of very busy young urban professionals, as well as young families from all levels of income. Instacart, however, already cracked the code for making grocery shopping and delivery extremely seamless and pleasant, and the fact that the analytics could all automatically materialize in user’s profile makes it seamless and much more desirable.
And even then, if people flocked to the idea of Nutriceipt, could our systems actually parse the data and know how much of any one particular item that someone purchased? If someone paid $5 for beef, but the receipt didn’t say how much it weighed, or whether it was on sale, or the exact type of beef, how could we give accurate and thus relevant data? How can we match a line on a receipt to a particular item, when every supermarket has different POS and database systems that have titled different products with different names? Instacart takes away all of these roadblocks, with complex systems in place that document food purchases with extreme accuracy, without the need for scanning for a receipt (an extra kinemetric step).
I humbly believe that Nutriceipt's new set of innovative features would be revolutionary for the world, prompting a surge of new loyal customers for Instacart in the form of Instacart Health. This is not a simple adjustment to the delivery process, or to the inventory system. It is a fundamentally novel offering not yet seen in the world.
Establishing Design Principles
I spent a few hours contemplating how I wanted the product to live and interact with its users. As any product is naturally an embodiment of its creators' opinions, I wanted to make sure that we knew which opinions (or rather, principles) reigned king. These would guide the design's iterative process.
General Design Principles:
We recognize that every person and body is different, and has different needs.
We maintain that many health principles, nevertheless, are universal.
We aim to simplify the user experience, while still maintaining legitimacy.
We aspire to design a tool, not a prescription.
We stay diet-agnostic. Keep to the science.
We amplify the signal above the noise.
We seek to create a feature that is insightful, informative, life-changing, and world-changing.
We shoot for progress, not perfection.
Aesthetic Design Principles:
Maintain Instacart brand standards with simple, flat vectors, Open Sans font, curated color scheme, and predesigned layouts.
Subtle, yet perceptible color changes in infographics for a clean, crisp aesthetic that is non-taxing on the brain.
Drafting the Prototype
I designed the layout from wireframes to final product on Figma, and used Marvel for the hands-on prototype seen below and Principle for the animated prototype (further down the page). I created several iterations before landing on the one you see here. Take a little test drive below! (For a full screen and mobile-friendly version, click here.)
One of the greatest difficulties of designing the experience of viewing this health data is to strike the right balance between presenting enough data to be engaging, relevant, and valuable (in a 10x kind of way), but not so much that it becomes overwhelming and excessive. Divided into categories, just like the Instacart aisles, the data can be previewed quickly in one vertical scrolling sweep, or viewed more holistically in its dedicated pages separated by category.
I first designed the landing page above the horizon line for Instacart Health. I knew that if Instacart Health were to succeed as part of the larger company, a consistent visual brand would need to be presented to the customer. I created a miniature, unofficial style guide of Instacart by collecting the font sizes, colors, spacing details, and other assets with a number or browser plug-ins and Chrome’s developer tools. I mimicked the Deals page header by implementing subtle, linear shadows on colored vector images of whole foods that symbolized vitality and freshness. I partnered the health page with an icon that had a similar aesthetic to the ones already designed on the menu bar, and featured a clipboard — representing a medical record — and an apple, a symbol of preventative health. Just to the right of the Health page label on the navigation bar, a purple “new” badge caught the attention of users to check it out.
On arrival, the Health page presents to users a series of charts and metrics that illuminate their consumption habits. I chose the humble pie charts and bar charts to both simplify and beautify the data analysis, in a way that directed attention to the data, not the medium. I designed by the principle of form following function, championing the user’s needs first by increasing the signal-to-noise ratio. I chose chart types based on the attributes of each category, depending on the ideal way of visualizing its data.
I then matched certain types of food with symbolic colors (using conventions where possible), with palettes extracted from aesthetically proven illustrations and designs I curated on Pinterest. The color for the plant category, for example, is green, while that for the animal category is brown. Similar associations were considered for all the colors chosen throughout the samples shown here. This was done to make the charts' individual components more readable and memorable. I layered numbers on top of the chart to add extra precision for users in understanding their data. And because the entire vision of the program is to improve health, I incorporated metrics on progress (increase % and decrease % from last time period) as well.
I wanted the feel of the page to be as diet-strategy-agnostic as possible. I knew it was impossible to be completely hands off in this department, since merely presenting data can already inherently imply certain biases. Nevertheless, I figured that Instacart has no interest in entering the war of what diet is “best,” and would rather help all individuals achieve their dietary goals, no matter what they were. Thus, the data that is presented has been stripped of as many political and philosophical opinions on how to eat as possible, while still relying on the latest scientific research as to which data points actually matter in optimizing for human health. In some cases, to make the data more relevant to users, I’ve included metrics on competing popular diets, like vegetarianism and paleo.
Copy is spare on these pages, to make it more pleasant to view and quicker to review. Nevertheless, explanations are still needed for much of the data. Since many of the proposed categories might be new to many users (and everyone seems to have a different definition of certain categories these days, like local, or seasonal), I still thought that it would be vital for users to understand how Instacart Health calculates their data. So I added an information icon button (an “i” inside a white circle) next to every chart title, which users can click to get more information about a particular chart. The pop-tip would explain the general definitions and boundaries of the various data categories.
From the pop-tip, users can click the “Source Data” link to see all their data behind the chart. A modal window would pop up into view, with the rest of the page shaded. I vacillated between designing an entirely new webpage or a modal for the data, but settled on a modal so that a user could more easily trace her steps as she navigates throughout the Health section. Having an entirely new webpage (without a set of breadcrumbs — a design I opposed due to the already stacked navigation portion of the site), I thought, would add unnecessary complexity to the navigation experience.
Users can edit the data behind the charts, à la Mint. I considered this feature a vital one, because in order for a user to really become attached to his data, he would need to feel that the numbers were not just precise but accurate. Sometimes, a user might buy food for a party, which does not represent his family’s dietary habits. He could go into the data and delete it. Or, maybe half of the butter was used to bake cookies for a bake sale. He could edit his data to reflect this reality. I designed the editing interface to make it visually clean with conventional affordances. When the user hovers over a particular line item, the edit pen appears, and clicking on the line item reveals editable form fields, to change the date, the name of the item, the weight, or the cost.
The first iteration of the Instacart Health page design included very few numbers, but I realized that users grasp onto quantifiable data and remember them more easily than a collection of different-sized, colored shapes. Knowing that your groceries are 5 parts plants to 3 parts animals, or 62% vegan, is a much stronger and more precise indicator of your diet's current state. Some numbers were a bit more ambitious, and granted an overall score for a particular category, such as the Sustainability Score in the Environmental Impact subsection, or the Nourishment Score in the Nutrients subsection. These scores were added to help users compile all the data within a subsection into a single, more comprehensible figure (similar to the single “% productive” metric on RescueTime).
Users can dive more deeply into each individual subsection by clicking the View All Data button at the end of the title of each subsection. On click, users travel to a page dedicated to that subsection, with even more charts (especially progress-related) that could be useful in helping them better break down and analyze their diet.
Additionally, I incorporated animation into the product’s design to more fluidly transition between different versions of the same chart. As users view the gradual changes between the different versions, they can notice the differences between the two data sets more easily.
Some other trade-offs I experienced during this design stage:
I realized late in the design stages that the colored pie charts were not accessibility-friendly, particularly to the color-blind (but also to the blind). Many colors could look the same to such users, and it could be difficult to distinguish one food category from another. Extra consideration must be taken to accommodate for both the color-blind and blind, to make the charts more readable.
The relative advantages of pie charts over bar charts and bar charts over pie charts were often difficult to pinpoint. They depict most of the information equally well. For a large number of categories, bar charts would help users compare different values more precisely, and I could easily incorporate two different types of data values (percentages and pounds, for example). Pie charts, however, leave enough room for progress metrics, and also happen to resemble the garden variety dinner plate, and nutrition experts have advised people everywhere to adjust their diet by changing the proportions of different food categories on their plate. By using pie charts, the site would build on top of a widely popular federal program.
Adding the green and red colors to the progress metrics was a difficult decision. The page was already loaded with color, and I thought adding more color would be visually distracting. Nevertheless, as one of the most (if not the most) important metrics on the page, I decided that it needed extra emphasis, and infused it with color. Even still, I believe either more (or more significant) progress metrics need to be added to the screen, to provide even greater context for all the stats.
It was an exercise in self-control to limit the number of charts and metrics on the home Health Page. I included so many in the first place to accommodate the many different kinds of dietary goals that Americans have. Even one person may have up to several dozen boundaries, as demonstrated by the 64 guidelines in Michael Pollan’s Food Rules. For this two-week design sprint, I did not have the time to create every subsection page, and curated the most relevant metrics and charts for the front page based on months of prior research.
A Side Feature, in the Mobile App
Lastly, initial feedback on the idea sparked the idea of a competitive social experience, similar to Fitbit’s friendly competition between users in number of steps taken in a day/week. Below is a simple prototype of what that could look like. Users could compete with one another based on a variety of different metrics, whether a plant vs. animal ratio, or a cost vs. carbon footprint ratio. To make the ratios more comparable, I converted them to a single number (a ratio of 4:1 would be 4, a ratio of 12:5 would be 2.4). In retrospect, using percentages may have been more comprehensible.
A few features that have not been fully fleshed out in the design, or are possibly down the pike, are listed here:
Custom charts, such as charts for those interested in reducing gluten or overall carbohydrate consumption, could be helpful for those with more unique needs. Users would be able to compare any number of data sets together, or track a single data set more closely, and choose which visual representation (pie, bar, etc.) to display their data. Users could also customize the data charts on their home Health Page, retiring some into their respective subsection pages, while elevating others onto the home page. This further personalization would make the data much more relevant to the user’s needs, although it could complicate the beginner or casual user’s experience.
Connecting to the APIs of 23andMe and recommending certain types of food based on your genetic profile, to avoid certain diseases more common to those with your specific genes, or to actualize your physical potential.
Add a “budgeting” or “goals” element to the Health page, so that users can also track how close they are to their goals. Notifications can be added to congratulate users who are on track to staying within their dietary budget.
Featuring ideal diets from non-profit organizations or medical associations (e.g. American Heart Association, American Cancer Society) that focus on diets particular to survivors or battlers of certain diseases and symptoms.
Many people buy in bulk, especially Costco products, which can spike your data. Buying even one bottle of olive oil can increase weekly calorie/fat metrics dramatically, but the impact on the data will peter out over time in monthly and annual reports. Nevertheless, a feature could be added to split a current food item’s nutritional profile into a certain number of weeks or months, to make the data more relevant on a week-to-week basis.
What’s Next, and Assumptions
There is a number of related assumptions Instacart would have to test:
The dietary report for an entire household produces sufficiently meaningful data for families
What we buy is more or less equivalent to what we eat
The dietary report for an entire household (vs. for one individual) is sufficiently meaningful
Becoming aware of and adjusting the ratios of different foods to one another in our diet is a desirable approach to diet change
Data on receipts (plus some background research) is sufficient to establish meaningful ratios
Though only accounting for groceries (as opposed to also tracking meals eaten out) would not be all-encompassing, for many individuals and families, this may still be enough
That Instacart either has access to or can gain access to all the data that can furnish these charts
That for those data points that are estimates (like water usage or miles traveled), it will still be effective and relevant
That the particular charts that were chosen are the charts that spark the curiosity of users
And – the ultimate assumption – that this feature will bring the conversion of legions of new loyal Instacart users.
User testing these assumptions would be important for the next step of Instacart Health’s evolution. The next step I took was some guerrilla on-the-street usability testing to explore whether this new service has the potential to convert ambivalent customers into loyal customers. The results were promising: 18 of 25 users said they would likely or absolutely use the service if it were paired with their current approach to grocery shopping. See the full research below.
Instacart Health would not just be a new direction for Instacart, it could be a new direction for American health. It is a win-win-win situation, with the company winning more loyal customers, customers winning better knowledge of their consumption habits (and greater health), and the country winning lower healthcare costs, and a reduction of the environmental impact of food consumption.
With Instacart’s prime position as middleman between consumers and almost every grocery provider, it can provide data to the user that no single entity or even delivery competitor can. The flexibility and comprehensive scope of its reach is peerless and defensible. By developing and launching the Instacart Health feature, Instacart will further establish itself as the best way to shop for groceries, for people anywhere. And it will save people not only time, but also their health, their money, their well-being, and ultimately, their happiness.
Disclaimer: I am not an employee of Instacart and this does not represent the official strategic direction of Instacart.
Asset Credits: Milky - Digital innovation (information icon), Sagar Unagar (progress metric arrow), Remy Medard (edit pen), Luca Reghellin (delete x), dw (source data icon), FreeVectorMaps.com (Map of World with Countries - Single Color & Map of United States of America), Alina Tong via Stocksy (Nutriceipt Mockup Background Photo), Freepik, and Flaticon