Research

I started by jotting down notes and ideas. This is a familiar problem to me because my husband and I usually do not have time or energy to make weekday dinners. From New York to San Francisco, we have tried many alternatives in pursuit of a great experience and a good meal. So it is with either bias or an informed perspective that I tackled this challenge.

Notes
Notes

Competitors

"Research" suggests that consumers are overwhelmed with choice when ordering food online. How do current products in the market deal with this problem? What are the ways in which they reduce the amount of decisions that a customer must make and the number of options from which they must choose? I listed current competitors in the mobile on-demand food space and then studied their products closely.

From left to right, the products below fall along a continuum between choice and constraint, from Postmates, which provides general delivery service from any restaurant or bricks-and-mortar store, to Bento, which specializes in one to two Asian cuisine meal packages daily.

Traditional Delivery

The more traditional apps offer ordering and delivery from the full menus of local restaurants:

Postmates
Postmates (PM)
GrubHub
GrubHub (GH)
Seamless
Seamless (PM)
Eat24
Eat24/Yelp (EY)
OrderAhead
OrderAhead (OA)

On-Demand Menus

Newer startups offer limited or curated menus:

Caviar/Square
Caviar/Square (CS)
Munchery
Munchery (MC)
SpoonRocket
SpoonRocket (SR)
Sprig
Sprig (SP)
Bento
Bento (BT)

Common features include:

FeaturePMGHSLEYOACSMCSRSPBT
Network of restaurants or central source
Many menu items or limited options
Saves recent/default delivery address
Saves recent filters and search terms
Retains current order/shopping cart
Highlights past orders
Charges flat or low delivery fee
Uses large, stylized photography
Shows ratings and reviews
Changes featured content regularly

Questions

To determine the success of each product and guess at the features that work and those don't, we can take a look at the market share and growth of each product. However, reliable public statistics are difficult to come by. A company may share new signups or daily active users to show growth but not share daily orders (conversions), which would expose weaknesses in their ordering funnel to their competitors.

The strongest evidence would come from qualitative research interviewing and observing people using these products and quantitative testing in a launched product or with prototypes. Qualitative questions to answer might include:

  1. When looking for food to order online, do you order from a restaurant that you've visited in person?
  2. How often do you try an unfamiliar dish or restaurant when ordering food online?
  3. When ordering unfamiliar dishes or from new restaurants, what factors into your decision (description, ratings, reviews, photography)?
  4. How often do you know the cuisine or dish that you want to order when you start your order online?
  5. How do you decide how much food to order?

Quantitative questions might be:

  1. How often do people order from the same restaurants in a given time period?
  2. Do people regularly order the same type of dishes (for example, meat and vegetables or pizza)?
  3. Do restaurants with better food photography get more orders?
  4. What are the most common search terms or parameters that people use? How do those vary with number of orders?
  5. Do people order enough food for one person, two people, or more? How many items are in an average order?

Analyze

Personas

I then considered the questions above from the lens of our experience. When my husband and I order food online, we take the following factors into consideration:

Convenience
The speed of the delivery, the availability of restaurants, the availability of dishes after dietary restrictions and allergies
Price
The per-person cost for an adequate amount of food, the total cost of the meal after taxes, tips, and fees.
Quality
The nutritional value and "healthiness" of the food (or lack thereof), the taste (as measured by popularity of the dish, the cuisine we're in the mood for, or the newness or familiarity of the dish)

Our basic needs don't change much from day to day: quick (delivered within 30 minutes), healthy (low-carb, protein and some vegetables) meals of reasonable portion sizes and prices ($5-15 per person) for two people. Because we are eating at home, we care less about the name of the restaurant (i.e. the source of the food) except as an indicator of food quality. Consistent great taste is very important but we don't want to eat the same thing everyday. We have different dishes and cuisines that we prefer to eat regularly. My husband likes meat sandwiches and I can't go a week without Asian soup and noodles.

The Traditional Experience

We can meet our needs with the traditional shopping cart experience offered by an app such as Eat24. But on a weeknight when our time is short and we are too tired to make many decisions, we often give up on the many-step process. There are simply too many options to consider: $3.00 delivery fee or free delivery but $20 minimum, 3.5 -stars $ Mexican or 4-stars $$ Mediterranean, 2 Nachos with Chicken with Green Salsa or 1 Super Quesadilla with Carne Asada with Red Salsa. With its power search functionality and extensive listing information, Eat24 might be a better tool for exploring the universe of dining options or ordering from restaurants that you regularly frequent.

Select Address
Select Address
Browse Restaurants
Browse Restaurants
Filter/Sort Restaurants
Filter/Sort Restaurants
Browse Menu
Browse Menu
Select Item
Add Item and Choose Options
Add More Items
Add More Items
Order Details
Order Details
Delivery Details
Delivery Details

The Newer Model

We much prefer the limited rotating menu, fast delivery experience offered by apps like Munchery and the majority of our weekday food budget has gone to them over the past six months. For logged-in users, the app stores dietary preferences, delivery information, and payment information. By limiting food choices to packaged meals, they reduce the decision-making equation to 1) does the food in the photo look tasty 2) does the description make it sound tasty 3) do other people find it tasty. The process might take as few as two steps to payment, from adding an item to delivery and order details.

Browse Menu
Browse Menu
Select Meal
Select Items
Order Details
Delivery Details (a)
Order Details
Order Details (b)

Assumptions

Based on a review of current competitors and consideration of my own experience, I concluded that the following features are important to reduce choice and simplify ordering food online for busy people:

Maximize Convenience

  • Store delivery information (phone number, email address, and default address — either the billing address or the most recent delivery address)
  • Show by default only restaurants open and available for delivery and remove distance and "open now" as a factor
  • Remove delivery time as a factor — baseline guarantee of fast delivery time with more accurate estimate in order details (people might care less then)
  • Store previous and favorite orders and taste preferences (e.g. cuisine)

Clarify Price

  • Store payment information
  • Remove delivery fees and minimums as a factor – flat low delivery/service fees or delivery subscriptions with optional tipping post-delivery
  • Display total cost per item after taxes and fees — what you see is what you pay
  • Display portion size estimates or filters to allow customers to order a suitable amount of food

Communicate Quality

  • Store dietary preferences and show only food that is relevant and available to order
  • Give maximum visual weight to food photography and let people "eat with their eyes"
  • Show social proof with number of favorites, number of orders (trending/popular/friends), or ratings/reviews
  • Enable food sources to differentiate from each other with content (food stories) and custom branding — restaurant or chef profiles

Caveats

The recommendations above come with certain caveats and constraints. Some are easy to implement while others require much more human investment and input. For example, I suspect stylized photographs of food are crucial to Munchery's growth as a service. Scaling that content production process to hundreds and thousands of small businesses would be quite the challenge. However, it is not impossible and the return to investment might be worth it as Airbnb has demonstrated with their product strategy. In their early days, they sent a professional photographer to every listing to enforce a perception of quality in their marketplace

Some may also face business constraints. For example, delivery subscriptions priced too low may make it impossible for a product to operate or improve its service. Even with the operational and financial resources of Google behind it, local delivery service Google Shopping Express still requires delivery minimums for subscribed users. Until recently, Munchery offered a flat annual delivery subscription that they have since rescinded.

Ideate

Using Paper by FiftyThree, I started sketching interaction ideas for the initial browse and select experience. All of these ideas depend on both high-quality visual content to attract engagement and increasing levels of personalization (machine learning) to strengthen conversion. They downplay the restaurant as an organizing schema and focus on selling dishes or food items. Finally, they assume that that "browsing" finite options is more effective than "exploring" infinite options when ordering food online.

1. Curated Categories

Curated Categories
Curated Categories

This sketch illustrates the Netflix model of recommendations applied to browsing food options. The recommendations would come from an algorithm of previous orders, popular items, dietary preferences, availability, and the other factors I discussed in the previous section. First-time users have the option to fill out a taste profile as they would on Netflix, selecting cuisines or categories of foods they favor (Southern, Breakfast) and perhaps even responding to photos of individual dishes (fried chicken, waffles).

This view assumes that the system has information about the preferences and tastes of logged-in or returning users. People see a list of categories (automatically created but improved with human curation, e.g. Breakfast for Dinner, Rainy Day Soups) and they can scroll horizontally through the options within each category. Selecting an item opens it to the details view or adds it to your order.

Potential Pros and Cons

  • It condenses many options into a small space.
  • People can quickly process and rule out groups of options.
  • Photo content needs to be clear and distinguishable in a small format. This might be difficult to execute with many restaurants.
  • Categories may be ambiguous or confusing because they could cover a wide range of criteria (season, cuisine, course, etc.).

2. Related Plates

Related Plates
Related Plates

This sketch shows a curated list of dishes with image size varying with portion size or course (i.e. side dish and desserts are smaller than entrées). Dishes that go well together would appear near each other. Because the view is limited to only several items at a time, human editors can help curate options to bump up seasonal or special items (summer salads) and related foods (salads and fresh fruit). I used round shapes to suggest the form of plates but it could also be a Pinterest-like masonry grid of different-size rectangles.

This idea stems from one of my pain points with Munchery. They offer great salads, soups, and sushi items that they call "side dishes." I often find it difficult to tell from the photograph or the price if the amount of food in the side dish is enough for one meal for myself (sometimes it is). Side dishes are usually in a category below main dishes so I can't compare them visually and I might miss them entirely.

Potential Pros and Cons

  • It highlights non-main dishes and may encourage ordering more food.
  • It is easier to assess the number of items to order for a given number of people.
  • It relies heavily on taste profiles and human curation to show the most relevant items earlier.
  • Different-size content blocks might feel chaotic and confusing.

3. Matched Meals

Matched Meals
Matched Meals

This sketch describes an interaction similar to the Tinder paradigm of showing one option at a time with a binary choice. Swipe right to bring back a previous option, swipe left to reveal a new option, tap the card to reveal more details or to select the item to order. While rotating through this carousel of options, you still see previews/peeks of previous and next options in the background. Options will show in order of "relevance" and recommendation based on your taste profile.

This interaction simplifies the consumer's choice to a yes/no question with each item and makes explicit the process of eliminating options. Because it offers "one meal at a time," either restaurants or editors have to package dishes and courses together into reasonable and desirable meals. I thought the Ness restaurant rating mobile app (sold to OpenTable and now defunct) executed this very well, especially in showing ratings and social connections.

Potential Pros and Cons

  • People see only one choice at a time.
  • Including additional features or actions with the card might make it confusing or cluttered.
  • People might dismiss a good option too quickly to see what's next.
  • It might be difficult to retrieve a dismissed option.

Refine

Flow

Select Meal
Select Meal
Place Order
Place Order
Add Meal
Add Meal

I designed the wireframes (with color) above in Sketch to examine some of my ideas in the target format. I learned that it's difficult to fit much textual content on a swipeable-card with a large photo and call-to-action. Thus, this might require linking to a further details view or creating a different design altogether depending on whether or not people have enough information to make a decision.

These wireframes also allowed me to explore some ideas, such as optimizing the flow for ordering a single item per order. Let's say that most people order only one meal or item at a time because most people order only for themselves. Instead of using a shopping cart or bag and describing the primary action as "Add," I might direct people to the order details page immediately after they select an item. Adding more items would become a secondary action on the order details page. Users adding more items would see visual cues that signal that they have started an order, such as updated UI language and colors. For people selecting more than one item, it might make sense to just update the order count in the header (rather than going straight to order details) until they're ready to proceed.

I would define and refine how people change options. For example, does editing the billing address happen within the order details or in a new view? I would experiment with decluttering the UI by testing more native interactions for making changes. For example, swiping left on an item to delete it from the order or swiping right to increase its count or save it to favorites.

Interactions

Pick and Drop
Pick and Drop

I would experiment with building animations and creative interactions into the experience. For example, we can play around with potentially more "delightful" ways to add items to your order other than a button. Perhaps when you select an item with a long press, a bowl illustration rotates or slides in from the bottom of the screen into which you can drag the item. Once it's added, the bowl shrinks back into the corner of the screen until you're ready to place your order.

Visual Identity

Logo Explorations
Logo Explorations

I started with the idea of using a simple bowl for a logo. The bowl is a food container found in many cultures, ancient and modern. It adapts to a range of functions, from sharing to storing, and accommodates a range of content, from beverages to bread. I thought this could make a good visual metaphor for a product that aims to be universally available and locally adaptable.

I dig the sketched effect that Paper creates (top right in image above) but that might not translate well in other forms and across all mediums. So as I drew the sketches from previous sections, I drew other quick variations: a lump of food in a bowl or a tongue in a laughing mouth, a bowl composed of two semicircles, a bowl with a lump of food peeking over the top, a literal drawing of a ladle. In the end, I reduced the shape to a tilted semicircle/half-disk shape. One could interpret it as a bowl, the scoop of a ladle, a laughing mouth, or an orange slice.

Assess

When we have refined enough details to launch the product, I would return to the questions posed in the initial section and collect data to answer additional questions:

  • How many people open the app but do not make an order?
  • How many people select/add an item but do not complete the order?
  • How many first-time users place an order in their first visit(s)?
  • How often do returning users place orders?
  • Which restaurants and items are the most popular?
  • Do people understand how to select or change options?
  • Do people understand how to select the initial item?
  • Do people understand how to add additional items?

Then I would continue to iterate and refine the experience. Looking forward, I would start exploring how to integrate the experience into other contexts and platforms. For example, if we cannot eat the food immediately or the meal requires some preparation upon arrival, the oven could automatically start pre-heating ten minutes before the delivery lands. When I am in my kitchen looking for food to eat, my refrigerator might recommend entrées to order based on its contents such as beverages or side dish ingredients.