Carefully Curious

UX Design
Visual Design

The tl;dr:

I know you're probably not going to read this whole page. So, to summarize:
This was a short exploratory project over several months to design a solution for a unique area for improvement: how to suggest new flavors and new menu items to Starbucks customers based on their buying habits. I was one of two designers on a team of several product owners and a developer. We worked collaboratively, often onsite, to brainstorm solutions, make prototypes of varying levels of fidelity, and test them with users. We delivered a set of viable options to weave into their existing mobile app that they could later develop and A/B test with.
Now you can just focus on the pretty pictures.

Starbucks has a diverse menu of beverages that boasts hundreds of thousands of ingredient and flavor combinations. Seasonal specials are always being added. Despite this versatility, most customers have around three go-to drinks that they order over and over.

How can Starbucks suggest new drinks to their customers in a way that feels personal and intentional?

At Formidable, I worked with Starbucks to help them design a brand-new feature of their mobile app to solve this problem. It dovetailed into their existing ordering flow and had two faces: entering your preferences, and browsing your recommendations.


Personas & Journeys

A lot of time spent was in workshops analyzing current data around user trends and purchasing habits. From a persona perspective, we opted not to create full personas for such a niche exploration (as to not conflict with personas already in use in the broader Starbucks ecosystem), and instead make a ranking system of curiosity. Any user could naturally fall into a label of safe, curious, or adventurous based on their own habits and desires in the moment. Our solution should cater to anyone on that spectrum.


Along with knowing those tendencies, the other important thing was knowing whenin the user journey were the right times to serve up these recommendations. This solution had to dovetail nicely into a user's existing process to placing an order, or prompt them to start a new order; anything besides a popup modal that would be instinctually closed.

The solutions we narrowed down to test with were a mix of tasteful prompts and suggesting modifications during an order. A push notification or inline block on the app home screen can enter the user into our "preferences quiz". From their menu screen and item details screen, we can suggest new ingredients to "twist up" their usual order.


Gather Preferences

For this feature to be worthwhile, drink recommendations need to truly feel personalized to the user – they need to take in both personal taste and dietary requirements. We did some extensive ideating over how to gather all this information from the user in a way that was quick and wouldn’t feel like any other bland survey or too cumbersome. We ended up with two prototypes:

1. Two questions to gather dietary needs and types of drinks you enjoy. From there, dive into sub-sections for each category for taste preferences – things like sweetness and caffeine levels, milk types, flavors, etc.
Our research with early prototypes told us that putting a lot of questions in a long list is pretty daunting. Most questions here have sub-questions that display progressively. Options about my milk preferences should only appear if I say I take milk in my coffee first.

2. Here, we tested a way to have the whole survey on one page. We wanted to see if users got the perception of the experience feeling shorter, as they could do the whole thing in one shot instead of going to different pages.
The first action is to choose which drink categories you’re interested in, which populates the accordion sections in the next state. Like the first, this ensures that the user only has to answer questions that pertain to them.


Variants of this quiz were explored, testing how the density of information felt splitting out the questions onto their own screens, rather than grouped or one page like in the prototypes above.


Display Recommendations


Recommendations are shown in a new “For You” tab alongside the existing tabs in the order experience. Previous testing enlightened us that users were really pleased with the idea of showing their usual drink order, a slight variation of it, and something new.

We came up with three variations of this trio to further test on. The items suggested will change based on the time of day or weather: your usual coffee in the morning, maybe a sweet from the pastry display in the afternoon, then something iced if it’s warm in the afternoon. Each item comes pre-populated with any ingredient preferences that the user has identified.


Some adventurous concepts were done to reenvision to product details page, to encourage those slight modifications and show the ingredients in a drink.


Within the above list of recommendations, we’ll pepper in these blocks so the user can fine-tune their tastes. These may be new questions, seasonal ones (do you like pumpkin flavored drinks?) or questions that they may have skipped in the initial survey.

Now What?