Pink Pie: A Case Study in UX Research

Dave Ostergren

Jun 23, 2020·14 min read

This project challenged me to focus on delivering an e-commerce experience for a local business and I was excited to use my design thinking skills to solve a real business need. My project partner MAYAH MCGOWAN and I chose to focus on Pink Pie, a local bakery specializing in mini-pies, as our “client” for this project. Fortunately for us, we were able to speak with the owner of Pink Pie, Michael, whose perspective afforded us tremendously valuable insight into Pink Pie’s current business context.

Like many local businesses, Pink Pie has been forced to adapt to a changing marketplace in response to the COVID-19 pandemic. Prior to the outbreak, Pink Pie relied heavily on the tourist market. Pink Pie’s brick & mortar location in Miami’s Wynwood district, a tourist favorite, allowed the business to capitalize heavily on this market segment. But with travel restrictions and regional lockdowns in place, that market evaporated.

In response to this shift in the market, Pink Pie began offering nationwide shipping by partnering with a third-party e-commerce platform. With so much uncertainty around when the tourist market will return in full to Miami, the current priority for Pink Pie is to increase the volume of nationwide shipping orders.

Now that we had a clear business objective in mind, it was time for us to evaluate the current user experience. Designers often conduct heuristic analyses on existing products and websites in order to find opportunities to add value by innovating or optimizing within the existing framework.

Given how subjective heuristic analyses are, it is useful to rely on an existing set of standards and definitions to make sense of the findings. Many designers, myself included, employ Jakob Nielsen’s 10 Usability Heuristics for User Interface Design while conducting a heuristic analysis of an existing site or product. While performing my analysis of Pink Pie’s website, I used a Google Chrome plug-in called UX-Check to easily keep my thoughts organized.

Here’s a very condensed summary of my Heuristic findings:

• Multiple opportunities to streamline ordering user experience
• Stylistic opportunity to embrace minimalist aesthetic through more efficient use of screen real-estate
• Opportunity to include more flexibility and efficiency of use during “Add to cart” phase (ex. provide quantity steppers earlier in the flow)

By this point, we’ve familiarized our self with the business context and objectives as well as the current user experience of ordering through the Pink Pie website. We were now ready to speak to customers of Pink Pie to understand their perspectives and motivations. In order to find users, we employed a guerilla research tactic: We found users who had engaged with Pink Pie’s recent Instagram posts and got in their DMs to ask if they’d agree to an interview with us. Here are some quotes that stayed with us:

“Everyone wants that (Amazon) Prime treatment.”

-A Pink Pie customer explaining her expectations of free/cheap shipping.

“every time she’s come down here she’s made a point to like go there.”

-Another customer sharing her Sister-in-Law’s loyalty to Pink Pie as a reason for why she chose to send her a pie from Pink Pie as a gift.

“I wanted to order for my wife on our Anniversary, but I needed something that showed more it was for my Anniversary, not just a message.”

-Customer sharing about their need for an occasion-specific gift greeting.

Define

There was definitely a pattern emerging in the data. It’s all too easy to get attached to what seems like a good idea early in the process, especially if there seems to be a pattern presenting itself.

In order to avoid getting attached to early ideas, I like to use a strategy known as a “parking lot”. This is a technique I first put into practice as a teacher. I would encourage my students to leave post-it notes on a corkboard that we referred to as the “bright idea board.”

I encouraged my students to leave their seats as soon as they had a good idea — a suggestion or inspiration for a current project — and place it on the bright idea board. We also had a question board, which operated the same way but was for holding onto questions that were maybe a little off-topic or hard to form at the moment.

The idea may seem a bit elementary, but there is good science to back it. Our brains can only hold about 5–9 “items” in short terms memory at a time. By writing something down with the intention to review it later, you have freed your brain from the burden of remembering it. This means more brainpower for designing!

Our conversations with Michael and his customers gave us a lot of great data to work with. Now we had to make sense of it all. Affinity mapping is a great tool for visually grouping data.

Affinity maps are only one type of visual organizer in the designer’s toolkit. Here is a task-analysis flow breaking down the sequence of steps a user will go through while completing the task of ordering a pie from Pink Pie online

Task analyses are useful for visualizing the user experience step by step and understanding the current state of the user’s experience.

Canvases like these are helpful when It comes time to make sense of the data collected during the Define phase of the design process. By visualizing the current user/customer experience it becomes much easier to understand the motivations behind their decision making. Journey maps, for example, are useful for pin-pointing where in the current process a designer can add value by introducing a well-designed solution to a user’s problem. Value Propositions are handy for a similar reason — ensuring that the product to be designed is aligned with the desires and expectations of the user or customer.

After completing a few definition exercises, we identified 3 potential problem statements to work with. Here is the problem statement we settled on tackling:

Our gifting customers are choosing competitors over us because they feel our online prices through [third party] are too high.

If we can solve this problem, it would impact gifting customers positively by reducing their cost to purchase our pies as gifts.

It would also benefit our business by increasing our shipping sales volume and revenue while generating and converting in-house customer leads.

With a problem statement clearly defined, we were just about ready to begin ideating possible solutions. But first, we flipped the problem statement on its head by rewriting it as a “how might we” statement:

How might we align our product offerings to better meet our customers expectations?

Same concept; much more empowering tone. Now we’re ready to get ideating.

Note: The problem statement above is centered heavily around pricing and strategy. When working with other professionals, designers may often encounter the belief that the job of a designer is “to make things look pretty.” If you are reading this, you probably know that that’s just the surface of what designers are capable of. It will often fall on the designer, in this case, to show — through explanation & example — that UX is a process for aligning business goals with user interests.

Ideation

Brainstorming can be a lot of fun! This is where the problems of today start to give up ground to the solutions of tomorrow. It can feel freeing to explore solutions after diving so deeply into the realm of problems.

Brainstorming can also be a little …meh. It’s all too easy to get stuck — as in either completely out of gas or stuck on one idea that seems really good (be careful of those). I’ve found that healthy brainstorming sessions share three traits:

• collaboration — ideation is a team sport. the more brains in the storm the better.
• speed — ideas should arise and be captured quickly. spending too much time on one idea can suck the life out of the brainstorm.
• (minimal) structure — a rigid agenda has no place in brainstorming. But, timeboxing brainstorm activities can help keep the ideas fresh.

Here are a couple of my favorite brainstorming techniques:

• Crazy 8’s — The idea is that each person will create 8 sketches representing 8 different solutions to the same problem statement. Typically 30 seconds is given for each sketch. Once all 8 sketches have been finished, each person in the brainstorm shares their sketches with the group. Sure, most of them probably suck, but that’s part of the process! Boby Haryanto has an interesting article on conducting this exercise remotely.
• Highs & Lows — Highs and Lows has two rounds: one for highs, and the other for lows. Each round should last about 60 seconds — more than that and you are likely overthinking. For round 1 (highs), imagine how you could spend the most money possible on a solution to your problem statement. No holding back, the bigger and more expensive the better. For round 2 (lows), come up with as many terrible ideas as possible. The more painful and inefficient for users and stakeholders alike, the better. I love this exercise for encouraging the brain to go wide and really push the boundaries a little.

After a few rounds of brainstorming, you’ve probably got a whiteboard (or a miro board) full of sticky notes. It’s time again to start narrowing our focus.

Develop

I’m a big fan of the MoSCoW method. MoSCoW is an acronym for “Must have, Should have, Could have, and Won’t have.” It’s handy for sorting brainstorming artifacts and ideas based on how closely they align to a How Might We statement. In other words, the MoSCoW method helps organize solution ideas according to how well they may actually solve the problem.

Completing the MoSCoW organizer above helped us appreciate the value of a gift-centered experience for our users.

Earlier I shared the value proposition canvas that we created for this project and explained how we relied on this to tool to ensure that our finished product would be aligned to the user’s needs and interests.

The Value Proposition canvas hinges around the concept of “Jobs to Be Done,” a framework for understanding consumer behavior that was coined by Clay Christensen. The underlying idea is that when consumers purchase a product, they are doing so to fulfill a need or want. They are in a sense hiring the product to fill a role.

In the video above — which you should totally watch after this — Clay Christensen recounts a story of conducting user research and observation to understand why so many McDonald’s customers were purchasing milkshakes at 7am. By applying the Jobs to Be Done framework, Christensen is able to uncover an entirely new market space for McDonald’s milkshakes that the company did not realize it was competing in.

This same approach helped us to understand that one of the most common “jobs” that users were “hiring” Pink Pie for was to give gifts to friends and loved ones. Pink Pie wasn’t just competing against other bakeries, they were competing against Hallmark and Ferrero-Rocher!

Here is the gifting Job to be Done for Pink Pie as we framed it:

When I am purchasing a gift I want to pick a fun and creative gift so I can give my friends/family/partner something that is unique and enjoyable.

We’re nearly ready to start prototyping now. But before we can get to that, it’s important that we define exactly what it is we are going to be building, and why. This will be our MVP, or Minimal Viable Product. An MVP should contain exactly as much functionality as is needed to address the original problem statement. It exists as a starting point — something to be iterated on over time.

Here is the MVP definition for our proposed enhancement to Pink Pie’s ordering experience:

We are building a gift-centered pie ordering experience for customers who are seeking unique and creative gifts because people want to share creative pies that they love with others.

A design project is a lot like a scientific experiment; there’s research, testing, measuring, analysis, and of course — a hypothesis. The MVP statement serves the role of the hypothesis in design. It is the idea that we are testing. In order to test successfully, we must consider what user results will validate our hypothesis, and what user results will invalidate it. These anticipated results are commonly referred to as metrics, though you may also hear expected results, and success/failure criteria.

Here are the metrics for success and failure as we defined them for this project’s MVP:

Success

• Site Traffic increase
• Gift order flow traffic & completion rate
• % of gift orders v. total order volume
• Revenue Generated by Gift flow feature
• User Return Rate

Failure

• Order flow drop-out vs. conversion rate
• Hanging cart data (items added to order but not submitted)
• Gifting specific (critical) feedback

A note on failure: Failure here does not mean catastrophic meltdown, but rather that we may have missed the mark slightly. Failure metrics should be thought of as indicators that something can be optimized further, better aligned to users' needs/expectations, or that there is a need to pivot.

Deliver

In the world of user experience design, there is no shortage of things to test. For this project, we focused on testing our prototypes for usability. Usability, as the name of the term suggests, is a measure of how effectively a tool can be used to complete a task. The best way to test for usability is to find someone to use your product!

An initial product design is typically referred to as a “loFi” prototype, for “low fidelity.” Best practice for loFi prototypes is to create design with pen(cil) and paper to keep the focus broad rather than detail-oriented.

In keeping with best practice, our LoFi was drawn up on paper. I first thought that this would be a pretty significant blocker to testing, especially with the current pandemic making “come pretend to use this paper app I made while I stare over your shoulder” a bit of a hard sell. But then I remembered that I live in 2020, and there are plenty of online tools for solving problems just like this!

We used Maze to test our paper prototype, and I gotta tell you I was aMazed by the experience!

As a side note, I absolutely love how just meta the idea of a tool designed by designers to help designers test the tools they are designing is.

Testing with Maze was (long dramatic pause) really cool, and helped us see a lot of our blind spots in the design. When you are the one building or designing something, you spend so much time “inside” of it that you can miss shortcomings that are really glaring once they’ve been pointed out.

A very wise man taught me that it only takes 5 tests to reveal 85% of a product's usability defects. So, we settled on testing with 5 users.

Now, you may very likely be wondering how would someone “use” a paper prototype — I know that I was. The answer is surprisingly — maybe even annoyingly — straight forward. Here’s how it works:

1. Testers create “tasks” or “missions” for the user to accomplish. This comes complete with fictional context for why the user is completing these tasks. The trick here is to be specific enough so the user knows what to look for, but not overly specific as to give them step by step direction. That would kind of defeat the whole purpose of testing for usability.
2. Via some “back-end magic” on Maze’s part, the user's interaction data such as screen taps and time on each screen is captured to be analyzed later.
3. Ideally, the researcher is able to observe facial expressions and body language for further insight into the user’s thinking. The researcher may also request that the user narrate their thinking (personally I feel like this may impact the authenticity of the user's actions a bit, but I’m a UX noob so take that with a grain of salt).
4. Survey questions (verbal or written) at the end of the testing session are commonplace for understanding where the user struggled with the interface and why they made the choices they made during the test.

This is really just the tip of the iceberg and should not be considered comprehensive. If you're interested in a guide on quality usability testing procedures here’s a great article by Yuna Orsini.

Once we had concluded our testing, it was pretty clear where the gaps were in our design were. Most user’s struggled with the card-based options we presented them with to choose between a “pre-fab” template gift greeting or a custom greeting builder. Despite the instructions prompting users to pick the template, most attempted to select the custom option. So in our second round of prototyping, we dropped the template altogether and doubled down on the custom greeting builder.

Stepping Back

We covered a lot of ground in this 4-day design sprint. Of course, if you are willing to “move fast and break things” it’s impossible to get everything to right in the first go around. Looking back on the project, there are certainly a few things I’d like to learn from and incorporate into my process moving forward.

Blind Spots

1. We began the project speaking to our stakeholder, Michael. Like I mentioned at the top, the insight into the business he offered was invaluable. Stakeholder engagement is critical if a designer or design team truly hopes to align the business objectives and user interests. With that said, our early conversations with Michael set a lens through which we viewed customer behavior. We focused heavily on the pain points he mentioned and allowed that steer our understanding of the user. The significance of the gifting use case was lost on us at first even though it was something that each interviewee mentioned. Ultimately, we realized this in time to pivot and accommodate that use case, but we very easily could have left that on the table.
2. In our early competitive analysis, we viewed the market space in a rather limited scope. We placed an outsized emphasis on the physical product of Pink Pie and neglected to consider the social and emotional “jobs to be done” (again, gifting) of our users. This prevented us from clearly defining the “blue ocean” that Pink Pie could occupy in the market early in the project

Next Steps

The sprint is over, but there are plenty of opportunities to expand on the work we have started.

If I were carrying this project for another sprint, here are the top priorities I would set for the next sprint:

1. Enhance and test the MidFi prototype. I would employ a design system to create a consistent and on-brand mid-fidelity prototype. Once complete, I would conduct a second round of usability testing. I would also incorporate some testing criteria to verify (or falsify) our assumptions that people are interested in a gift-specific ordering experience.
2. Revisit market analysis. Now that we have committed to pivoting towards a gift specific feature, I would take another pass at researching the market to include indirect competitors (that is, competition for gifts — not necessarily food-based competitors). I would conduct secondary research into the market as well as qualitative and quantitative research into consumer behavior when purchasing gifts.

Thanks for joining, I hope you had as much fun reading this as I had writing it. Please feel free to connect with me on Linkedin, and if you enjoyed my writing go-ahead and follow me on Medium — I’ll have more case studies like this one coming up soon.