🎓 Academic
Gen Z already knows how to find a recipe. The question was whether an app could make them confident enough to finish cooking it — and want to try again.
My Role
UX Designer
Team
Tools
Figma, Figjam, UserTesting.com
Timeline
9 weeks
I cook on pure instinct, zero measurements — delicious results on most days, occasionally regrettable ones. That made this project personally interesting. I designed the onboarding flow, the recipe preview screens, the step-by-step cook-along including the timer system, and the post-cooking documentation screens. The cook-along was the hardest — getting the timer behaviour, the video and text toggle, and the step progression to feel like it was with you in the kitchen rather than pulling your attention away from the stove.
THE PROBLEM
Research shows that learning to cook between 18 and 23 significantly improves long-term nutrition outcomes. And yet — Gen Z is one of the least confident generations in the kitchen. Not because the information isn't available. TikTok, YouTube, Instagram, food blogs — recipes are everywhere. The problem is that finding a recipe and actually finishing it are two completely different experiences, and nothing in the market was designed for the gap between them.
Yes, Chef! addresses that gap. Recipes come from a curated set of chefs with detailed, structured guidance — not user-generated content. The quality bar is high. But quality alone doesn't build confidence. The design problem was everything around the recipe: how you discover it, how the app adapts to what you actually own and can eat, how you get through the process without giving up, and what happens when you're done.
COMPETITIVE ANALYSIS
We evaluated 15 apps before narrowing to 6 for focused analysis — direct competitors including Parsnip, Meal Lime, Kitchen Stories, and Pick Up Limes, and indirect competitors Beli and Locket. Mapping them on two axes — individual vs social use, and prepping ahead vs documenting the past — revealed something immediately: the space we wanted to occupy was almost empty.
Most apps do either meal planning or recipe content — not both, and not with community. Recipe discovery, step-by-step instructions, and meal planning were the most common features. Real-time cooking support, social sharing built into the cooking flow, and genuine habit-building were almost entirely absent. The cooking app market is saturated with apps that get you to the recipe. None of them were designed to get you through it.
That framing shaped everything that followed. The question wasn't "how do we make a better recipe app." It was: how do we make an app that stays with you during the part that's actually hard?
RESEARCH
51 surveys and 10 interviews, split evenly between novice and seasoned cooks. We mapped insights across six themes: confidence, recipe decisions, ingredient management, social sharing, progress tracking, and community. The finding that changed our direction came from the confidence cluster — it wasn't a lack of skill that stopped people from cooking. It was anxiety before they started, stress during the process, and hesitation about whether the result was worth sharing. The friction was emotional, not informational.
— Usability Test Participant
Confidence, not skill, was the design problem. Every screen needed to reduce a moment where someone might give up — before they start, during cooking, and when they're deciding whether to try again. That became the test we ran every design decision through.
A second finding shaped the documentation feature specifically: 80% of both novice and seasoned cooks regularly share food photos with people they know — on Instagram, Snapchat, Beli, Locket. The assumption that only confident cooks share was wrong. Gen Z documents food regardless of skill level. That made in-app documentation feel essential rather than optional, and it's why we built both a public sharing mode and a private personal journal into the same flow.
We also confirmed something the competitive analysis had already suggested: 87–96% of both groups discover recipes through social media, not cooking apps. The existing apps were being used as references after the discovery happened elsewhere. That told us the market gap wasn't in discovery — it was in everything that comes after.
ITERATION
We ran four rounds across 27 participants — in-person pilot, lab sessions, and two rounds on UserTesting.com. Each round targeted the flows that had failed in the previous one. The metric that made our improvement most visible was total task time: between rounds one and two it dropped by 51 seconds. Not a dramatic number — but it represented less time confused and more time cooking.
The biggest single change was the timer system. In round one, participants couldn't tell which timer belonged to which step — they'd start a countdown and immediately lose track of what it was measuring. Labelling every timer to its step eliminated that confusion entirely in round two. The second change was the documentation screen — adding the wrap-up page and placeholder hints brought completion rates up in every subsequent round. Users weren't skipping it anymore because they finally knew it was there.
DESIGN DECISIONS
Every screen, every interaction, every line of copy was tested against one question: does this feel like a doctor's waiting room, or someone's kitchen table? These three decisions defined which answer we got.
Before anyone sees a recipe, we asked five things: dietary restrictions and allergens, appliances you actually own, ingredients you prefer to avoid, cooking goals, and a short knowledge quiz. Each question had a reason. I'm lactose intolerant — existing apps surfaced dairy-heavy recipes constantly, which erodes trust immediately. The appliance question mattered just as much: not everyone has an air fryer or a rice cooker, and recommending a recipe that requires equipment you don't own is a different kind of give-up moment, one that happens before you've even started. The quiz assessed baseline knowledge so the app could calibrate guidance from day one. Users responded positively — they felt understood before they'd cooked anything.
The original preview page asked users to commit to cooking before giving them enough to decide whether they wanted to. Dense layout, unclear hierarchy, a low-contrast call-to-action nobody tapped. I iterated it four times. The final version surfaces cooking time upfront in distinct phases — prep, cook, rest — shows which ingredients you already have in your pantry, flags allergens, makes substitution options visible before you commit, and adds cooking tip cards: short technique explanations you can browse before you start. The "I'm ready to cook" button became the highest-contrast element on the screen. Small change, meaningful signal — it tells the user this is the moment of commitment.
The original cook-along had a single view-only timer. You could watch it count down. You couldn't pause it, run multiple timers, or tell which step it belonged to. Novice cooks multitask — they're boiling water while chopping vegetables while checking if the pan is hot enough. The redesigned cook-along runs multiple labelled timers simultaneously, each expandable for a detailed view with pause and resume controls. It also offers video demonstration alongside text instructions, so users can switch based on how they absorb information mid-cook. After these changes, total task time dropped 51 seconds between rounds one and two.
In early testing, users completed the cook-along and tapped away without realising there was anything left to do. The documentation step was invisible. I redesigned the end-of-cook screen as a clear completion moment — a wrap-up page that signals the tutorial is finished before leading into reflection. Text boxes got placeholder hints so users knew what to write. They could rate their attempt, upload a photo, log a private note in their personal cooking journal, or share publicly. The choice between public and private mattered: not everyone wants to share a first attempt, but almost everyone wants to remember it.
RETROSPECTIVE
The cook-along works for someone who roughly knows what they're doing. But a genuinely novice cook who doesn't know what "fold in" means, or can't tell if oil is hot enough, hits the step-by-step instructions and is still lost. We added tip cards on the recipe preview and Train quizzes on the homepage — but both happen before you start cooking. What the cook-along actually needed was contextual help inside the step itself: a tooltip on "fold in" that opens right there, without leaving the screen. That's the version I'd build next.
The curated chef model is the right call for quality — structured, reliable, detailed guidance. But it creates a ceiling on what you can cook. A Gen Z user who finds a recipe on TikTok or gets one from their mum can't bring it into the app. In a follow-up version I'd explore importing and adapting external recipes into the Yes, Chef! format — keeping the tip cards and guided structure while opening up the content library.
Running unmoderated tests on UserTesting.com taught me something I didn't expect: writing precise task instructions is a design skill. In moderated sessions, participants could ask questions when confused. On UserTesting.com they interpreted tasks differently and took paths we hadn't anticipated. The tasks that generated the most useful data were the ones written most precisely. That's a lesson about communication I've carried into every research activity since.
