Designing a nutritional reflex, one scan at a time.
Pet owners wanted clarity. Labels had no language. Pawka had the science but no unified experience - scoring felt opaque, trust felt earned nowhere, and every aisle became a decision paralysis.
Scope of Work
A market no one could read.
Pet owners face 200+ references per aisle, ingredient lists nobody outside a lab can parse, and marketing claims with no way to check them. Pawka closes that gap with a single scan.
Consumers face cognitive overload in the aisle. The information gap makes rational decision-making nearly impossible in the 4 minutes and 20 seconds they spend at the shelf.
Pawka acts as a nutritional assistant. Scan a product and the interface decodes its composition into an actionable score - personalized to your animal's species, breed, age and weight.
Deliver a readable, credible experience across three radically different user profiles - from the skeptical expert to the time-pressured parent who just wants a green light.
Design Problem
How do you make complex nutritional content instantly intelligible and actionable for a diverse audience - in a rapid purchase situation - while preserving scientific rigor, full independence, and meaningful personalization?
Field first. Assumptions after.
12 semi-structured interviews, 3 in-store observation sessions, one competitive benchmark. Field data replaced assumption, not the other way around. Every decision downstream traces back to a specific finding.
The global score out of 100 resolves the cognitive block triggered by technical labels. A green/orange/red color code delivers a visual answer in under 3 seconds - no reading required.
92% - label opacity83% of testers demand an independent source. The interface exposes the scoring criteria, their weighting, and veterinary validation directly from the product sheet - transparency as a feature.
83% - need for trust75% face decision overload in the aisle. The scan function isolates a single clear sheet from 200+ references in 3 seconds. The interface suggests better-rated alternatives nearby.
75% - in-aisle overloadOnboarding collects species, breed, age and weight. This data contextualizes every recommendation and generates a pet-specific score on each scan - Laja (2019) UX personalization model.
JTBD - personalizationGuest mode allows 5 free scans per day with no sign-up. Account creation is only triggered at the right moment - saving favorites or viewing scan history. Value before commitment.
Nir Eyal model (2014)Three profiles. One product.
One primary persona, two secondary, two tertiary, two anti-personas - seven profiles pinned to every design decision that follows.
"I want to be sure I'm buying the best thing for my dog without falling into greenwashing."
"If my vet validates it, I buy it. If an app tells me the same thing simply, I'll listen too."
"If the app hides its methodology, I uninstall in 48 hours. If it shows it, I become an ambassador."
From discovery to loyalty.
5 phases, the Gibbons model (NN/g, 2018), emotion scored from −2 (frustration) to +2 (enthusiasm). Every friction point and design opportunity below comes from the dips, not the peaks.
Sees an Instagram post about pet food greenwashing. Searches for a product scanning app for pets.
Reads Pawka reviews on the App Store. Compares methodology transparency with other apps.
Downloads the app. Signs up. Creates her dog's profile - breed, age, weight.
Scans kibble in the aisle. Reads the score + explanation. Compares 2 available alternatives nearby.
Shares a product score on Instagram. Subscribes to alerts for controversial ingredients.
9 frictions. 3 absolute priorities.
Each friction below traces to a specific emotional low point on the journey map above. Severity, 1 to 5, based on how often it happens and what it costs in conversions. Fix the blockers first. Retention comes after.
Catastrophic. Demanding commitment before demonstrating value destroys conversion. Fix: guest mode with 5 free scans per day.
Major. Claire wants objective proof, Jean-Luc needs reassurance. Fix: color scale + text label + one-tap link to the scoring methodology.
Major. Only 50% direct success in usability tests, 4.7 misadventure clicks on average. Fix: persistent "How is this scored?" button on every product sheet.
Major. Claire is standing in an aisle, phone in hand, needs to decide in 2 minutes. Fix: geo-filtered alternatives with local availability and estimated distance.
Medium. Only 3 mandatory fields in the initial flow (species, breed, age). Everything else is optional and collected progressively over sessions.
Medium. Théo checks criteria before committing. Fix: public methodology FAQ page on the web. Direct link from the App Store description.
From architecture to pixel.
Hub-and-spoke mobile architecture. Scan as the central action. 5-tab bottom bar. Two modes - Guest and Account - with progressive conversion toward Premium.
A contract between design and development.
Atomic Design methodology. Foundations → Atoms → Molecules → Organisms. A structured UI Kit reduces handoff time, guarantees visual consistency, and documents every design decision as the product scales.
Grotesque
Buttons, badges, icons, inputs - all states documented
Criteria card - reusable across nutrition analysis and comparator
Full product sheet, login module, subscription cards
Empirical proof, not gut feeling.
Methodological triangulation, 4 sequential methods, 6 participants - 2 per persona - running from a five-second first glance to a 28-word semantic differential.
Measures first impression. Validates instant score readability and value proposition comprehension at 2 critical screens.
5 tasks covering critical paths. "Think aloud" protocol. Reveals mental models and areas of confusion that metrics alone cannot detect.
10-item Likert questionnaire. Single 0-100 score. Benchmarked against sector averages. Validates usability before launch.
28 semantic differential word pairs. Measures Pragmatic Quality, Hedonic-Stimulation, Hedonic-Identification, and Global Attractiveness.
of participants recalled the score and color code from the product sheet after 5 seconds. Core design hypothesis validated.
average SUS score - 68th sector percentile, above health/nutrition average (68.4). Level: "Good."
AttrakDiff Pragmatic Quality - Self-Oriented quadrant: the app is perceived as both useful and identity-bearing.
"I understood the score immediately - it's exactly like the Nutri-Score. Orange means not great, I wouldn't have bought it."
"I could explore without giving anything away. That's reassuring for someone like me who's suspicious of apps."
"I'd put it as a widget on my iPhone to access it fast without even opening the app."
| Task | Direct Success | Assisted | Avg Time | Target | Misclicks |
|---|---|---|---|---|---|
| T1 - Onboarding + pet | Good result: 83% | 17% | 2m 10s | <3min | 2.3 |
| T2 - Scan + read score | Good result: 100% | 0% | 28s | <60s | 0.2 |
| T3 - Scoring method | Needs improvement: 50% | 33% | 1m 45s | <60s | Needs improvement: 4.7 |
| T4 - Favourites + history | Good result: 83% | 17% | 35s | <60s | 1.1 |
| T5 - Free vs Premium | 67% | 33% | 1m 05s | <60s | 2.8 |
WCAG 2.1 AA from day one.
Accessibility integrated at the conception stage, not retrofitted. Fixing an accessibility issue in design costs 10× less than in development, and 100× less than post-launch. Audited with Stark and A11y Annotation Kit across 14 WCAG 2.1 criteria.
All interactive elements meet the 44×44pt minimum touch target (Apple HIG). The central scan button is 56×56pt. 8pt minimum spacing between adjacent interactive elements prevents accidental taps.
Interface copy targets CEFR B1-B2 level. Technical terms always paired with plain-language explanations. The score reduces 200+ references to one clear decision - directly addressing the paradox of choice. Constant visible feedback eliminates interface anxiety.
The score gradient (red to green) is the classic deuteranopia risk. Mitigated by two design decisions: a systematic text label on every score level (Critical, Poor, Acceptable, Good, Excellent), and differentiated luminosity values across score tiers - not just hue.
| Color Pair | Ratio | Normal Text | Large Text | Usage |
|---|---|---|---|---|
| Saddle Brown on Snow | 9.20:1 | AAA | AAA | Body text, titles, labels (primary) |
| Snow on Saddle Brown | 9.20:1 | AAA | AAA | Reversed text, filled buttons |
| Saddle Brown on Pale Yellow | 7.36:1 | AAA | AAA | Titles and badges on yellow bg |
| Saddle Brown on Mint Green | 6.28:1 | AA | AAA | Headings on green bg |
| Alert Red on Snow | 4.82:1 | AA | AA | Error messages, critical scores |
| Premium Gold on Snow | 3.12:1 | Fail | AA | Icons and large text only |
| Snow on Mint Green | 1.46:1 | Fail | Fail | Never used for text - excluded from tokens |
4 quarters. Measurable targets.
Every milestone is grounded in test data. Fix conversion blockers first. Then iterate on retention, growth, and international expansion - each quarter with a quantified objective.
- Scoring method button P0
- 7-day Premium free trial
- Loaders with progress + copy
- A/B test Premium CTA placement
- iOS / Android deploy
- Product comparator
- iOS / Android widget
- Offline mode - 50 cached products
- Video onboarding
- Story share template
- Multi-species (exotic pets)
- Alternatives marketplace
- Gamified progress
- Geofenced push notifications
- UK · DE · ES
- Petstore partner API
- Cultural color code adaptation
- Contextualised prescription
11 - Key Learnings
83% of testers demanded proof before trust. Transparency became the product.
12 interviews said the same thing in different words: nobody trusts a company grading its own homework. That's why every scoring criterion sits in the open on the product sheet, weighting included. It's also why 100% of testers still remembered the score five seconds after looking away. Pawka isn't betting on being liked. It's betting on being checkable - and in a market this saturated with greenwashing, checkable is the only kind of trust that survives an aisle, a skeptical buyer, and four minutes and twenty seconds to decide.
More projects
Got a cool idea?
This spot is waiting for your success story.
Twiggli
Product design for a family-sharing mobile app - helping parents and kids stay connected through shared memories, routines, and thoughtful UX.