Communication UX

Mobile Friendly

2025

Cass Connect

96–98% of ad-driven users left on the first screen.

Using GA4 funnel data and session replay,

I identified where users dropped off and redesigned the flow around mood and context.

Client

Cass

Role

UX Lead — Strategy, IA, Wireframe, Prototype

Year

2025

Platform

Responsive Web

Background

"People don't search for a bar — they search for a vibe."

Brief

Cass needed a venue discovery platform that could turn ad traffic into retained users.

Problem

Existing apps filtered by location or rating but users chose venues by mood, occasion, and social context.

Opportunity

Design a mood-first flow that feels like a friend’s recommendation not a search engine.

GWI · KEYWORDS · SOCIAL LISTENING → PERSONA

We didn't guess who the user was.
We found them in the data.

Sentiment

83%

Positive — Strong brand affinity, with negative spikes around health and excess drinking.

What they talk about

24%

Location mention — Location was a recurring topic and a clear UX entry point.

Where they are

70%

Instagram Discovery — Instagram was the primary venue discovery channel for this cohort.

Primary Persona

New Drinker

A digitally connected socializer exploring new tastes and expressing personal style.

3.02M

projected pop · Age 20–34

Pain Points → Design Triggers

✕No vibe/ambience filter — Users could search by category or proximity but not by mood.

✕Weak Atmosphere Cues - Photos did not communicate the venue vibe quickly.

✕No social shortlist — Users could not save and share a curated night-out list.

Design Direction

Mood-first. Context-aware.
Cass-curated.

Emotion
Before Location

Start with mood, not search.

Vibe in
One Second

Use full-bleed photos and minimal text to make atmosphere instantly readable.

Brand
Without the Ad

Show Cass venues as contextual recommendations, not banners.

Wireframe — Key Screens

From IA to interaction, before any hi-fi

Four core screens were tested before visual design to validate flow clarity, momentum, and the retention loop.

"Where to drink?" was not the question
— "what mood am I in?" was.

Research

Keyword analysis and social listening mapped real consumer needs.

Benchmark

Naver, Kakao, and Catchtable audits showed where context was missing.

Design

A Mood → Browse → Save flow was validated through user testing.

User flow

Landing
- Instant Mood Rollout

Start with age check, then reveal a mood card to set the tone

Personalized Discovery
– Set Your Vibe

Choose a vibe and location to get matched venue suggestions.

Engaging Exploration
– Swipe, Discover, Decide

Swipe through venues and decide quickly

Easy Social Planning
– Share the Moment

Share the selected venue with friends to coordinate the night out.

Brand Connection

Cass-Curated, Not Ad-Driven

Cass-curated spaces appear as contextual recommendations.

Users left before they understood the service.

GA4 funnel data and session replay revealed three drop-off points each pointing to a different design failure

96–98%

First-screen exit rate

Ad-driven users left before interacting

↓

① Clarity & Guidance

Hero copy + 1–2 step onboarding needed to set expectations

44.06%

Steps 2→3 conversion

Users picked a mood, then lost momentum.

↓

② Engagement & Motivation

Post-action feedback, reward signals, and loading animation

0.26%

Step 5 reach rate

A major drop-off revealed a trust and relevance gap

↓

③ Depth & Trust

Real photos + AI summary + brand sponsor info on place cards

Solution 01 — Home screen

Make the First Action Obvious

Swipe became the entry point

when users missed how to start.

AS-IS — Home screen

No swipe signal, no state feedback

User flow: No signal → hesitation → drop-off

HIGH — Order Bias Top 3 drove 47.3% of choices

HIGH — Weak progress motivation Progress lacked completion push

MID — Hidden State Selected moods were unclear

TO-BE — redesigned

Swipe action copy + dot indicator + chip state

Solution 02 — Place card

Add Context Before Decision

Real photos and quick summaries

helped users decide faster

AS-IS — Place card selection

Illustration only — can't judge atmosphere

User flow: Illustration → Uncertainty → External search → Exit

HIGH — Low Trust Illustrations made the real vibe unclear

HIGH — No Feedback Actions gave no clear response

MID — Weak Context Missing summaries increased effort

TO-BE — redesigned

Real photos + AI summary chip + loading animation

User decision flow: Real photo check → AI summary → instant action → visit decision