faircado logo

Redefining visual search: from MVP to star feature

faircado is a fashion aggregator app on a mission to make second-hand shopping as easy as buying new. By 2025 it had over 300k users, mostly Gen Z.

Featured by App Store

My Role

Product Designer

UX Researcher

Timeline

6 months

2025-2026

Team

Team lead

Space

E-commerce

Aggregator app

My Impact

  • Built the company's first structured user research process.
  • Introduced qualitative research to a team hearing it for the first time.
  • Shaped a year-long roadmap, tech sprints, and business strategy through continuous insights.

context

Why visual search?

Short answer: we were not nailing text search.
Each marketplace uses its own taxonomy and keywords, making standardized text search nearly impossible.

At the same time, recent reports from Pinterest and Google showed Gen Z shops visually first. Both facts pointed us in the same direction: build an image search MVP.

Article: The End of the Search Bar? How AI Visual Search Is Redefining How Gen Z Shops
Stat card: 80% of Gen Z relies on Google Search for shopping discovery, research, and decisions
Stat card: 69% of Gen Z agree that visual results are more helpful than text when deciding what to buy

PROBLEM

The MVP engaged.
But didn't sell.

Image searchers returned 3x more often than text searchers. Potential was there, but users scrolled results, favorited, opened product pages, and didn't buy.

Despite the potential, only ~40% of users ever tried image search, and the ones who did use it weren't converting.

So my challenge became:

How can we improve the image search MVP to drive sales and grow adoption beyond 40%?

MY APPROACH

Evaluating the problem

I led the research and investigation by digging deeper into the problem end-to-end with 4 approaches:

Session recording tool logo
Session recording tool logo

User Tracking

I've analyzed 50+ user behaviour recordings

Interview participant
Interview participant
Interview participant

User interviews

Conducted 30+ user interviews over 6 months

Amazon logo
Pinterest logo
Google logo

Benchmarking

Analized main players in the visual search sphere

Survey

Collected 6 months of qualitative data

The data told us what.
I needed to find out why.

The in-app survey revealed that users' top complaint was 'Didn't match my image', but the team didn't know the specifics.

So I brought qualitative research into a team hearing it for the first time.

Survey prompt asking 'Happy with these results?' with 'Didn't match my image' selected as the reason
Survey prompt asking why items were left in the bag, with 'Still thinking about it' and 'Just browsing' selected

In-app survey launched with engineering

TAKEAWAYS

Three insights
guided my work:

Didn't know I could
image search
80% of behaviour recordings showed users missing the entry point to image search. Interviews confirmed users weren't aware of it.
Awareness

Adoption was an awareness problem

How can I make image search impossible to miss?

Results didn't match
my image
#1 complaint, with over 4,000 mentions. Interviews revealed the model missed detail-heavy items e.g. embroidered jeans and v-necks.
Match accuracy

Our AI model lacked match accuracy

How can I make it catch details and give users more control of search?

I was still thinking
about it
70% of users who didn't buy were “just browsing” or “still thinking”. Interviews found timing and budget slowed down decision-making.
Momentum

Conversion was a momentum issue

How can I create urgency and make it easy for users to search again?

FINAL SOLUTION

One feature, extended across three fronts

Over the next 6 months, I collaborated with engineering to revamp image search across the following fronts:

01.Awareness

Awareness

Redesigned onboarding to spotlight image search from day one

Video walkthrough teaches users to select one item per photo
Interactive onboarding screen showing real, favoritable search results
Interactive onboarding with real, favoritable results
Registration screen continuing the onboarding story
Storytelling continues into registration screen
Awareness

Added 2 new entry points to the homepage so image search is visible

Homepage before adding new image search entry points
Before
  1. Camera icon in the search bar
Homepage after adding new image search entry points
After
  1. Camera icon in the search bar
  2. New “Your images” section on the homepage
  3. New camera icon in the bottom navigation

02.Accuracy

Match accuracy

Retrained the model to catch key details and refine results with text

Model training example highlighting logos, embroidery, and collar type
Logos, embroidery and collar type were the most mentioned by users
“Add to search” allow users to refine visual search with text
make it blue
short sleeve
like this v-neck
Match accuracy

Launched 'Find similar' to turn every item into a sharper, more engaging search

Find similar button added to a product page
“Find similar” button added to product pages
“Find similar” feature in action, creating an engaging search loop

03.Momentum

Momentum

Added urgency badges to push 'just browsing' users to act

Product page with a fire urgency badge reading Almost taken, hurry up
Fire badge signals items already in other users' bags
Product page with a lightning urgency badge reading Popular item
Lightning badge flags items other users are favoriting
Momentum

Built two features so 'still thinking' users easily pick up where they left off

Recent searches feature showing past image searches
“Recent searches” feature
Image history page collecting past search photos
Image history page

DISCARDED IDEAS

Some of the many 'NOs'

For each front, I came up with many ideas that didn't make the cut due to complexity, tech constraints or lack of resources. Prioritizing what moved the needle took careful thinking and real tradeoffs.

Dialog shown on app open, prompting users to try image search

Dialog on app open

Removed after testing showed Gen Z skip dialogs entirely

Recommendation feed of visually similar items

Recommendation feed

The algorithm underperformed and was discontinued

Fairnando, an AI shopping agent chat assistant

AI shopping agent

“Fairnando” became too complex to build, so it was shelved

IMPACT

Image search became
the star feature

Once overlooked, image search became the heart of the app, and each solution pushed it further. Below are the highlights:

Adoption of image search grew to 89%

Image search adoption climbed from ~40% to a peak of 89% in September 2025

“Find similar” feature drove 3x higher sales

Similarity search converted at a 3-4x higher rate than text search; search-history image, at 2x.

Urgency badges had 3x higher engagement

Urgency badged items were added to bag and favourited 3x more than un-badged ones.

I've noticed faircado is pricier than eBay, but I don't mind because the experience is way better.
faircado user

faircado user, the Netherlands

LEARNINGS

Lessons that stuck

MVP is not enough.
MVE is the new bar.
Users are increasingly tech-savvy with high expectations. A Minimum Viable Product is no longer enough. Minimum Viable Experience is the real bar.
When AI solves speed,
direction takes over.
Teams can now ship features simply because they can, rather than ask what's driving impact. Speed is no longer the differentiator, judgment is.
Never overlook a key
feature over another
After adoption of image search peaked, it settled on 70%, leaving 30% of users relying on a broken text search. Optimizing one feature doesn't excuse a gap in another.

Next Project

eco.mio

Automating sales demos to cut 15h of work monthly

How I built a client-ready demo maker that turned 5 hours of work into 30 minutes at a B2B climate tech company.

  • AI Product Engineer
  • Streamlining a cross-team process
  • 5-hour process → 30 minutes
  • Climate tech | B2B SaaS