Onsite Search

A search that starts with inspiration and ends with an itinerary.

Onsite Search

Onsite Search

Onsite Search

A search that starts with inspiration and ends with an itinerary.

A search that starts with inspiration and ends with an itinerary.

Product Design

Product Design

Mobile App

Mobile App

UX Research

UX Research

ROLE

ROLE

Product Designer, UX Researcher, Developer

Product Designer, UX Researcher, Developer

TOOL STACK

TOOL STACK

Figma · React, Cursor · Claude Code · Anthropic API · Microsoft Azure API · Vercel

Figma · React, Cursor · Claude Code · Anthropic API · Microsoft Azure API · Vercel

DURATION

DURATION

TIMELINE

5 Months

5 Months

OUTCOME

OUTCOME

Presented at

Presented at

· Digital Product Design & Cross-functional teams

· Digital Product Design & Cross-functional teams

TEAM

TEAM

Solo

Solo

MoMA's digital archive holds over 106,000 works. An internal report told the real story, 80% of users left search without finding what they wanted, 35% of searches happened on mobile, and 861 people scanned a QR code for Starry Night while standing inside the building, asking the same three questions: where is it, is it here today, what should I see next. Search had no answers. It returned a flat list and stopped. I spent my fellowship designing for that person, the one with a phone in their hand and a painting somewhere in the building they can't find.

MoMA's digital archive holds over 106,000 works. An internal report told the real story, 80% of users left search without finding what they wanted, 35% of searches happened on mobile, and 861 people scanned a QR code for Starry Night while standing inside the building, asking the same three questions: where is it, is it here today, what should I see next. Search had no answers. It returned a flat list and stopped. I spent my fellowship designing for that person, the one with a phone in their hand and a painting somewhere in the building they can't find.

80%

80%

left search without finding what they wanted

left search without finding what they wanted

35%

35%

of searches happen on mobile, often mid-visit

of searches happen on mobile, often mid-visit

861

861

onsite QR scans for Starry Night alone, asking where to find it

onsite QR scans for Starry Night alone, asking where to find it

The problem: search breaks for someone standing onsite at MoMA

The problem: search breaks for
someone standing onsite at MoMA

Search works for a desk. It breaks for someone standing in the building. Half of MoMA's searchers are academic researchers. Fifteen percent are planning a visit. Then there's a third group the data kept surfacing but the product ignored: people already onsite, phone out, mid-visit.

Search works for a desk. It breaks for someone standing in the building. Half of MoMA's searchers are academic researchers. Fifteen percent are planning a visit. Then there's a third group the data kept surfacing but the product ignored: people already onsite, phone out, mid-visit.

Every query returned the same flat list. That works if you know exactly what you want. Most people don't.

Every query returned the same flat list. That works if you know exactly what you want. Most people don't.

The result page made it worse: every answer was one tap too deep

The result page made it worse:
every answer was one tap too deep

Search wasn't returning the wrong thing. It was returning the right thing in the wrong shape. Someone standing in the building searches "Vincent Van Gogh" and gets 735 results: title, medium, archive documents, old exhibition records. The gallery location is on the artwork page. The on-view status is on the artwork page. The audio guide is on the artwork page. Every answer requires a tap.

Search wasn't returning the wrong thing. It was returning the right thing in the wrong shape. Someone standing in the building searches "Vincent Van Gogh" and gets 735 results: title, medium, archive documents, old exhibition records. The gallery location is on the artwork page. The on-view status is on the artwork page. The audio guide is on the artwork page. Every answer requires a tap.

The answer exists. it's just buried one tap too deep for someone standing in the building

The answer exists. it's just buried one tap too deep for someone standing in the building

How might we help someone move from vague curiosity to a confident plan they can walk?

How might we help someone move from vague curiosity to a confident plan they can walk?

Competitive audit: museums vs. the apps people actually use daily

Competitive audit: museums vs. the apps people actually use daily

Before I designed anything, I audited direct museum competitors and the discovery platforms people use every day: Google, Amazon, Spotify, Pinterest, Netflix. The museums told me what not to do, most prioritize archival depth over visitor usability, built for lookup, not exploration. The consumer platforms told me what to do. The pattern repeated everywhere: the strong ones don't just return results, they guide the next decision.

Before I designed anything, I audited direct museum competitors and the discovery platforms people use every day: Google, Amazon, Spotify, Pinterest, Netflix. The museums told me what not to do, most prioritize archival depth over visitor usability, built for lookup, not exploration. The consumer platforms told me what to do. The pattern repeated everywhere: the strong ones don't just return results, they guide the next decision.

The best platforms don't show you results.
They help you decide what to do next.

The best platforms don't show you results.
They help you decide what to do next.

The reframe: search isn't a lookup, it's the start of a visit

The reframe: search isn't a lookup, it's the start of a visit

I stopped treating search as a tool and started treating it as the entry point into learning, planning, and walking the museum. That single shift turned the project into one connected system: a search-to-itinerary experience. Three pillars hold it up.

I stopped treating search as a tool and started treating it as the entry point into learning, planning, and walking the museum. That single shift turned the project into one connected system: a search-to-itinerary experience. Three pillars hold it up.

DESIGN DECISION 1

DESIGN DECISION 1

The Empty State:
three ways in before you type

The Empty State:
three ways in before you type

Onsite visitors arrive without a query in mind. An empty search bar is the coldest possible thing to hand them, it's a blank box asking a person who came to be inspired to do the work of being inspired first. I replaced the empty state with three ways in before you type: popular pills for artworks, artists, and movements; thematic collections you can tap straight into; and ready-made itineraries like "The Unmissables" and "Slow Looking." The first tap is already a step into the collection, not a search for one.

Onsite visitors arrive without a query in mind. An empty search bar is the coldest possible thing to hand them, it's a blank box asking a person who came to be inspired to do the work of being inspired first. I replaced the empty state with three ways in before you type: popular pills for artworks, artists, and movements; thematic collections you can tap straight into; and ready-made itineraries like "The Unmissables" and "Slow Looking." The first tap is already a step into the collection, not a search for one.

DESIGN DECISION 2

DESIGN DECISION 2

The Digest Card:
a result shaped to what you searched

The results page matches what you
actually searched for

MoMA's current search returns one layout for everything, artist, artwork, movement, document, it's all the same stacked list. I built the Digest Card to fix that. Your top result becomes a card built for the kind of thing you searched. Search an artist: short bio, works on view, gallery locations.

MoMA's current search returns one layout for everything, artist, artwork, movement, document, it's all the same stacked list. I built the Digest Card to fix that. Your top result becomes a card built for the kind of thing you searched. Search an artist: short bio, works on view, gallery locations.

Built for what you searched, not just what you typed. Search an artwork: the piece, its location, on-view status. Search a movement: definition, key artists, defining works. The answer arrives with the search instead of three taps past it.

Built for what you searched, not just what you typed. Search an artwork: the piece, its location, on-view status. Search a movement: definition, key artists, defining works. The answer arrives with the search instead of three taps past it.

Artist — "Claude Monet"

Artist — "Claude Monet"

Search an artist and the result leads with who they are and what's here today. Three works on view, each with a gallery number. A recommended itinerary at the bottom so the visit starts before you leave the search.

Search an artist and the result leads with who they are and what's here today. Three works on view, each with a gallery number. A recommended itinerary at the bottom so the visit starts before you leave the search.

Artwork — "The Persistence of Memory"

Artwork — "The Persistence of Memory"

Search an artwork and the result becomes the artwork page. Full image, on-view status, gallery location, and an audio guide tap, all without leaving search. The answer is in the result, not behind it.

Search an artwork and the result becomes the artwork page. Full image, on-view status, gallery location, and an audio guide tap, all without leaving search. The answer is in the result, not behind it.

Movement — "Expressionism"

Movement — "Expressionism"

Search a movement and the result orients you to the idea first, then the people behind it. A short definition, the key artists on view, their works. Discovery mode for someone who knows a feeling but not a name.

Search a movement and the result orients you to the idea first, then the people behind it. A short definition, the key artists on view, their works. Discovery mode for someone who knows a feeling but not a name.

DESIGN DECISION 3

DESIGN DECISION 3

The Itinerary: a running plan that follows you across the app

The Itinerary: a running plan that follows you across the app

This is the through-line that ties the system together, and it was the piece visitors responded to most. Every search ends with a recommended itinerary built around what you searched. Once you create one, a banner travels with you across every page, ready for the next checkmark. One tap adds the next work to the running plan, and you keep adding as you browse, the way you'd add songs to a playlist. Search becomes a visit, not a list.

This is the through-line that ties the system together, and it was the piece visitors responded to most. Every search ends with a recommended itinerary built around what you searched. Once you create one, a banner travels with you across every page, ready for the next checkmark. One tap adds the next work to the running plan, and you keep adding as you browse, the way you'd add songs to a playlist. Search becomes a visit, not a list.

From a single search to a plan you can walk

From a single search to a plan you can walk

Every search ends with a recommended itinerary. Build one and a banner follows you across every page, ready for the next checkmark. One tap adds the next work to your running plan. You start with a name and leave with a route.

Every search ends with a recommended itinerary. Build one and a banner follows you across every page, ready for the next checkmark. One tap adds the next work to your running plan. You start with a name and leave with a route.

USER TESTING

USER TESTING

5 users. 4 tasks.

5 users. 4 tasks.

Split between casual visitors and art enthusiasts. A few things landed as expected. One comment changed the direction entirely: the itinerary should work like a Spotify playlist, where you just keep adding.

"It's very rare that I know what I want so recommendations are great."

"It's very rare that I know what I want so recommendations are great."

-Curious Visitor

-Curious Visitor

"Appreciated the audio guide for crowded rooms where wall text is hard to read."

"Appreciated the audio guide for crowded rooms where wall text is hard to read."

-Art Enthusiast

-Art Enthusiast

"Itinerary should work like a Spotify playlist, where I keep adding artworks.”

"Itinerary should work like a Spotify playlist, where I keep adding artworks.”

-Art Enthusiast

-Art Enthusiast

"Attention was immediately drawn to Explore all Women Artists."

"Attention was immediately drawn to Explore all Women Artists."

-Curious Visitor

-Curious Visitor

What I'd build next

What I'd build next

Multiple saved itineraries, one per visit or mood

Multiple saved itineraries, one per visit or mood

Search by image, snap a work and find similar pieces on view

Search by image, snap a work and find similar pieces on view

Cross-device handoff, plan at home, pick up onsite

Cross-device handoff, plan at home, pick up onsite

Store integration, surface the catalog and prints tied to what you saved

Store integration, surface the catalog and prints tied to what you saved

Reflections

Reflections

Designing for exploration is not designing for lookup

Designing for exploration is not designing for lookup

Discovery needs different patterns than transactional search. Helping someone explore means guiding them, not just retrieving for them.

Discovery needs different patterns than transactional search. Helping someone explore means guiding them, not just retrieving for them.

System thinking beats screen thinking

System thinking beats screen thinking

This project pushed me past individual screens into one connected journey across search, planning, and physical navigation. The value lived in the connections, not the components.

This project pushed me past individual screens into one connected journey across search, planning, and physical navigation. The value lived in the connections, not the components.

Discovery needs restraint

Discovery needs restraint

More intelligence does not mean more interface. Knowing when to guide and when to step back mattered as much as anything I added.

More intelligence does not mean more interface. Knowing when to guide and when to step back mattered as much as anything I added.

© 2026- Shreesa Shrestha

© 2026- Shreesa Shrestha