Social
Redesigning Picsart's Browse page for a shifting user base, from an undifferentiated feed to a personalized, research-validated discovery experience for creators and prosumers.
| Company | Picsart |
| Product | Browse — Inspiring Feed |
| My Role | Senior Product Designer · Research · IA · Interaction design · Validation |
| Platform | iOS · Android · Web |
| Timeline | 2024 |
| Users | Consumers and prosumers (solopreneurs, small businesses, marketers) |
| Key Decision | One-column feed with two-tab navigation, chosen after testing three layouts with six users |
| Outcome | Design validated across four research cycles · Prototypes advanced to A/B testing |
Picsart's prosumer segment was growing, but the Browse page hadn't been updated to reflect that. The gap between the audience Picsart was building toward and the experience the feed was designed for was a product strategy problem, not just a design one.
As the sole designer, I ran four structured research cycles end to end, from unmoderated value interviews to prototype testing with six users, then redesigned the full feed architecture, content tagging system, and personalization model.
A research-validated feed redesign with a documented rationale for every decision. Prototypes advanced to A/B testing. A prior shift to one-column layout had already shown a meaningful increase in Time-to-Image engagement in Picsart's own product data.
I left Picsart at the end of this project, so I don't have post-launch metrics. What I have is a clear record of what was validated and what was handed off.
One-column, two-tab structure tested with users and confirmed through Cycle 2 testing and existing behavioral data.
No decision in this redesign came from a preference or a stakeholder opinion. Each one connects to a specific research finding.
Feed architecture, content tagging, and personalized carousels designed for the professional user segment Picsart was growing toward.
Multi-action CTA puts Remix, Try, and Edit in front of users during browsing, not buried two taps away.
Validated designs handed to engineering as A/B test candidates. Live measurement happened after my tenure.
One data point from before this redesign is worth noting. Picsart's own analytics showed a meaningful increase in Time-to-Image engagement after a prior shift from two-column to one-column layout, with minimal impact on editor opens. That data informed the layout decision but didn't solely determine it.
A feed that shows less but shows the right content is more useful than one that surfaces everything. Fewer, better-matched items beat more items on every metric that matters.
Different content types serve different intentions. The mistake in previous versions was separating by asset type. The right separation is by user goal.
Picsart isn't a mood board. Remix, Try, and Edit are what make it different from a social feed. The Browse page should surface those actions, not bury them.
Users want their interests reflected. They also want to be surprised. A feed that only shows what someone already likes gets boring fast.
Seeing the same content in top positions for weeks tells users the system is broken. Recency and rotation aren't nice-to-haves — they're the minimum.
Picsart built its Browse experience for consumers looking for visual inspiration. By 2024, the platform's growth was shifting toward prosumers, solopreneurs, small business owners, and marketers who came with fundamentally different browsing intent. The feed hadn't been designed for that difference.
Previous attempts to address this had missed the root cause. Separating content by asset type addressed format, not intent. Gross engagement signals surfaced what was broadly popular, not what was relevant to a specific segment. Both approaches treated the feed as a single-audience product.
I ran research in tight, focused cycles not one large study, but four rounds each answering a specific question. These are the findings that shaped the most important decisions.
Talking to users made the split obvious fast. Consumers browsed broadly — scrolling until something caught their eye, saving it, moving on. Prosumers came with a job to do. They were looking for a template, a style direction, something they could actually use for a project. When the feed couldn't surface that, they stopped scrolling. The same interface was working fine for one group and actively frustrating the other.
When content felt arbitrary or off-topic, users lost trust in the feed fast. A simple signal like "suggested" or "for you" was enough to reset that. It didn't need to explain the algorithm — just acknowledge that one existed.
In the Spaces community research, users were clear that a fully personalized feed would get boring. They wanted content tuned to their interests alongside content that surprised them. Pure relevance without variety removes the reason to browse. What surprised me most was that users didn't want maximum personalization. They wanted relevance, but they also wanted discovery.
Carousels were valued for variety — seeing creators, communities, and templates in one row. But appearing too often, they broke the scroll rhythm. Solopreneurs especially found them distracting when they showed up back to back with content cards.
When users encountered Try, Edit, and Remix in the layout prototypes, they used them. A save-only CTA had trained users to treat Picsart like a passive mood board. Surfacing creative actions in the feed reframed it as a starting point for creation.
Each research cycle was designed to answer a single question. I worked alongside research partner Dr. Charlotte Pyatt-Downes across multiple cycles.
Cycle 2 tested three distinct prototypes with six users (three solopreneurs and three consumers). The differences weren't cosmetic. Each prototype represented a real structural decision.
Five of six users preferred two tabs. All preferred one-column over two-column.
Each decision had a real alternative I tested. I didn't reject options because they felt wrong, I rejected them because the research made the right answer visible.
Should the feed show more content at once, or focus users on one piece at a time?
Layout testing in Cycle 2 was the deciding input. Six users tested three different layouts. The combination of their stated preferences, the crowding problem at carousel intersections, and Picsart's own analytics data from a prior layout switch made the one-column choice clear.
Should Inspiring and Following content share one feed, or live in separate tabs?
The preference signal from testing was strong and consistent. What made it easy to commit was understanding why users wanted the separation. It wasn't about content type. It was about cognitive mode. When they opened the feed to get inspired, they didn't want to be pulled into the social layer. That distinction had been missed in previous approaches.
Should the feed surface creative actions (Remix, Try, Edit) inline, or keep a single save CTA?
The prototype testing made the gap obvious. Users described Picsart as a creative app, not a browsing app. But the save-only CTA was treating them like passive consumers. When creative actions appeared on the card, users reached for them. That was enough.
Each design decision connects directly to a documented user problem.
Algorithmic and social content served different user intentions but lived in one undifferentiated feed
Clear separation by intent — curated discovery in Inspiring, creator and community content in Following
Two-column layout combined with frequent carousels felt crowded and prevented content focus
One piece of content at a time, with a cleaner scroll rhythm and stronger focus per card
Users couldn't tell why content appeared — felt arbitrary and eroded trust in the feed
Transparent ranking signals help users calibrate what the feed is doing and why
Carousels were generic and unrelated to user preferences, making them feel like filler
Carousels matched to onboarding preferences and past behavior — relevant communities, templates, and creators
Save was the only prominent action; Remix, Try, and Edit were buried or unknown
Creative capabilities surfaced inline, turning browsing into an entry point for creation
Every major decision was tested before it moved forward. I didn't present a single direction and ask for sign-off.
Research partner Dr. Charlotte Pyatt-Downes collaborated across multiple cycles, contributing to research design and analysis in Cycles 2 and 3.
The hardest part of this project wasn't the design. It was designing for an audience that was still arriving. Picsart's prosumer base was growing, but the product analytics were still weighted toward the existing consumer majority. Making decisions for a future user segment, against data that reflected the past, meant leaning heavily on qualitative research when the numbers were telling a different story.
Four research cycles sounds like a lot. But each one was tight and targeted, designed to answer one question and move on. That cadence worked well, and it's something I'd bring to any content-heavy design problem. Waiting for certainty before testing is almost always slower than testing early and adjusting.
The one thing I'd push harder on is the learnability cycle (Cycle 4). Feature discovery for Remix and Replay was identified as a problem early. The multi-action CTA addressed visibility, but the deeper question — how do users come to understand what these features actually do — was never formally tested. That's the loose end I'd tie off first.