FLICKR · GROWTH DESIGN
Reviving Communities
Onboarding, first-week milestones and group health: a three-part strategy to help new users find, join and stay in the communities most tied to Pro subscriptions.
ROLE
Lead & sole product designer
TEAM
1 PM, 1 engineer
SCOPE
Sign-up onboarding, first-week retention, group discovery
STATUS
Onboarding shipped. Milestones designed and handed off.

IMPACT
13x
increase in 7-day activation for new users
+73%
more group joins from new users after launch
+35%
more awareness of groups and how they to use them
01 — THE INSIGHT
Community members were 3x more likely to subscribe. New users never found their community.
People active in Flickr Groups were three times more likely to subscribe to Flickr Pro, yet new users weren’t joining groups. Digging in, the problem had three layers.
They didn’t get it
New users didn’t understand what groups were or why they should join one.
They couldn’t find them
Groups were buried and hard to search or browse, so there was no obvious way in.
Many were dead
Thousands of inactive groups made discovery feel abandoned, even when users did look.
02 — THE STRATEGY
One problem, three moves
Because the problem had several layers, my PM and I proposed a phased plan: build a mental model of groups from the first minute through the first week, and fix the supply of good groups underneath.
PART 1 · ONBOARDING
Introduce groups at sign-up
Personalized group recommendations based on the interests people pick.
Shipped
PART 2 · MILESTONES
Reinforce in the first week
A lightweight checklist that nudges the actions that predict retention.
Designed
PART 3 · GROUP HEALTH
Surface only living groups
Hide dead groups, badge new and popular ones, and archive legacy collections.
Designed · in development
03 — PART 1: ONBOARDING
A three-step onboarding built around groups
Flickr had no real onboarding. I designed a short flow: pick your interests, join recommended groups, then set up your profile. Each step feeds the next. Interests power the recommendations, and joined groups fill the home feed from day one.

Step 1 · Choose interests with photo tiles

Step 2 · Join recommended, active groups. Step 3 is a quick profile setup.
Key decisions
01
Cut the welcome tour
Our first version opened with a “Welcome to Flickr, here’s how it works” screen. It had a big drop-off, so I removed it and got people into the product faster. Experimenting and reviewing data with product leaders helped me understand what solutions could be proposed.
02
Photos, not bubbles
Early designs used text chips for categories. For a photography community that felt generic, so I switched to photo tiles: more on-brand, more inviting and easier to scan. We wanted new users to know how much Flickr values photography.
03
Recommend only living groups
I worked with engineering on the recommendation logic so new users only saw active, relevant groups. A dead group on day one contribute more to our retention issue.
04
Treat skip as a lever
After launch we saw drop-off at the interests step. I added a skip option and adjusted when it appeared as data came in, balancing completion against personalization.
05
Ship on a modern stack
Instead of extending the legacy sign-up code, we built onboarding in Next.js. That let a team of three ship faster and keep iterating after launch.
Select interests
Abstract art
Animals
Pets
Wildlife
Food & drink
Home & garden
Music
Nature
Landscapes
Sports
Travel
Street
Film
Cameras & gear
Architecture
Before · generic text chips

After · photo tiles that feel like Flickr
RESULT
Onboarding shipped and drove a 13x increase in 7-day activation. Group joins and group retention rose. But joins alone weren’t enough: we needed new members to keep coming back.
04 — PART 2: MILESTONES
Turning a first join into a first-week habit
Onboarding gave new users a mental model of groups. Milestones builds on it during their first week with a small, dismissible checklist of the actions that unlock Flickr: join more groups, post to a group, complete your profile, leave feedback.
Aligning the team first
Before designing, I led a brainstorm with the Growth team to agree on what makes milestones feel helpful, not intrusive. We gathered patterns from products we admired, then grouped them into core values, each tied to a stage of a new user’s journey.
DAYS 0–2
Belonging
Make the first action easy. Don’t ask for too much up front, and connect people through shared interests.
DAYS 3–4
Validation
Prove the action mattered, through natural nudges, new content to engage with, and feedback on their photos.
DAYS 5–7
Momentum
Create reasons to return: a progression journey, challenges, and others’ achievements as signals to keep going.
DAY 7+
Pro value
Introduce Pro as acceleration, not interruption. Uploading more becomes a natural reason to upgrade.
Each milestone maps to a value: joining groups builds belonging, posting to a group earns validation, visible progress creates momentum, and the Pro offer arrives once the value is clear.

From the working session: patterns the team gathered, sorted under each core value

Collapsed on the feed: visible, never blocking

Expanded checklist with the next step highlighted
Helpful, not in the way
No pop-ups or ads. A compact widget that collapses, snoozes for 24 hours or can be dismissed for good.
Lead with the payoff
Every step says why it matters, like “Photos in groups get 3x more views.”
Built on onboarding
Suggestions build on the interests and groups people chose at sign-up.
CONSTRAINT
Designing around legacy systems
Flickr couldn’t recommend people or photos, so I couldn’t write steps like “like a photo you’ll love.” I scoped milestones to actions the platform could reliably support and kept the flow short.
EXPERIMENT
Testing the payoff
Finishing every milestone triggered either a celebration moment or a 10% Pro offer for a test segment, so we could learn what actually moved retention and conversion.
STATUS
Designed, specced, handed off
I designed every state, including snooze, dismiss confirmation and completion. Using Figma to hand off to engineering for build and QA.
05 — PART 3: GROUP HEALTH
Fixing what users find when they get there
Recommendations only work if the groups are alive. Collaborating with engineering and PM, we defined rules to categorize every group, then designed badges and placements across discovery and search.
Inactive
No activity in 2+ years and no admin
Hidden from search and discovery
New
Recently created and active
Surfaced so newcomers can join early
Popular
High recent activity
Badged across discovery and search
The Archives
Inactive, but holds legacy photography like national parks
Preserved on a new Archives page I created, since these photos still bring people in from search engines
06 — WHAT I LEARNED
Joins are a start, not the goal
What was next
Onboarding moved group joins, but a join isn’t participation. The next phase I was designing added lightweight ways to take part in groups, like reactions and likes in discussions, so a new member’s first contribution feels low-stakes.
What I’d take with me
The biggest gains came from designing a user journey, not a screen. Connecting sign-up, the first week and the health of the groups themselves into one journey is what made each piece work.
NEXT PROJECT