Winnie is a marketplace for child care and backed by a trusted community of parents and providers.
projectWinnie's home feed needed a redesign in order to help users search and discover local child care providers. Based on support tickets and App Store Reviews, it was clear users were confused by seeing a feed consisting entirely of parenting discussions. Instead users expected to find local daycares and preschools immediately after opening the app.
Although Winnie was making a big investment to collect local child care data across different markets, it was not being surfaced in the home feed. This created a negative experience for new users who were presented with a feed of discussions instead of affordances for local search and discovery. Monetization was also a consideration because child care providers pay to be promoted in search results. So additional advertising inventory could be created in the home feed.
My hypothesis was that If I added affordances for local search and discovery to the home feed then I could increase the number of place searches as well as place detail views. As a result, the home feed experience would match user expectations and paying partners could reach more potential customers.
results72% vs. 42% of new users completed a local search in their first app session. This +30% lift reflected that new users could search and discover local places faster, right from the home feed.
The number of place detail views did not meaningfully increase because the CTR from search results to place detail stayed flat. So even though users were creating more searches, they were not seeing results that matched their intent.
To inform my design decisions, I surveyed a random sample of 1K users who had used the iOS app in the past three months.
A majority of users said their goal in using Winnie was to find local child care and activities which made sense since Winnie's maine value prop is local data.
Only 18% of users said they wanted advice or support from other parents. This made me realize that I could push down UGC in the home feed since most users didn't find value in it.
A majority said they wanted to use search queries (14%) and explore a map (26%) to see a ranked list of local providers. This meant that shortcuts to create search queries and explore a map could potentially be added to the top of feed.
Q: What was your goal in using Winnie?
Q: What would be your ideal way to accomplish this goal?
Q: What result would best satisfy your needs?
Q: Would you intend to purchase any of the following?
At the top of feed, a user would be able to quickly create a search based on intent. Below search affordances, a user could browse local providers. The rest of feed would show parenting discussions and static resources.
I started off by exploring different ways to display search shortcut buttons which served two purposes. Not only could users see search results faster but they would also understand the variety of local data that was available for search.
Displaying "filtered" search buttons such as Montessori didn't make sense because we didn't have strong data collection in every major market. As a result, lots of users would end up in an undesirable low data or empty state.
Then I delved into different ways to help users discover local places and also manage places that they had previously viewed and saved. However due to backend engineering constraints, it was out of scope to show viewed places.
To increase CTR from feed to place detail, I proposed personalizing the places shown in the segmented control based on their search intent. During onboarding users can state what types of local providers they are looking for. The default state would show a mix of child care and kids activities.
The Nearby button launches a full screen map view with a carousel of places. As a user scrolls through the carousel, the associated map pin scales up to provide feedback. The other search buttons would trigger a search with results based on the category.
Below the search shortcuts, a segmented control would allow the user to toggle between featured and viewed places. The default state for the featured places section would show local child care providers based on the user's device location. Some randomness and a quality filter (description, reviews) was added so this section felt curated. If the device location couldn't be detected then the user's home location would be used. If no places are found, the radius would be expanded until the max limit is reached.
