Pilters
Summary
At one point, filter usage on Android was low, and required a number of taps into the filter menu to utilize. After an initial win by exposing price filter as pills, also known as “pilters,” we ran a number of subsequent experiments exposing additional pilters on Search. Our hope was that by making filters more prominent, buyers would more effectively find the items they’re looking for.
After experimenting with a number of pilter assortments, we finally saw a $2.3 million win and a 1.14% lift in conversion. We’d go on to make pilters sticky to the top of Search results, which was a $200k win, 5% engagement lift, and 0.6% lift to conversion.
Company
Etsy
Date
Oct. 2022-Feb. 2024
My role
UX/UI Prototyping Strategy Research
The challenge
Because the research tells us… buyers are interested in using filters, but find the filters unreliable and complex to configure
We believe that… improving the functionality and options of filters
Will result in… buyers finding items that meet their criteria faster
In early 2023, filter usage on Android was low and research suggested that Android buyers were more sensitive to price than their iOS counterparts. As a result, we saw a win by exposing price filters on the search page as “pilters” (pill filters), to cut down on the number of taps required to access filters, increase usage, and ultimately increase conversion.
Between this win and the wins of other teams, we saw Filter work lead to large incremental GMS wins over 2023 and 2024. That being said, we believed there was lots of opportunity to expand filters on the Android surface.
Solution
We believed by making filters more prominent and easier to use, buyers would be able to find the items they want to purchase easier and more quickly.
In the first few iterations, we played with a number of variables to expand our pilter set:
- Updated pilter UX to align with iOS and new design system guidelines
- We learned that iOS saw a huge ($9.2 million) win moving by moving "Free Shipping" to the front of the pilter offering
- We later added estimated delivery date (EDD), despite deprioritizing it early on (because it isn't available internationally and research suggested buyers don’t totally trust the dates)
- After a $3.8 million iOS win by suggesting the “Personalizable” pilter to buyers, we pivoted to try moving the "Personalizable" pilter to the first slot, as well as moving active pilters to the front of the list for continuous visibility
In the final iteration, we made the following changes:
- We moved the price filter to the first spot, which allowed the other pilters to “shine”
- Historically, we know Android buyers are particularly sensitive to price (and more specifically, value), so placing the price option first was critical to the success of this feature
After our enhanced pilter win, we learned that iOS saw a $4.5 million GMS win by making pilters sticky when at least one pilter was applied. We subsequently tested the same pattern, which led to a similar win of $200k annually. The goal was to keep only relevant search filters closeby.
Key learnings:
- The positioning of pilters matters. By prioritizing one pilter, we are by definition deprioritizing another
- We have continued our experimentation around pilters, such as Etsy’s Pick (inspired by an iOS win with Etsy’s pick as the third pilter option), which was not a win
Introducing an "item format" pilter that allowed buyers to filter digital and/or physical items, allowed users to better toggle between item types they are looking for. We believe the pilter allowed users to limit search results to physical items, boosting relevance and item price.
Results
After four initial tries, we finally realized $2.3 million Annualized GMS (gross merchandise sales) win and a +1.14% conversion lift.
- We believe moving the price filter to the first slot was the main driver of GMS (gross merchandise sales)
- The benefits of the other pilters was being overshadowed by deemphasizing the price filter; by moving the price filter up, we then saw the increased GMS impact from the other pilters
- We saw a decrease in percent visits with a filter BUT an +18% increase in mean visits with a filter
Making pilters sticky when at least one pilter was applied resulted in a +$200k annual GMS win. We also measured a 5% engagement lift, driven by lifts in usage for each pilter type. Given a 0.6% lift to conversion rate, it's likely these sticky pilters helped users better manage their search mid-search.
Finally, introducing a pilter to see digital and/or physical items was a $1.0 million win and generated a GMS lift driven by increased average purchased item price (+1.5) and a +1.3% lift to ACBV. In addition, there was an increase in the average number of times any filter was applied.