© 2020 by Ben Roach. All images and logos owned by their respective brands.

Case study 03: Amazon

Reachable refinements

I've omitted certain information to comply with my Amazon non-disclosure agreement.

A little context

During my first two years at Amazon, I led the Search Navigation program for the retail website. This meant I was responsible for the end user experiences for things like filtering, browsing, and sorting search results to help customers narrow from a broad set of items to just those which meet their needs.

This case study looks at a newly launched feature for our customers in India.

Working backwards

When I first took ownership of this work stream, I began by prioritizing a backlog of projects that were rooted in customer needs by reviewing a database of existing customer frustrations. It was quickly apparent that the mobile filtering experience was the cause of a significant number of customer problems and usability issues.

I started digging into videos of customers interacting with the current experience and reading customer feedback that had been left on the website. This was a great way to identify areas where there were opportunities for improvement.

High-level takeaways

I learned that pain points with customers attempting to apply search filters are consistently among the most commonly cited shopper frustrations during mobile app studies. These frustrations boiled down to two core themes:

  1. Poor discoverability – many customers simply did not know where to find the filter link in the mobile experience.
     

  2. Filter usability – there were a myriad of usability issues in the current experience. These were more prominent on mobile web than on native app.

Search results (mobile web)

Filter menu (mobile web)

Diving deep into data

After familiarizing myself with the customer problems, I looked at all of the data I could find for the filter menu. Unsurprisingly, there were a small number of filters that were used at a much higher rate than the others, so I decided to prioritize making it simpler to apply those filters.

I also learned that filtering was much less common on mobile than desktop. This was likely because our mobile filters were hidden behind a menu, whereas options were easily accessible in the desktop experience.

It was also interesting to learn that customers who used filters were converting at a higher rate than those who did not use filters – although I would caution we cannot determine whether this is correlation or causation.

I had access to a number of highly useful internal user studies from Amazon user researchers. These reports helped me get closer to our actual customers, their needs, and how they were struggling with the current experience.

Additionally, I read the Baymard's Product Lists & Filtering to gain a deeper understanding of best practices when implementing filters for e-commerce. And yes, I read all 502 pages of the report which was very technical and dry, but worth it.

Surfacing choices

One way to improve a key problem with discoverability of filters was a solution which surfaces them on a filter bar, rather than hiding them behind a link.

Previous pattern with options hidden behind link.

Proposed pattern with most useful options surfaced.

Ergonomic reach

Beyond the known customer problems, I saw other opportunities to improve the experience. I noticed that our controls were in a difficult to reach area of the mobile screen, so I explored moving them to the bottom.

I also looked at ways I could improve the display and affordance of existing filters...
I worked with a researcher and put three iterations through usability. Overall customers were thrilled, but based on feedback I proposed a few UX simplifications.
Final solution

In the end, I landed on a final solution that addressed our two main frustrations – discoverability and usability – and based on usability testing I was confident in the UX. Some of the key improvements included:

  1. I introduced a new sticky bottom navigation bar that improved discoverability of filters by making them available all the time on the search page.
     

  2. I introduced a new bottom sheet paradigm which also gave users the ability to multi-select (something that was not previously available).
     

  3. I worked with engineering to increase the relevance of filters for some of our top queries and dynamically display the most useful filters directly in the bottom navigation bar.

India-first experimentation

I suggested and got alignment to test a proof of concept with our customers in the India marketplace. T​he top reasons I argued for this were:

  1. India is a strategically important country for Amazon.
     

  2. India is has a high rate of mobile-only shoppers, meaning mobile filter frustrations were more acute here.
     

  3. Our mobile experience for India is in English, but India recognizes 22 major languages meaning many customers are non-native English speakers. This can often mean they have more difficulty figuring out the right words to use in their search queries, and and by improving the filtering experience I could make it much easier for these customers to find the right items.

The results

This experiment moved all of the meaningful business and customer engagement metrics in the right direction. Site-wide revenue had a very significant increase and refinement usage increased +32.85%.

Case study 01: Sears Parts Direct

The quest for a better home page

Case study 02: Fiat Chrysler Automobiles

A modular campaign framework