The Problem

Typing is intensive and consumers have started dropping off the site before finding the item they are looking for.
‍
Image recognition for product search has not been implemented by any beauty competitor. We have the opportunity to implement a successful tool where other brands have not.

Develop a tool that will allow consumers to identify BECCA's products by taking/uploading a photo of a product.

We began by conducting a landscape analysis to get a deeper understanding of the market, our competition, and what opportunities we could take advantage of.

We then defined how to measure the success of our goals with some KPIs.

What percentage of the consumers use image recognition vs. text search?

Measurement: Number of taps to camera icon vs. # of typed searches
Success: 6-Month Pilot period: 1-5% of searches are image recognition

What percentage of users add to bag with text search, image search, and without using search?

Measurement: Percentage of users that used image recognition that add to bag  from the results page
Success: Percentage of users that are adding to bag using image recognition vs. text search are on par with one another

Do users complete the image search after tapping the camera icon?

Measurement: Number of users who tap the camera button vs. the number of users who complete the image recognition steps
Success: 70% of users complete the image recognition steps

Does the image recognition capability provide consumers with results she’s looking for?

Measurement: Number of taps to: ‘Try Again’ button/ top 4 search results in search results page.
Success: Gain learnings on Google’s APIs accuracy in matching consumer images to Becca’s products and accuracy of training data

I then conducted initial user testing to get a background understanding of our users existing habits and routines. I used this data to help define our case studies and have a deeper connection to our persona.

I collaborated with the UX lead and we conducted a second round of user testing to validate our assumptions and found these key insights.

We begin our design process by creating a user flow and customer journey.
‍

Once we collect insights from testing, we validate our assumptions and begin high-fidelity designs.

Our Solution

Although consumers find the concept and feature of image recognition search to be beneficial, they had trouble finding and accessing it.

• Users were mistaking promotion of the feature for a product ad
• Users were dropping off when products could not be found via image they submitted
• Users expressed confusion if search was not successful
• Users are so focused on searching for a product they are used to typing it in out of habit and do not think to search via image.

Our post-launch insights revealed to us that there was more we could do. to improve our metrics we will:

• Increase awareness of the tool by including visual cues on the brand site.
• Run user tests on new iterations and proceed for v1.2 launch.

The goal of testing BECCA’s visual search tool post-launch is to understand areas where users are dropping off. These learnings will be incorporated into future executions of visual search with other ELC brands, as this feature transitions to Product Development.  Becca’s image recognition (IR) search feature has recently been launched and we want to discover opportunity spaces that we can take advantage of to improve engagements and awareness of the feature. 

After running our tests, we analyzed the data to come up with an iterated design as a solution to the drop offs and confusing users were facing.

I ran a sketching session and began to design some wireframe sketch solutions which then evolved into hi-fidelity designs.