After multiple years of working closely with Amazon’s advertising customers, we identified that the majority of their budgets (roughly 80%) were committed to offline media. Offline media that had no mechanism to track customers or performance metrics. Our advertising customers were frustrated with the limited budgets they had left to spend with Amazon, so we saw an opportunity in visual search that would leverage their offline media to drive online sales. We challenged ourselves to understand the visual search domain, reviewing studies on the impact of WeChat in China and previous attempts at unique, branded barcodes.
How can we design a barcode that users trust and intuatively understand? How can we separate Amazon from its competitors? We researched a variety of different barcodes types, and our competitor’s visual search codes. We dug deep into Amazon’s existing visual search knowledge to provide insight into the constraints of camera technology.
SmileCode was not an Amazon initiative, it was a passion project worked on in rare moments of downtime. I sketched ideas and mocked concepts to keep my colleagues engaged, inspired and motivated. We’d meet regularly to discuss ideas, ideate and strategise a plan to secure sponsorship from Amazon's senior leadership team.
Between myself and a design technologist, we produced multiple prototypes exploring how we could use Amazon’s proprietary platforms to generate unique barcodes and distribute them for offline media. This level of effort and commitment to the project got the attention of our A9 visual search colleagues and kick-started the essential collaboration between design and visual search engineers.
With internal sponsorship secured, ongoing conversations with patent lawyers and the support of our A9 colleagues, we held a design sprint to better understand camera technology limitations and define the visual design of the code, branding and naming.
We ran an internal survey and tested mocks with potential users. 80% noticed the code, 65% understood that that code would take them to somewhere on Amazon, and 25% figured out how to scan the code. These results showed us the initial barrier to adoption and informed the need for the code to be Amazon branded. After an unsuccessful day deliberating a name, we simply asked our test participants what they would call this unique code, and “SmileCode” was suggested multiple times.
A key learning from the sprint was to understand how mobile cameras interpret barcodes and differentiate between an active or inactive bit. We devised a masking pattern that would ensure each randomly generated code appeared balanced in it’s distribution of elements. With key aspects of the code now defined, the A9 team picked up the task of integrating SmileCode scanning into the existing Amazon camera app, that would ultimately drive users to products and product landing pages.
With Black Friday approaching I made contact with Amazon's PR team and negotiated the opportunity to test SmileCodes at a pop-up event in Soho Square. The event hosted 2,450 visitors and achieved 2,695 search sessions – of which 887 users successfully used the camera app and were redirected to a URL, resulting in a scan through rate of 36%. The inclusion of SmileCode was complementary to the event, but it's presence received positive press response and it was listed as a top 2018 trend by CARAT.
My colleagues and I worked closely with Amazon's legal team to produce a patent around the usage of SmileCode, which was approved by the USPTO in July 2019. We subsequently received Amazon Inventor awards for our efforts and SmileCode was considered a key offering within Amazon's suite of advertising products.
SmileCode maintained it's trajectory over the following year, partnering with magazines, product catalogues, Amazon boxes and lockers, successfully driving users to product and landing pages from the camera app.
