fbtechreview.comAUGUST 20198in myviewAt Amazon, we strive to be Earth's most customer-centric company, where customers can find and discover anything they might want to buy online. As we grow our selection and business to delight our customers, we also encounter complexity as we operate as a brand owner, a retailer, or as a supplier. We solve this complexity by constantly innovating on behalf of our customers to better serve and protect them, and we have implemented technological solutions within food safety for relatively straightforward tasks like slicing deli meat, as well as complex tasks like ingesting customer feedback and executing food recalls. We look at a high-risk process or event and ask how we can eliminate the risk while driving innovation that reduces the chance for human error.Listening To Our Customers at ScaleWe invest heavily into mechanisms to listen to our customers and detect when something has gone wrong. We make data-driven decisions but also respond rapidly to customer anecdotes. On the rare occasion when data and a customer anecdote disagree, we work on a solution with the tenet that our customer is right. Through automation, we aggregate 30 million pieces of customer feedback a week globally in over 40 languages. These interactions include customer contacts or feedback data such as product reviews, customer return comments, Customer Service chat, social media account, etc. We have leveraged LEVERAGING TECHNOLOGY AND AUTOMATION TO KEEP OUR CUSTOMERS SAFEautomation to handle both the scale and the data extraction challenge. We rely on Machine Learning (ML) and complex software- based logic to understand context across languages and unlock the meaning of customer comments without the need for users to review them one by one.This first pass of `labeling' customer interactions is critical to separate the true signal from the large amount of customer interaction data. We then back it up through further automation in the form of ML to determine the feedback's relation to a food safety concern. When the ML judgment has high confidence, we automate the action, otherwise, we rely on human Subject Matter Expert (SME) review. Our SMEs have accurately assessed safety risk hundreds of thousands of times over the last several years, and all outcomes of their judgment is then utilized to further enhance our ML automation efforts.Instituting Technology to Ensure Food Processing StandardsSlicing deli meat is a relatively straightforward task, however, the physical handling of ready to eat food is one of the highest risk process control points from a food safety perspective. We implemented a physical process control by leveraging software logic in our deli slicing operation to ensure our customers receive safe, correct product, at the correct BY CARLETTA OOTON, VICE PRESIDENT, HEALTH AND SAFETY, SUSTAINABILITY, SECURITY & COMPLIANCE, AMAZON
< Page 7 | Page 9 >