ACE Case Study Operations Round 1

ACE Challenge 2015

Amazon Confidential Confidential

Preliminary Round: ACE Case Breakers Case Study: Operations Introduction It is a bright autumn morning in 2205 and Gaurav Maurya, SVP – Amazon Enceladus 1, is gazing outside the window and reminiscing about how Blue Origin 2 had transformed Enceladus into a bustling human colony in a short span of 50 years, much like some of the developing countries back on Earth. Also, continuing Amazon’s legacy, Amazon.en has become Enceladus’s most customer-centric customer -centric company in a short span of one year and customers love to buy all all sorts of products on it every day. Gaurav is scheduled to meet the Board of Directors back on Earth in 3 days to present the first year report. For this meeting, Gaurav has decided to focus on operations, and on how different verticals within operations are helping raise the bar on customer experience. Gaurav enlisted four of his team members to share their stories. As Gaurav prepares for the meeting, he has asked you to take a look at the reports that his team has put together and help him answer some open questions and come up with fresh ideas on how to solve the problems. Scenario 1: Like on Earth, Amazon’s goal is to offer Enceladus’ largest selection on Amazon.en. Deonn, who recently joined Amazon.en Amazon.en wants wants to accelerate accelerate the addition addition of new products in the catalogue. While Amazon Amazon already already provides tools for vendors to submit their items in the catalogue, the error rate remains high. On an average, vendors fail to create 15 products for every 100 products submitted. This is a huge opportunity miss for Amazon, especially since the plan is to add ~20 million items by the end of next year. Assume that each item generates ENC₹ 15 in revenue each year. Deonn wants to launch a service offering for Amazon’s premium ven dors and her goal is to reduce the error rate to 2%. Additional benefit due to this service is reduction in lead time for vendors to launch their product on Amazon to 30 days from the usual 60 days that it takes at present. She is proposing to develop a team of Catalog Specialists to support the premium suppliers in adding their products to the catalogue. She estimates that a Catalog Specialist can help submit ~200 items per day. A Catalog Associate typically earns

1

Enceladus: Enceladus is the sixth largest moon of Saturn Blue Origin is a privately funded company set up by Jeff Bezos

2

Amazon Confidential Confidential This case and the numbers published in this case are not representative representative of any Amazon businesses within India or in a ny other region where Amazon has offices or operations and should not be used for any public or media consumption.

ACE Challenge 2015

Amazon Confidential

ENC₹ 45,000 per year. Overall, her goal for FY 2206 and FY 2207 is to create ~3% (every year) of Amazon.en catalogue using this service. Can you help Deonn in preparing a business case by projecting revenue and cost for 2206 and 2207 and assist her in arriving at go or no-go decision? (Note: Please provide your answer in a tabular format providing a break-up of your revenue and cost calculations. Also, please make assumptions as required and clearly state them.) Scenario 2: On the launch day of Amazon.en, the analytics team noticed an interesting trend in the buying behavior of the Amazon customers in Enceladus - most of the orders being placed on Amazon were being done using compromised/stolen cards. This was the first time that Amazon was seeing such a massive fraud-attack right at the launch. While doing a root-cause analysis, it was found that the machine learning algorithm (used for predicting fraud orders) was not mature enough to capture these fraudulent buying-patterns. Nearly 10K orders were placed on Amazon on the very first day of its launch, and there was a chance that nearly 70% of them were fraudulent. Amazon.en has an annual-goal for minimizing the bad-debt received due to fraud. Hence, minimizing the loss from this fraud attack was pertinent without impacting genuine customers. A snapshot of the above situation at Operations is as follows: Machine Learning Algorithms had failed to detect fraud, hence the orders either had to be manually investigated for fraud or had to be passed without fraud-detection (Batch-Pass). Batch-Passing of orders (without detecting fraud) involved very high risk and could lead to huge losses for Amazon. On the other hand, manual investigation of orders for fraud would have reduced the probability of fraud to 0.5%. However, the team could manually investigate a maximum of 1,500 orders each day. This effectively meant that the remaining orders would have to b e batch-passed. Please see Appendix 1 for relevant data. Despite all these challenges, Amazon was able to handle this fraud situation effectively. Please answer the questions that Gaurav had scribbled on the report: 1.

What is the correct mix of the orders that were sent for manual investigations and the ones that were batch-passed? (Note: Please state all your assumptions clearly and also include the work excel sheet for the solution)

Amazon Confidential This case and the numbers published in this case are not representative of any Amazon businesses within India or in a ny other region where Amazon has offices or operations and should not be used for any public or media consumption.

ACE Challenge 2015

2.

Amazon Confidential

Based on the attached data-set, what is the maximum loss that Amazon could incur from this incident? (Note: Please state all your assumptions clearly and also include the work excel sheet for the solution)

3.

How can Amazon prepare better and prevent such an ev ent from happening in the future? (Note: Please limit your response to 200 words and try to support y our arguments with data points, if possible)

Scenario 3: Gaurav is working on launching the Mayday 3 feature for Amazon.en customers during the upcoming holiday season. This feature has proved to be very successful in other planets in terms of increasing customer delight. A new Mayday Team of Customer Service Associates (CSAs) needs to be built to handle these contacts. Amazon is hoping to sell 25K devices (with the new feature) during the holiday season. Mayday contacts per unit sold is generally very high in the beginning, because customers see a new button and have a tendency to see what happens when they press it; some customers also need help to register their devices. Since most of the gift boxes will be opened on King's Fest Day (Dec 26), the Mayday team needs to plan capacity for Dec 26. The average time a CSA takes to handle a Mayday contact is 10 minutes. Each CSA taking these contacts works 9 hours a day including all breaks. The employee is expected to take a 30 minute lunch break and 2 additional breaks of 15 minutes each. On an average, a CSA spends 80% of the working time available on calls with the customers and the remaining 20% time waiting for the calls. Mayday contacts inflow pattern of King's Fest Day in 2204 for Amazon.Ti 4 is given in Appendix 2. The Titan Retail Team sold 0.1 million Mayday enabled devices in the 2204 holiday season. Based on the above data, assuming that Amazon.en customers have the same Mayday calling behavior as customers in Titan and making reasonable assumptions, answer the following: 1.

How many Mayday contacts will Amazon.en get on Dec 26? (Note: Please answer in 1 table showing calculations and the final number)

2.

All of Saturn’s moons have 24 hour days. Based on your contact forecast, how many people will you hire for the Mayday team to take care of during the holiday season maintaining a 100% service level? Also, one CSA works for 5 days in a week. (Note: Please answer in 1 table showing calculations and the final number)

3

Read more about Mayday here: (Link) Amazon.Ti Operates on Titan. Titan is the largest moon of Saturn.

4


ACE Challenge 2015

3.

Amazon Confidential

Propose a staffing plan (no. of people in different shifts) based on the given data? Mayday Team works in 4 shifts: 6 AM to 3 PM, 10 AM to 7 PM, 5 PM to 2 AM and 9 PM to 6 AM. (Note: Please answer in 1 table showing calculations and the final number)

4.

In addition to hiring new people, what can the Amazon.en team do to handle this volume? ( Note: Please answer in 2 lines)

Scenario 4: Much like on Earth, the holiday season on Enceladus is a season for shopping. Amazon receives a high volume of orders from customers ordering gifts for their loved ones. To handle the sharp in crease in volume, Amazon’s warehouses (FCs5) need to be appropriately stocked with items to fulfill demand in close proximity to customers. Iriba, who manages the logistics network foresees a huge challenge for Amazon.en during the upcoming holiday season for inbound shipment transfers across different Amazon FCs. During the holiday season, there is a skew in demand across regions and as a result one FC may run out of a product while another may have that product in excess. This situation is countered with a reactive solution – transfer items across 6 major FCs. Warehouse transfers are expensive, and amount to ~15% of the money spent on transportation during the holiday season. Iriba's challenge is to select the right carriers for these warehouse transfers. She needs to select carriers for 9 major routes (lanes) to minimize the transportation cost without compromising on the customer experience. Also, the Retail Team must know about the availability of products to ensure that customers get their products as soon as possible. Please help Iriba come up with a carrier scorecard based on the following aspects: Each logistics provider (carriers) offers the contract rate that varies for different routes. This is the primary cost of transportation. The other cost is associated with a delay in delivering shipments to warehouses. Delayed shipments results in cost of labor re-planning at warehouses. One more component that results in extra cost is not having shipment visibility. Carriers are expected to send shipment status updates during transit. This helps Amazon track shipments, re-plan and communicate to customers in advance in case of exceptions. Carriers missing out on providing required status updates result in extra communication downstream that bears cost. For detail please refer to Appendix 3.

5

Amazon’s

warehouses are also known as Fulfillment Centers (FC)


ACE Challenge 2015

Amazon Confidential

Questions: 1.

Present a carrier scorecard in an appropriate table that Iriba can use readily during the holiday season (Note: Please provide your response according to the format provided in App endix 3).

2.

Iriba’s hunch is that cost of no visibility is grossly underestimated. Help Iriba identify at least 5 potential areas that should be considered to account for estimating the cost of no visibility. (Note: Please limit your response to five bullet points and no more than 150 words in total).

Appendix 1: Data for Scenario #2 Refer to sheet Worksheet for Scenario 2-ACE Case Study-Operations-Round 1 ‘

’


’


’


ACE Case Study Operations Round 1

Recommend Documents