### Final Grade & Feedback Q1: 15/15 Q2: 15/15 Q3: 15/15 Q4: 15/15 Q5: 10/10 Bonus: 0/10 [Only censoring mentioned, no distribution assumption] **Total: 70/80** Yedioth Case Report 15.778 Introduction to Operations Management Sloan Fellows Section A Group T-Rex Alecia Asiamigbe Rahul Bandekar Ido Levy Christie Lucagbo Marco Musazzi July 31, 2025 1. In the current distribution model, where each retailer is supplied once a week independently of all other retailers, what would be a good method to compute the quantity shipped to each retailer to guarantee that 99% of customers will be served? Apply your approach to compute recommended quantities to the 50 retailers (explain the methodology in the body of the report and provide the results in appendix). We took the following approach: • First, we grouped the data by retailer using the values in the column “Customer Number” (from retailer 1 to retailer 50). Then, we calculated the expected value (or mean) of the weekly demand per retailer using the historical data in column “Sales”. • Next, we calculated the standard deviation of weekly demand per customer • We calculated Q*i or the target quantity to be shipped to each retailer i per week at 99% service levels and rounded up the quantity: Q*i for retaileri = Meani + (K* StDevi ) where: K = 2.326, which is ~Z-score for 99% service levels • We calculated the estimated Returnsi for each retailer i, which is given by 2.326*StDev i and rounded up the quantity • Finally, we summed up (1) Q*i and (2) Returnsi for each retailer i=1 to 50 to arrive at the total production requirements and total expected returns to meet 99% service levels. This gave us the following results: • Total Production Quantity without pooling: ~🚨419 units🚨 • Projected Returns without pooling: ~213 units Note that we are 🚨underestimating the expected value of demand, because we do not know what the actual value of demand would’ve been for weeks where a retailer had a sell-through🚨 (this number can be estimated through statistical modeling but would require additional assumptions on the demand curve shape). See Appendix A. 2. If Yedioth could implement full pooling among all of the 50 retailers what would be the estimated benefit in terms of total production levels and returns if the required service level is 99%? (Note: Full pooling means that somehow all of the retailers could be supplied in-real-time from the same pool of inventory.) If Yedioth could implement full pooling among all 50 retailers, then we would take the following approach to calculate total production quantity Q* at 99% service levels: • Calculate the expected value (or mean) of total weekly demand by grouping the data in column “Sales” on a weekly basis • Calculate the standard deviation of total weekly demand • Calculate the Q*: Q* = Mean + (K*StDev) where: Mean = 189.59, K = 2.326, StDev = 19.99 • Calculate estimated Returns, which is given by 2.326*StDev This gives us the following results: • Total Production Quantity Q* with full pooling = ~🚨237 units🚨 • Projected Returns with full pooling= ~47 units Therefore, by implementing full pooling, Yedioth could reduce total production quantity by ~182 units (419 units minus 237 units) and reduce returns by 166 units (213 units minus 47 units). See Appendix B. 3. Suppose that one could implement full pooling only among retailers that are treated by the same sales agent. What would be the potential benefit in terms of production levels and returns, assuming 99% service level. Compare to your #2 answer. Assuming that Yedioth could implement pooling only for retailers handled by the same sales agent, then we would take the following approach to calculate total production quantity at 99% service levels: • Group the data by sales agent using the values in the column “Sales Agent” (from sales agent 1 to 10). Then, we calculated the expected value (or mean) of the weekly demand of all retailers handled by each agent using the historical data in column “Sales”. • Next, we calculated the standard deviation of weekly demand per sales agent • We calculated Q*j or the target quantity to be handled by each sales agent j per week at 99% service levels: Q*j for sales agentj = Meanj + (K* StDevj ) where: K = 2.326, which is ~Z-score for 99% service levels • We calculated the estimated Returnsj for each sales agent j, which is given by 2.326*StDev i • Finally, we summed up (1) Q*j and (2) Returnsj for each Sales Agent j=1 to 10 to arrive at the total production requirements and total expected returns to meet 99% service levels. This gives us the following results: • Total Production Quantity for pooling per Sales Agent only = ~🚨287 units🚨 • Projected Returns for pooling per Sales Agent only= ~97 units If Yedioth could only implement pooling among retailers handled by the same sales agent, then total production quantity increases by ~50 units (287 units minus 237 units) and total returns increase by ~50 units (97 units minus 47 units) vs if Yedioth could implement full pooling across all 50 retailers. Pooling by sales agent gives partial benefits, i.e. better than independent, not as efficient as full pooling. See Appendix C. 4. Propose more realistic processes/strategies that leverage the fact that the sales agent visits each retailer in the middle of the week. What would the benefit be of these processes/strategies? • Only deliver safety stock for the first half of the week, and then use midweek visits to deliver safety stock for the second half of the week based on actual sales from the first half. Presently, only 6% of all weekly orders had added quantities beyond what was originally distributed (127 of 2108). This indicates that the initial distributed amount is usually sufficient to serve as safety stock for the entire week. Let X = demand for first half of the week and Y = demand for second half of the week. When we allocate safety stock for the entire week upfront, we are protecting against the combined uncertainty of total weekly demand, or allocating for the Variance of (X+Y) combined. However, if we already know what the value is for the first half of the week (during the sales agents’ visits), then there is no uncertainty in the value of X, and we only have to estimate for Y. Since the total uncertainty in the second half is less than the uncertainty we face before the week began, the safety stock required for the second half will always be lower than what we would have needed had we forecasted the full week in one go. This is why, by stocking only for the first half of the week upfront and allocating additional inventory midweek to meet the variance for the second half, we can expect returns to be lower and sellouts more unlikely. This helps Yedioth lower costs associated with collecting and transporting surplus magazines from the retailers, especially if there is not much “seasonality” in sales through the week. • Shared safety stock per sales agent. We already know that the required safety by pooling together the demand of all retailers under the same agent is lower than preparing safety stocks for each retail store. Rather than allocating safety stock for each store, Yedioth can allocate safety stock for each agent and train agents to understand how best to deploy their safety stocks between the stores. 5. What do you think are the organizational challenges that Assaf will have to address? • 🚨Change resistance from agents🚨. Sales agents will be worried about the overall lower safety stock. They may interpret this as an increased risk of missing out on sales, and therefore missing out on their sales incentives. Assaf will need to spend considerable time explaining to the agents why the pooling scheme (and mid-week replenishment scheme) are better. Assaf may also want to think about introducing incentives to agents for not overstocking. • Change resistance from retailers. Likewise, retailers may worry that lower inventory levels delivered to them might mean an increased risk of missing out on sales, and therefore missing out on profit. • Information management. Optimizing the sales process would require the company to create a well-functioning information management system that constantly and accurately informs distribution decisions. This may potentially pose a challenge for a small business relying on traditional ways of communication. Furthermore, an information system might be costly and/or impractical (training/time). Appendix A (Question 1) Appendix B (Question 2) Appendix C (Question 3)