### Final Grade & Feedback Q1: 15/15 Q2: 15/15 Q3: 15/15 Q4: 15/15 Q5: 10/10 Bonus: 5 [Censoring discussion in Method 3 + Distribution analysis in Method 2] **Total: 75/80** --- 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III The Yedioth Group Case Questions 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. Methodology 1: With simple assumptions: Current distribution model requires Yedioth to estimate the distribution of the “weekly demand” of each retailer individually. To estimate the distribution, we made two assumptions: 1. Weekly sales are independent observations; 2. The sales figures are the best representation of demand; 3. For each retailer, the weekly sales follows normal distribution With the assumption, we will calculate the supply that guarantees 99% service level. That is: P (D<=Supply) = 99% Given D follows normal distribution, we have Supply_i = mu_i + 2.32 * sigma_i. Where Supply_i is the number of magazines to be distributed to ith retailer, mu_i and sigma_i are mean and standard deviation estimated with the observed sales records from previous years. Using the method, we calculated the distribution of magazine (see detailed breakdown in appendix I). In total, we need 419 magazines per week. 🚨correct ~419🚨 --------------------------------------------------------------------------------------------------------------------------Note: Method 2 and Method 3 are 2 different approaches we tried, in order to relax the assumptions made in Method 1. Due to time constraint, we didn’t solve the final answer for Method 2 and didn’t validate the final number for Method 3. Methodology 2: Violation of “normal distribution” assumption In reality, demand (for simplicity, we assume demand = sales) doesn’t follow normal distribution. In fact it might even have some seasonality. Here we will ignore the time effect on the demand. 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III For example, using the histogram of retailer 1’s weekly sales as the proxy of PDF (the bar chart), CDF (the line chart) of the demand’s distribution: Given the histogram, using the empirical distribution, we can estimate that supply of 8 could provide 99% service level (approximately). Using this method, we estimated the supply per each retailer (see appendix II), in total weekly supply is 370 🚨empirical distribution analysis🚨 (see detailed breakdown in appendix II). Methodology 3: Violation of assumption “The sales figures are the best representation of demand;” When the retailer has a “sold-out”, we don’t know if the full demand is met, because there might be customers that haven’t been fully served. So the observed sales numbers are not the actual demand (lower or equal to). 🚨censoring awareness🚨 In this case, we will not know the distribution of the actual Demand, thus cannot use observed data to estimate the ideal supply at 99% service level. Meaning, we won’t know the actual mu of demand, and std of demand. Below is how we think the problem can be tackled with some “additional” assumptions who are less critical as the assumption of “sales = demand”. 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III At high-level, we will use the current “service level” to estimate the actual mean of demand: Under the situation of “sold-out”, Supply <= Demand, in the situation of “return”: Supply > Demand. That means, through the observation data, we know empirically: P (Demand <= Supply) which is equal to 1 – “Sell Through rate” Here we apply two assumptions: • • Actual demand follows normal distribution We use variance of the observed sales as the proxy for the variances of demand Then we have: P (Demand <= Supply) = P ((Demand – mu) / sigma <= (Supply – mu) / sigma) = P(Z<= (Supply – mu) / sigma) Here, sigma = variance of observed sales for each retailer, Supply = actual supply which is the Distributed + Added. Given, P(Z<= (Supply – mu) / sigma) = (1-sell through rate) => (Supply – mu) / sigma = Φ⁻¹(1 sell through rate) => mu = Supply - sigma * Φ⁻¹(1 - sell through rate, mu, sigma) We can solve mu with the function above (however we don’t know how to calculate mu with excel so we didn’t finish the calculation), as all the parameters above are known except for mu. With the adjusted mu, we can calculate Supply to serve at 99% service level = Supply - sigma * Φ⁻¹(1 - sell through rate, mu, sigma) + 2.32 * sigma 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%? For simplicity, the calculations for Q2 and Q3 will use: Methodology 1: With simple assumptions. Under this perfect pooling scenario, we will be able to pool all the weekly observations from 50 retails to estimate the distribution of weekly demand (in total). We first calculated the weekly total sales, then calculated the mean and standard deviation for the weekly total sales. With the statistics, we can calculate the number of magazines to be distributed to the entire 50 retailers (as one pool). Below are the statistics and the final answer: 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III Mean of the weekly total sales: 189.59 Standard deviation of the weekly total sales: 19.99 Solving the equation: P(D<=Supply) = 99%, where D is the total demand (across all the 50 retailers) and Supply is the total supply to the entire 50 retailers, We have Supply = mean(observed demand) + 2.32*std(observed demand) = 235.9600782 (236 rounded). 🚨correct pooling🚨 Comparing the result from question 1: 419, we concluded that using the “perfect pooling”, we estimate the result through improved prediction of variability. Therefore we can reach the same customer service level with way lower production levels,, therefore less returns of unsold goods (we can compare the results actual sales, which is about 190 per week, the new estimation is much closer to the 190). 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. With the new setting, instead of 1 pool of 50 retailers, we will end up with 10 pools of 5 retailers. The methodology of calculating supply per agent (or per 5 retailers managed by agent i) stay the same as 2. By calculating the mean and standard deviation of weekly total sales, of ith pool (5 retailers managed by agent i), we got the supply to ith pool. In total, the 10 pools (50 retailers) require 293 magazines per week. 🚨acceptable agent pooling🚨 (see details in Appendix III) Although method 3 requires more magazines, 287, the increased number of magazines (287236=51) is manageable, especially when there’re 10 agents in total (so 5 magazines wasted per agent). Operational wise, method 3 is also more doble than method 2. Potentially method 3 has a lower operational cost than method 2 as well. So if full pooling is achievable among retailers managed by the same agent, method 3 is the best choice among 1-3. 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III 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? • • • • • Delayed Two-Phase Distribution: Initiate a first shipment based on projected demand. During a mid-week visit, the sales agent assesses on-site inventory levels and sales performance. A follow-up delivery is then made to replenish fast-moving or sold-out items. Enhance Tight Feedback Loops and Early-Week Sales Data Reporting: Sales agents report real-time or near-real-time sales data early-to-mid week to a central dispatch system. This will trigger targeted replenishment based on actual demand trends. Predictive Restocking Using Historical Sales Data: Leverage historical sales trends, seasonality, and regional demand patterns to anticipate inventory needs in advance. This data-driven approach enables proactive shipment planning and minimizes both stockouts and excess inventory without requiring real-time field updates. Pilot a Vendor-Managed Inventory Model with Key Retailers: Let sales agents directly manage their order quantities based on past and midweek sales data. This would transfer forecasting responsibility to those who have the best real-time knowledge. 🚨multiple specific mechanisms🚨 Segment the Retailers Based on Sales Patterns: Use mid-week visits 5. What do you think are the organizational challenges that Assaf will have to address? • • • • • Cultural Resistance and Change Management: Addressing resistance to shifting longstanding distribution and production practices, particularly within a traditional, familyrun business culture. Employee and Agent Buy-In: Convincing staff to move away from buffer inventory and personal judgment toward data-driven decision-making over intuition or reliance on excess inventory will be critical. IT Investment: Deploying systems for sales tracking, mid-week data reporting, and inventory visibility across the supply chain. Process Redesign: Realigning roles, workflows, and incentive structures to support multi-phase distribution and pooled inventory strategies. Collaboration Barriers and Resistance to Change: Building trust and information sharing among decentralized retail partners who may be hesitant to engage with centralized planning efforts will be essential. 15.778 Introduction to Operations Management • Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III Lack of Retailer Incentives: Retailers currently have no incentive to improve their forecasting accuracy or reduce returns. This perpetuates a culture of over-ordering and inefficiency. New penalty / incentive structures are needed. 🚨multiple stakeholders and barriers🚨 Appendix I: Retailer # Count of Count of Months Average of Sales StdDev of Sales Date (Date) Rounded Supply at 99% SL Supply 1 2 4.266666667 5.447368421 1.763261441 1.940985657 45 38 45 38 8.35743321 9.950455145 9 10 3 4 5 6 1.608695652 1.804347826 3.086956522 3.608695652 0.744707575 1.185459409 1.457945083 1.731771872 46 46 46 46 46 46 46 46 3.336417226 4.554613655 6.469389115 7.626406395 4 5 7 8 7 8 9 10 11 3.926829268 14.43478261 4.733333333 3.173913043 4 1.602969805 3.913805616 1.615268061 1.371201384 1.862951485 41 46 45 46 35 41 46 45 46 35 7.645719216 23.51481164 8.480755235 6.355100254 8.322047445 8 24 9 7 9 12 13 14 15 16 4.066666667 2.244444444 2.456521739 4.56097561 6.52173913 1.009049958 1.111010096 1.129533271 1.581909929 2.008195768 45 45 46 41 46 45 45 46 41 46 6.40766257 4.821987867 5.077038928 8.231006645 11.18075331 7 5 6 9 12 17 18 19 20 21 22 23 24 25 26 27 4.782608696 1.860465116 5.847826087 3.25 2.782608696 4.195652174 5.136363636 1.565217391 8.739130435 3.152173913 2.162790698 1.698535886 0.804197185 1.763423329 1.740422296 1.009137002 2.145999536 2.097516905 1.147987106 2.38918659 1.475238456 1.703361279 46 43 46 44 46 46 44 46 46 46 43 46 43 46 44 46 46 44 46 46 46 43 8.723211952 3.726202585 9.93896821 7.287779727 5.123806541 9.174371098 10.00260286 4.228547477 14.28204332 6.574727131 6.114588865 9 4 10 8 6 10 11 5 15 7 7 28 29 6.652173913 1.942857143 3.107105772 1.258917769 46 35 46 35 13.8606593 4.863546367 14 5 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III 30 31 32 33 34 35 8 4.804347826 2.586956522 3.673913043 2.217391304 3.953488372 2.683281573 1.939296152 1.239643084 1.606297991 1.298084803 1.661231448 46 46 46 46 46 43 46 46 46 46 46 43 14.22521325 9.303514899 5.462928477 7.400524382 5.228948047 7.807545331 15 10 6 8 6 8 36 37 38 39 40 2.673913043 3.608695652 3.404761905 3.043478261 9.065217391 1.549972728 1.58434369 1.083344501 1.413530202 2.506898694 46 46 42 46 46 46 46 42 46 46 6.269849772 7.284373013 5.918121147 6.32286833 14.88122236 7 8 6 7 15 41 42 43 44 45 3.695652174 3.735294118 3.065217391 3.375 1.47826087 1.473764118 2.093407018 1.466699604 0.916125381 0.936640105 46 34 46 8 46 46 34 46 8 46 7.114784928 8.5919984 6.467960472 5.500410884 3.651265914 8 9 7 6 4 46 47 48 49 50 6.352941176 5.512195122 1.428571429 2.928571429 4 1.920895513 2.026350799 1.065107404 1.268814451 1.224744871 34 41 35 14 25 34 41 35 14 25 10.80941877 10.21332898 3.899620606 5.872220955 6.841408101 11 11 4 6 7 Grand Total 4.137096774 2.939137332 2108 2108 393.3001803 419 Appendix II: Retailer Supply estimation using CDF: Supply = 99 percentile of Customer Number Sales CDF Percentile Count <= Value Total Observations P_99_Flag 1 2 3 4 5 8 10 3 5 7 1 1 1 1 1 100 100 100 100 100 45 38 46 46 46 45 38 46 46 46 TRUE TRUE TRUE TRUE TRUE 6 7 8 7 7 22 1 1 1 100 100 100 46 41 46 46 41 46 TRUE TRUE TRUE 15.778 Introduction to Operations Management Dolphins Study Group, Section A Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III 9 10 11 12 13 14 8 5 8 6 4 4 1 1 1 1 1 1 100 100 100 100 100 100 45 46 35 45 45 46 45 46 35 45 45 46 TRUE TRUE TRUE TRUE TRUE TRUE 15 16 17 18 19 7 10 8 4 10 1 1 1 1 1 100 100 100 100 100 41 46 46 43 46 41 46 46 43 46 TRUE TRUE TRUE TRUE TRUE 20 21 22 23 24 8 4 8 9 6 1 1 1 1 1 100 100 100 100 100 44 46 46 44 46 44 46 46 44 46 TRUE TRUE TRUE TRUE TRUE 25 26 27 28 29 30 31 32 33 15 6 6 15 4 13 9 6 8 1 1 1 1 1 1 1 1 1 100 100 100 100 100 100 100 100 100 46 46 43 46 35 46 46 46 46 46 46 43 46 35 46 46 46 46 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 34 35 36 37 38 39 40 41 42 43 44 4 7 6 6 5 6 14 6 9 5 5 1 1 1 1 1 1 1 1 1 1 1 100 100 100 100 100 100 100 100 100 100 100 46 43 46 46 42 46 46 46 34 46 8 46 43 46 46 42 46 46 46 34 46 8 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 45 46 4 10 1 1 100 100 46 34 46 34 TRUE TRUE 15.778 Introduction to Operations Management 47 48 49 50 Total Supply 8 3 6 6 Dolphins Study Group, Section A 1 1 1 1 100 100 100 100 Group Members: Hanyun Li, Daigo Ito, Ademir Xavier, Mamdouh Almutairi, Eze Burts III 41 35 14 25 41 35 14 25 TRUE TRUE TRUE TRUE 8 9 370 Appendix III: Agent 1 2 3 4 5 6 7 10 (blank) Grand Total Average Weekly Sales 22.43 15.83 12.80 17.61 28.04 22.17 11.22 15.65 27.98 15.85 Std Weekly Sales Supply at 99% SL Supply at 99% SL rounded 5.18 3.96 2.80 3.38 6.38 4.32 2.50 3.31 5.70 34.44 25.02 19.30 25.46 42.85 32.19 17.02 23.34 41.21 35 26 20 26 43 33 18 24 42 4.11 25.39 286.2318357 26 293