Study Guide: Supply Chain Management - Sales Forecasting and Inventory Management

SUPPLY CHAIN MANAGEMENT: SALES FORECASTING AND INVENTORY MANAGEMENT

What is Sales Forecasting?

• Definition: Sales forecasting is the process of gathering and analyzing information to estimate future sales.

• Purpose: A sales forecast serves as an effort to estimate future sales to customers, acting as a key input to supply chain management. It is the initial step in planning supply chain operations.

• Role of Supply Chain Managers: Supply chain managers are key participants and users of sales forecasts because they develop plans to allocate enough assets to meet anticipated sales.

• Types of Assets:

  • Labor

  • Trucks

  • Warehouse space

  • Production capacity

Reasons to Forecast Sales

• Proper forecasting determines the necessary level of production to meet sales demands.

• Consequences of Miscalculating Forecasts:

  • Too much production leads to excess inventory.

  • Too little production results in stockouts and lost sales.

  • Both scenarios incur higher costs for the firm.

• Sharing sales forecast data is beneficial since one firm in a supply chain depends on the sales forecast provided by another firm.

Two Basic Forecasting Methods

• Qualitative Forecasting: Based on the judgment of knowledgeable forecasters. Better used when historical sales data is scarce or deemed unreliable.

• Quantitative Forecasting: Based on projections of historical sales data; most effective when past sales are a valid indicator of future performance.

Qualitative Forecasting Methods

Consensus Panel

• Formed by a group of experts who collaboratively determine a sales forecast.

• Weakness: Consensus can be influenced by experts with dominant personalities, instead of the best estimates.

• An alternative is averaging individual estimates to minimize bias.

Delphi Method

• Variation of the consensus panel where experts work independently and anonymously to avoid bias from dominant personalities.

• Process:

  • Facilitator prepares a questionnaire for experts who answer independently.

  • Facilitator compiles responses into a summary report.

  • Experts review the summary and revise their answers.

  • The process continues until a consensus is achieved.

• Effective across a variety of topics.

Sales Force Estimate

• Forecasts are derived from the judgements of salespersons who have direct contact with customers.

• Advantages: Sales force is best positioned to predict customer needs and market trends.

• Bias Risks: Salespersons may under-forecast or over-forecast based on their quotas or perceived needs for support of their products.

• Suggested approach: Evaluate and compensate sales teams not just based on meeting quotas but also for accuracy in sales forecasts.

Quantitative Forecasting Methods

Moving Average (MA)

• Definition: A simplistic forecasting method using the average sales from a predetermined number of past periods.

• Example: Predicting October sales with a MA of three months involves averaging sales from July, August, and September.

• Computation:

  • If sales were 130, 170, and 120, the October forecast is:
    $ext{Forecast} = rac{130 + 170 + 120}{3} = 140$

• Key Decision: Selecting the optimal number of periods for accuracy; larger periods dilute responsiveness to recent sales changes.

• Shortcoming: It does not account for trends or seasonality.

Exponential Smoothing (ES)

• Overview: A popular forecasting method utilizing a smoothing parameter to weigh past sales data, allowing it to consider a longer history of data than MA.

• Limitations: Not ideal for long-term forecasts; best suited for short-term (1-3 periods).

• Forecast Calculation: Based on last period's actual and forecasted sales with an error correction: $F<em>{t+1} = ext{α}A</em>t + (1 - ext{α})F_t$

  • Where F represents forecast, A actual sales, and α the smoothing parameter.

• Smoothing Parameter: Value between 0 (no impact of past errors) and 1 (consider only previous period's actual sales).

Trend Correction in Exponential Smoothing

• Second-order smoothing adjusts forecasts when sales trends are evident.

• Trend Value: Calculated by the change between forecast values over time.

• Adjusted Forecast: Adds/subtracts trend value to the forecast.

Seasonality in Exponential Smoothing

• Adjusts forecasts with a seasonality index to better compare sales across different seasons.

• Example: If high season sales are 20% above average, the Seasonality Index is 1.2.

• Adjusted sales are obtained by dividing high season sales by the index:
  $ext{Adjusted Sales} = rac{ ext{High Season Sales}}{ ext{Seasonality Index}}$

Integrating Qualitative and Quantitative Forecasts: S&OP

• Definition: Sales & Operations Planning is an integrative process for aligning multiple functions within a firm to enhance forecast accuracy and resource allocation.

• Components: People, Process, Technology.

• Collaboration: Enables sharing forecasts with suppliers and customers, leading to a united demand plan.

S&OP Process Steps

1. Gather sales and marketing data for baseline demand forecast (6-18 month horizon).

2. Assess current inventory strategies and initial supply plan according to marketing consensus.

3. Conduct cross-functional team meetings involving key stakeholders to finalize operating plans.

4. Distribute and implement the finalized S&OP plan through operations and sales teams.

5. Measure results and assess the effectiveness of the S&OP process.

Integrating Qualitative and Quantitative Forecasts: CPFR

• Definition: Collaborative Planning, Forecasting, and Replenishment involves synchronized efforts and information sharing among supply chain partners to enhance supply chain performance through joint decision-making.

• Comparison with S&OP: While S&OP focuses on internal alignment, CPFR extends this collaboration across multiple firms.

• Key Requirement: Establishing trust among partners due to extensive data sharing for effective implementation.

Standard CPFR Model Activities

1. Strategy and planning

2. Demand and supply management

3. Execution

4. Analysis

Measuring Forecasting Error

• All forecasts carry some degree of error; having a forecast is more advantageous than no forecast at all.

• Purposes of Measuring Forecast Error:

  1. Assess confidence in forecasts; less error results in higher confidence.

  2. Identify areas for improvement in forecasts.

  3. Focus management efforts on items with the largest forecast errors.

Factors Affecting Forecast Error

• Time Horizon: Longer forecasts tend to incur larger errors.

• Level of Aggregation: Errors are larger when forecasting at detailed levels versus aggregate levels.

Measures of Forecast Error

1. Mean Absolute Deviation (MAD): Average absolute errors across time periods; avoids sign issues.

2. Mean Absolute Percent Error (MAPE): A percentage measure allowing product comparison.

3. Mean Squared Error (MSE): Prioritizes many small errors over few large ones by squaring individual errors.

Measuring Forecasting Error: Detailed Descriptions

Mean Absolute Deviation (MAD)

• Definition: The absolute average error, calculated using:
  $ext{MAD} = rac{1}{n} ext{Σ}|A<em>i - F</em>i|$ , where Ai = actual value, Fi = forecasted value.

Mean Absolute Percent Error (MAPE)

• Definition: Gives percentage error allowing for comparison across products:
  $ext{MAPE} = rac{100}{n} ext{Σ} rac{|A<em>i - F</em>i|}{A_i}$

Mean Squared Error (MSE)

• Definition: Focused on squaring errors to attribute more weight to larger deviations:
  $ext{MSE} = rac{1}{n} ext{Σ}(A<em>i - F</em>i)^2$