How to design a Demand Planning process

Once the rationale for demand planning is defined, the organization needs to determine what to forecast. This requires an examination of the product or products required by and provided to the end-customer. This examination will include an analysis of the product portfolio and hierarchy.  In essence, this examination can be and would have already been captured by a working Marketing Plan or Sales Plan which addresses the following basic questions;

• What products are you offering to the market?
• Who are the target customers for the product offerings?
• What product families are these products grouped?
• In what territories will the products be offered?
• What is the build or purchase strategy for these products e.g. made-to-order, made-to-stock?
• At what pricing or pricing structures will the products be offered?
• At which lifecycle stage are the products in e.g. new, mature, obsolete?
• What ABC classification can be applied to each product e.g. high-value, high-use, medium-value, medium-use, low-value, low-use or combinations thereof?

Where to forecast?

After determining the products to be forecasted, the next step is to answer questions related to the logistics and distribution of the products to be offered to the customer. The questions below will be helpful.

• What is the delivery strategy for each product e.g. pick-up, door-to-door delivery, etc.?
• In what warehouses or distribution centers will the products be stored and distributed from?
• What ABC classification can be applied to each product e.g. high-value, high-use, medium-value, medium-use, low-value, low-use or combinations thereof in the specific warehouses or distribution centers?

From the above, the organization will be able to determine what products to forecast at certain distribution centres or warehouses. Normally, this would mean a specific stock keeping unit (SKU) and warehouse or what is called a SKU-warehouse combination.  However, there may be other combinations that some companies may prefer like SKU-region or SKU-customer combinations appropriate to the business environment the company is involved in.  Eventually, the organization also has to agree on the aggregation of products to consider for reporting requirements.

When to forecast?

The next step is to determine the time buckets with which to forecast products e.g. daily, weekly, or monthly. This decision will be dependent on the volatility of the market or the business the organization is engaged in. The organization then has to determine the time horizon to forecast. This can be the next 6 months, 1 year, 1 and a half years, etc. Finally, the organization has to agree on the time intervals of forecasting e.g. weekly forecasting, monthly forecasting, etc.

For organizations that utilise a Materials Requirements Planning (MRP) system, there are further considerations to be addressed in regards to time-based forecasting. These considerations may include how the forecast is split and how the forecast is consumed.

In forecast splitting, assuming the forecast is loaded in monthly time buckets, the MRP system will then split the forecast from monthly to weekly time buckets for detailed planning. For example, a monthly forecast of 100 units may be split into a weekly forecast of 25 units. This provides a basis with which the MRP will run and from which actions like builds, purchases and deliveries can be made and timed.

On the other hand, forecast consumption patterns provide direction to an MRP system on how to automatically consume or delete a forecast when a customer order, sales order, or demand is entered into the system. The forecast consumption capability ensures that the forecast is replaced and does not remain in the system and double-up and drive demand for products or parts.

How to forecast?

In the scope of demand planning, this activity aims to determine the best way to predict a more realistic level of future demand for products and/or services.  Armed with the knowledge of what to forecast, where to forecast and when – the following steps may help organizations on how best to forecast.

1. Organization – Organise a team responsible for determining, planning, challenging, and approving forecasts. Considerations may include:

• Demand Planning team – responsible for creating baseline forecasts, maintaining data and forecast integrity in the sales forecasting tool and/or ERP/MRP system as well as other related platforms as applicable, and communicating the forecast to other downstream users
• Marketing and Sales team – responsible for providing market information that impact sales and forecasts like promotional activities, pricing strategies, new product introductions, obsolescence, etc.
• Finance team – responsible for challenging sales forecasts in relation to overall company budgets and financial targets
• General Management – responsible for approving the forecast to ensure it is in line with strategic and financial goals of the company

2. Forecasting – An agreed process for cross-functional team determination of a single forecast

• Baseline forecasting – involves an analysis of historical sales, macro and micro economic factors, and the application of qualitative and quantitative methods to create the initial forecast – may take anywhere between 3 to 10 working days depending on the number and complexity of the products
• Collaborative forecasting – involves presentation of the baseline forecast and first-level discussions between the demand planning team, sales and marketing teams to determine the nature and extent of activities that impact the forecast e.g. promotional activities, product launches, etc. The baseline forecasts are thus adjusted to consider the inputs from the collaborative discussions.  Organizations may include other participants depending on the value of the inputs they can bring to the discussions.  Additional layers of collaborative discussions may also be considered especially depending on geographic divisions vis-à-vis forecast reporting requirements e.g. state or territory forecasts, national forecasts, regional forecasts, continental forecasts, etc.

3. Review – first-level review of the collaborative forecast by the functional managers of Demand Planning or Supply Chain teams, Sales and Marketing teams plus the Finance team. This is to ensure that the forecasts are challenged and in-line with operational, revenue and financial targets prior to approval by General Management.

4. Approval – high-level review and approval of overall forecasts of the company. General macro variables and long-term strategic inputs and plans may also be considered and discussed especially in relation to the forecast being approved.

5. Forecast Data Upload, Processing, and Storage – a forecast upload and processing schedule must always be maintained to ensure that ERP/MRP systems are updated accordingly with the new forecast. This is especially true when a forecasting tool/software exists (that stores the approved forecasts) that is separate or different from the ERP/MRP system to which the approved forecast has to be uploaded. A thorough review of the uploaded forecast must be conducted to ensure that it is accurate and is a 100% copy of the approved forecast.

Another item to consider is a system of storing forecast data that will allow for performance tracking over time and across different metrics and measurements.  In-house or cloud-based systems can be explored depending on the organization requirements.

6. Forecast Handover – the final and approved forecast should then be reported to all stakeholders as well as to downstream users e.g. manufacturing planners, schedulers and buyers to ensure that everyone is aware of the new forecast as well as the incremental changes from the previous forecast. Usually, a Forecast Handover Meeting is called to present and discuss the new forecast.

7. Forecast Reporting and Performance Measurement – a performance measurement system must be in place for the organization to determine forecast accuracy, forecast efficiency, and general improvement of the demand planning process. In accordance with the organization’s determination of “What to forecast?”, forecast reporting should also be aligned to conform to the level of aggregation required (e.g. by product family or sub-families, by territory, by distribution centre, etc.) that may also link to other KPI’s of the company like service-levels, inventory levels, etc. Reports can also be released to track the performance of different layers of forecasting (e.g. baseline forecast by the demand planner, first level collaboration, second level collaboration, addition of promotions or external events, etc.) in order to allow the organization to determine opportunities in the different layers of forecasting.

Several metrics may be used to track forecast accuracy.  Among the common ones are Mean Absolute Percentage Error (MAPE), Median Absolute Deviation (MAD), Mean Absolute

Scaled Error (MASE), etc. Tracking Signals and Forecast Bias are helpful tools in analysing historical forecast accuracy.  Some organizations may also find it necessary to measure the forecast efficiency of one statistical model over another  e.g. exponential versus average or between a baseline forecast and another layer of forecast collaboration i.e. baseline forecast versus first-level collaboration.  Common metrics used are Geometric Mean Relative Absolute Error (GMRAE), Forecasting Efficiency Quotient, etc.

Finally, the whole demand planning process may also be triggered to scale up to a higher level of performance by changing some key reporting parameters.  For example, if the current metric tracking (say MAPE) is calculated using Month-1 (actuals sales of May 2016 is compared to the forecast of May 2016 as determined back in April 2016) we can then improve by switching over to Month-2 (actual sales of May 2016 is compared to the forecast of May 2016 as determined back in March 2016) or even Month-3 (actual sales of May 2016 is compared to the forecast of May 2016 as determined back in February 2016).  This is particularly useful to align the forecast accuracy measurement with the average lead time of materials.  So if the average materials lead time is 3 months it is better to measure forecast accuracy using Month-3 instead of Month-1.  Tracking Signals and Forecast Bias may also apply a similar switch, i.e. from Month-1 to Month-3 and may include upper and lower control limits which may be tightened to show an increase or scale up in the performance monitoring and control process.

8. Resulting Benefit/Impact to other KPI’s and operational parameters – The improvements achieved in a working demand planning process will definitely need to be translated into other KPI’s and other important supply chain parameters. For example, if the forecast accuracy is improving, there may be a need to reduce or eliminate safety stocks that will help in reducing inventory levels and reduce working capital outflow.

An improvement in forecast accuracy will also reflect improvements in service levels and DIFOT targets.  Conversely, if target finished goods stock levels are attained there may be an opportunity to reduce finished goods lead times to customers – say if a stock is available to commit and deliver in 4 to 6 weeks then with the improvement of forecast accuracy and stock levels the commit and delivery lead time can then be reduced to a commit of 2 to 4 weeks – which will translate into a significant competitive advantage in the market.

In conclusion, organizations are encouraged to go through a review of the suggested steps in order to come up with a viable demand planning process.  The initial process is never the final and optimum one; however, it should be one that allows the organization to establish a baseline process from which to improve.  This can also pave the way for overall review and streamlining of the linkages of the demand planning process with the other supply chain functions and processes as well as other departmental functions and processes like manufacturing, sales, marketing, finance, etc.

Wilson Penafiel
A Supply Chain Management professional with more than 20 years’ experience in supply chain development and transformation, global customer support, operations management and business enterprise solutions gained from the utilities, electrical, electronics, pharmaceuticals, food, and consumer goods industries.
Possesses proven experience in managing end-to-end supply chain, including customer service, demand planning, supply planning, materials and inventory management, S&OP, purchasing, logistics and warehousing.
An Industrial Engineer and Six Sigma practitioner with proven track record in process improvement, quality and project management.