Having worked in the Demand Planning arena with FMCG companies for the last 3-4 years, I have come to realize that one size fits all does not work for defining the demand planning solution for an FMCG organization.
In a global CPG company, different countries have altogether different ways of working, and bringing them on a common template solution is a great challenge.
I have tried capturing some of the key challenges one faces while establishing the demand planning solution for customers:
1. Choosing Route to Market (Distribution Channel).
In developing markets, Traditional Trade(TT) has been the conventional route to market for sales. But over the last decade, Modern Trade(MT) has gained a substantial portion of the business.
In developed markets, on the other hand, traditional trade has subsided long back and MT is the only option for getting your products to end consumers. Big retailers like Metro, Walmart, Costco, Tesco, and Target are buying directly from Companies and selling to the customers.
If a major share of the business is happening via Traditional Trade, forecast accuracy will certainly be better if sell-out history (Sales happening from distributor to end retailers) is leveraged for conducting Demand Planning. But we may run into a few roadblocks if forecasting is conducted using sell-out history.
If these difficulties can be worked upon, the statistical forecast generated on secondary sales will be far more accurate than what we will get by using Sell in History for traditional trade RTM.
2. Getting Inputs for History Cleaning from Cross-Functional Teams.
History cleaning is required to remove one-time events that change the nature of sales. E.g. Demonetization in 2016 had a significant impact on sales in the Indian market but it would not have been a recurring event year on year. Similarly, the Coronavirus pandemic in 2020 would cause a major dip in demand this year for sure.
Also, other significant events like cannibalization of sales due to promotions, trade promotions not going as planned, etc. must be cleaned so that future forecast does not take those surges and dips into account while generating a forecast
It is a crucial activity for getting a good statistical forecast and must be done with utmost care to get an accurate demand forecast as history is what forms the basis of generating statistical forecasts in the system for any retail organization.
Although the demand planner holds the responsibility for conducting history cleaning, still all the inputs for carrying out cleaning must be taken from various cross-functional teams like Sales, Markets, CP&A, and Finance.
Getting these inputs timely from them is always a challenge and if these inputs are required for a certain period in history, it makes it even more difficult if organizations don’t consider keeping the actuals.
Furthermore, if sell-out forecasting is being carried out, Inputs will be required from the distributor end which makes it even more difficult.
All these inputs are necessary and crucial to increase the accuracy of the demand forecast. Hence adequate processes and measures should be established to capture these events as and when they occur so that they are readily available when required for cleaning of sales history.
3. Alignment of System Generated Baseline Forecast with Sales Teams
When it comes to generating the baseline forecast or the bottom-up forecast number in demand planning, sales teams’ input is pivotal as they are the ones closer to markets and driving the generation of revenue for the organization.
So a major challenge comes in the alignment of statistical forecast numbers with sales and marketing teams because country demand planners have generated their number using statistical forecasting techniques but sales teams have a different number for baseline which has been derived using growth targets and swings of the portfolio in the recent quarter.
4. Deciding on levels in product hierarchy for generating baseline forecast
Forecast accuracy will always be higher if statistical forecasts and baselines are generated at aggregated levels of Product and Customer Hierarchy.
But Some Markets have hard constraints in terms of discussing their numbers with cross-functional teams and running promotions at certain lower levels of Product groups. Hence, it is not feasible for them to generate a baseline forecast at some aggregated level in the hierarchy.
In one FMCG client, one market chose to generate statistical forecast at the Product Group level (where they combined different Flavors within the same pack size) as they were running the same promotions across flavors and it was favorable for them to generate baselines at that level. Yet, another market in the same geographical location chose to generate baselines at a level below (Different flavors forecasted separately ) as there was a lot of variance in sales patterns as well as promotions across flavors.
5. Alignment of customer hierarchies for conducting the Demand IBP Process
An even bigger challenge is to set up a customer hierarchy accurately for an efficient and accurate IBP process for Demand.
Forecast accuracy would certainly be higher if Forecast is conducted at the country level but that is not always feasible. Discussions with Cross-Functional Teams such as sales, marketing, and CP&A may happen at RTM levels such as sell-in and sell out or it can happen based on Geographical Setup (North/South or East/West), etc.
For one of our CPG clients, a huge market had a central demand planner for each category who managed country Level numbers and then allocated numbers to field demand planners who were managing RTMs and Regions.
Consensus meetings with cross-functional teams also happened at Country Level itself and hence the market chose to do baseline generation at the country Level though they had Modern Trade, Traditional Trade, E-Commerce, and ‘Away from home’ as different routes to market.
Another country, during blueprinting, chose to conduct forecasting at the channel level segregating its modern and traditional trade business by channels. But they did not see a tangible benefit in investing more time managing forecasts for 2 channels separately because they were discussing numbers with cross-functional teams at the country level.
So, at a later stage, they wanted to change the setup to execute Forecasting at the National Level. this resulted in a lot of rework and delays in project timelines.
Some developing markets were heavily dependent on traditional trade, but modern trade was also picking up which had altogether different ways of working for the 2 channels. Hence their baseline generations and forecast discussions were built around 2 channels.
Secondly, Sales teams in markets might be already using some sales promotions planning tool which will generate results at a certain level of the hierarchy. Synchronization of that hierarchy with the one which is being established in the demand planning solution is really critical because planners will need to upload promotions in the ERP system.
6. Accuracy of aggregated level forecast and disaggregation
As mentioned earlier, the accuracy of statistical forecast certainly increases when it is generated at aggregated levels of product and customer hierarchy.
But then at the same time, challenges arise in terms of disaggregation of that forecast to lower levels of hierarchies.
As a global process, most FMCG companies use the share of the business of the recent past to disaggregate volumes but that may not go well with Sales Teams which may want to uplift or reduce forecasts based on market inputs and revenue targets instead of a share of the business.
e.g. One heavily promoted market in one FMCG client was not ready to use the share of business for disaggregation of customer volumes as they were a heavily promoted market with promotions varying to a great extent by customers by each month. And we had to implement the concept of a Planner Proposal for Disaggregation in customer hierarchy in which users could provide ratios of their choice if they wanted disaggregation proportions different from the existing share of the business.
Another point to note was that the use of Share of business in the recent past for disaggregating forecast volumes may have its challenges while doing alignment with cross-functional teams and have an impact on forecast accuracy. This also was a great lesson learned for future implementations where a decision must be made on disaggregation.
To summarize, Demand Planning, in sharp contrast to Supply Planning, is a people's game much more than it is system controlled. There are a lot of stakeholders involved and alignment from all of them is needed to arrive at a consensus forecast. And, one needs to deep-dive into the organization's pain points during the initiation phase itself if a good and sustainable solution has to be implemented.
In a global CPG company, different countries have altogether different ways of working, and bringing them on a common template solution is a great challenge.
I have tried capturing some of the key challenges one faces while establishing the demand planning solution for customers:
1. Choosing Route to Market (Distribution Channel).
In developing markets, Traditional Trade(TT) has been the conventional route to market for sales. But over the last decade, Modern Trade(MT) has gained a substantial portion of the business.
In developed markets, on the other hand, traditional trade has subsided long back and MT is the only option for getting your products to end consumers. Big retailers like Metro, Walmart, Costco, Tesco, and Target are buying directly from Companies and selling to the customers.
If a major share of the business is happening via Traditional Trade, forecast accuracy will certainly be better if sell-out history (Sales happening from distributor to end retailers) is leveraged for conducting Demand Planning. But we may run into a few roadblocks if forecasting is conducted using sell-out history.
- Getting accurate sell-out history from the distributor for a certain period would certainly be a challenge. Organizations do need an agreement signed with distributors to provide them with correct sell-out history periodically and a traditional distributor might not be technologically equipped to do that easily or maybe not be interested in sharing these details
- The distributor’s data will be in a different format and a large FMCG organization will be working with a lot of distributors. In addition to that, the distributor can create his material codes to run promotions at this end to retailers and there will be no corresponding material existing in the company's ERP system. Getting all this worked up and converting this data to FMCG’s system’s language will pose a major challenge.
If these difficulties can be worked upon, the statistical forecast generated on secondary sales will be far more accurate than what we will get by using Sell in History for traditional trade RTM.
2. Getting Inputs for History Cleaning from Cross-Functional Teams.
History cleaning is required to remove one-time events that change the nature of sales. E.g. Demonetization in 2016 had a significant impact on sales in the Indian market but it would not have been a recurring event year on year. Similarly, the Coronavirus pandemic in 2020 would cause a major dip in demand this year for sure.
Also, other significant events like cannibalization of sales due to promotions, trade promotions not going as planned, etc. must be cleaned so that future forecast does not take those surges and dips into account while generating a forecast
It is a crucial activity for getting a good statistical forecast and must be done with utmost care to get an accurate demand forecast as history is what forms the basis of generating statistical forecasts in the system for any retail organization.
Although the demand planner holds the responsibility for conducting history cleaning, still all the inputs for carrying out cleaning must be taken from various cross-functional teams like Sales, Markets, CP&A, and Finance.
Getting these inputs timely from them is always a challenge and if these inputs are required for a certain period in history, it makes it even more difficult if organizations don’t consider keeping the actuals.
Furthermore, if sell-out forecasting is being carried out, Inputs will be required from the distributor end which makes it even more difficult.
All these inputs are necessary and crucial to increase the accuracy of the demand forecast. Hence adequate processes and measures should be established to capture these events as and when they occur so that they are readily available when required for cleaning of sales history.
3. Alignment of System Generated Baseline Forecast with Sales Teams
When it comes to generating the baseline forecast or the bottom-up forecast number in demand planning, sales teams’ input is pivotal as they are the ones closer to markets and driving the generation of revenue for the organization.
So a major challenge comes in the alignment of statistical forecast numbers with sales and marketing teams because country demand planners have generated their number using statistical forecasting techniques but sales teams have a different number for baseline which has been derived using growth targets and swings of the portfolio in the recent quarter.
4. Deciding on levels in product hierarchy for generating baseline forecast
Forecast accuracy will always be higher if statistical forecasts and baselines are generated at aggregated levels of Product and Customer Hierarchy.
But Some Markets have hard constraints in terms of discussing their numbers with cross-functional teams and running promotions at certain lower levels of Product groups. Hence, it is not feasible for them to generate a baseline forecast at some aggregated level in the hierarchy.
In one FMCG client, one market chose to generate statistical forecast at the Product Group level (where they combined different Flavors within the same pack size) as they were running the same promotions across flavors and it was favorable for them to generate baselines at that level. Yet, another market in the same geographical location chose to generate baselines at a level below (Different flavors forecasted separately ) as there was a lot of variance in sales patterns as well as promotions across flavors.
5. Alignment of customer hierarchies for conducting the Demand IBP Process
An even bigger challenge is to set up a customer hierarchy accurately for an efficient and accurate IBP process for Demand.
Forecast accuracy would certainly be higher if Forecast is conducted at the country level but that is not always feasible. Discussions with Cross-Functional Teams such as sales, marketing, and CP&A may happen at RTM levels such as sell-in and sell out or it can happen based on Geographical Setup (North/South or East/West), etc.
For one of our CPG clients, a huge market had a central demand planner for each category who managed country Level numbers and then allocated numbers to field demand planners who were managing RTMs and Regions.
Consensus meetings with cross-functional teams also happened at Country Level itself and hence the market chose to do baseline generation at the country Level though they had Modern Trade, Traditional Trade, E-Commerce, and ‘Away from home’ as different routes to market.
Another country, during blueprinting, chose to conduct forecasting at the channel level segregating its modern and traditional trade business by channels. But they did not see a tangible benefit in investing more time managing forecasts for 2 channels separately because they were discussing numbers with cross-functional teams at the country level.
So, at a later stage, they wanted to change the setup to execute Forecasting at the National Level. this resulted in a lot of rework and delays in project timelines.
Some developing markets were heavily dependent on traditional trade, but modern trade was also picking up which had altogether different ways of working for the 2 channels. Hence their baseline generations and forecast discussions were built around 2 channels.
Secondly, Sales teams in markets might be already using some sales promotions planning tool which will generate results at a certain level of the hierarchy. Synchronization of that hierarchy with the one which is being established in the demand planning solution is really critical because planners will need to upload promotions in the ERP system.
6. Accuracy of aggregated level forecast and disaggregation
As mentioned earlier, the accuracy of statistical forecast certainly increases when it is generated at aggregated levels of product and customer hierarchy.
But then at the same time, challenges arise in terms of disaggregation of that forecast to lower levels of hierarchies.
As a global process, most FMCG companies use the share of the business of the recent past to disaggregate volumes but that may not go well with Sales Teams which may want to uplift or reduce forecasts based on market inputs and revenue targets instead of a share of the business.
e.g. One heavily promoted market in one FMCG client was not ready to use the share of business for disaggregation of customer volumes as they were a heavily promoted market with promotions varying to a great extent by customers by each month. And we had to implement the concept of a Planner Proposal for Disaggregation in customer hierarchy in which users could provide ratios of their choice if they wanted disaggregation proportions different from the existing share of the business.
Another point to note was that the use of Share of business in the recent past for disaggregating forecast volumes may have its challenges while doing alignment with cross-functional teams and have an impact on forecast accuracy. This also was a great lesson learned for future implementations where a decision must be made on disaggregation.
To summarize, Demand Planning, in sharp contrast to Supply Planning, is a people's game much more than it is system controlled. There are a lot of stakeholders involved and alignment from all of them is needed to arrive at a consensus forecast. And, one needs to deep-dive into the organization's pain points during the initiation phase itself if a good and sustainable solution has to be implemented.
- SAP Managed Tags:
| User | Count |
|---|---|
| 16 | |
| 4 | |
| 4 | |
| 4 | |
| 2 | |
| 2 | |
| 2 | |
| 1 | |
| 1 | |
| 1 |
