Enabling Data-driven Demand Forecasting Across Fragmented and Disconnected Markets

Background:

Some months ago, a European leader in roofing and waterproofing solutions received an order for an e-commerce warehouse. Such a warehouse typically runs into several square kilometers of roofing. Hence it would count as a prized order. This is more so if the solution provider has deep experience and has multiple plants to address demand spikes at short notice.

Here, the solution provider had a 150-year-old history with over 125 manufacturing facilities in 40 countries producing 27,000 products. And yet, despite these advantages, meeting the order proved to be an inordinately challenging task. This is because of the following issues:

• Fragmented Supply Chain: The provider had grown through several rapid acquisitions—with each acquired company operating as an independent entity having its own business culture, processes and nomenclature for products
• Internal resistance: For the business, it has been very difficult to communicate and coordinate production across these entities to meet any ambitious business order
• Lack of alignment: This would impact at site level or company level and lead to low performance levels

The company quickly recognized that the problem needed to be addressed. Today, using technology, it has harmonized processes and functions as an integrated entity. Making accurate demand forecasts and creating precise production plans have come within its grasp.

This European roofing and waterproofing specialist, that is a client of ITC Infotech, is not a unique example. There are scores of organizations whose awkward business structures hinder operational efficiency and business delivery. Examining the underlying challenge faced by this organization, and the systematic approach used to create a solution, offers insights that can be useful to any manufacturing entity.

Challenges faced:

We will begin with the barriers the client faced in exploiting the full potential of its manufacturing capability. Many of these are familiar to manufacturing businesses:

1. Unstructured Demand Planning: Aside from being independent run, each of the 125+ plants had its own method of forecasting demand. At least one of the 125+ plants used only an Excel worksheet (manual). Some even less (gut feel if you were wondering)! Usually the planning tools were rudimentary, and all were offline making collaboration impossible.
2. Non-standardized Hierarchy: Although the plants sold the same products in different markets, each plant used a different nomenclature for SKUs and could not clearly or reliably communicate requirements across plants.
3. Sales driven forecasting: Even with a vast catalogue of 27,000 products, the business could not take last-minute orders as demand forecasting and planning was sales driven. The fragmented nature of the business meant that it was perpetually in firefighting mode.
4. Speed to market based on just-in-time manufacturing: There was uncertainty around the location and timeliness of supply of raw materials, parts and products.
5. Lack of Integrated Supply Chain: The plants, spread across geographies, did not have a unified collaboration platform – this resulted in the business being unable to leverage the real production capabilities it possessed.
6. Inadequate investment in training and resources: Most employees in the planning function were production planners, not demand planners. Their understanding of demand planning needed enhancement using data and analytical tools. Also, the client could not dedicate many resources to demand forecasting.
7. Tight Planning cycle times: The client did not have the luxury of using 10 to 15 days, that is normal for most other businesses, to come up with demand forecasts. They needed forecasts created in 2 days.
8. Inaccurate forecast changes: Planners were unable to figure out whether sales director inputs were introducing any bias in the system and reducing the overall accuracy.
9. Inaccurate monitoring of all SKUs and stock levels: Inventory forecasting and demand planning becomes much more difficult when stock levels are hard to monitor or lack visibility. Without easy access to stock level reports a business can quickly build up unnecessary excess stock levels or they can have stock outages as demand spikes for specific products.

Implications of Inaccurate Forecasting:

Not solving the demand forecasting problem had serious consequences which were as follows:

• High Inventory: Due to inaccurate sales predictions many plants ended up with higher production than could be absorbed by their markets.
• High Cost: Finished goods had to be kept in warehouses, adding to cost overheads.
• High Risk of Scrapping: Roofing material is prone to deterioration when in storage, resulting in waste and loss.
• High Inventory Hold: Finally, every business wants to avoid keeping capital locked in inventory.
• Increased Expediting Costs: Also underestimating demand can cause inflated expediting costs to secure the rapid supply of raw materials.

Solutioning by ITC Infotech:

When ITC Infotech was brought in to provide the tools and systems to overcome the challenges, a global demand planning template was created on the Anaplan cloud planning platform (you guessed it, cloud makes real-time collaboration effortless).

The solution was designed so that the client could do the following:

• Analyze historical sales
• Perform statistical forecasting
• Make necessary adjustments to the generated forecast
• Compare budget vs forecast and
• Measure the forecast accuracy at various levels

To meet the diverse requirements of the client, for every role and responsibility, the Anaplan solution had to be enhanced and re-engineered. Our Anaplan experts had to decide which components to use, which to modify and where to apply automation.

What emerged was a solution designed for users who had no prior expertise in data science that demand forecasting traditionally calls for. To assist the team, algorithms had to be created that used the best parameters to automatically generate forecasts. Now, when team members log into their systems, they have immediate access to accurate, data-driven demand forecasts they can use with confidence.

The machine generated forecasts can be adjusted in real time using human judgment and intuition. Forecasts can be evaluated for accuracy and are visible across the entities of the business. Orders, now based on an 18-month rolling forecast, can be farmed out accurately while an additional layer of analytics enables leadership to monitor country-wise performance.

If a forecast proves to be flawed, planners can dig into the root cause of the failure and identify if the wrong algorithm was used, if the sales inputs were poor or the marketing intelligence was imperfect.

Benefits:

Currently the solution has been rolled out in 7 of the 40 countries the client operates in. The result of using technology is evident due to following benefits:

• Forecast accuracy has improved by 5% overall.
• Planning cycle times have reduced from 12 days to 2 days.
• Lost sales have been reduced by 10 % overall.
• The client now has access to granular and accurate forecasts and there is substantial improvement in visibility into the demand.
• It can now collaborate in real time between sales and supply chain teams using the cloud platform.

The single takeaway, for ITC Infotech, from the engagement has been that changing planning cultures is difficult but using technology to design solutions that account for user maturity levels makes it easier. The solution needs a user-centric approach to succeed.

Author:

Pavan KM
Lead Consultant, DATA
ITC Infotech

Rajeev Charaliyil
Sr Project Manager, DATA
ITC Infotech

Gayithri Krishnamurthy Joise
Sr Project Manager, DATA
ITC Infotech