Closing the Loop. Now that we have taken action based on facts, analysis and modeling, let’s measure and monitor our results so that we can learn
I woke up early at my hotel in Beijing to prepare for a long day ahead: I was working in the retail industry.
I entered the meeting room right on time. Once I had met the group and they settled in their seats, I began, “If you are like most retailers, you may be wondering about four questions or in other words your key performance indicators.” I went on to ask the audience the four questions:
How do I increase my fill rates without increasing inventory costs?
Is it possible to decrease inventory levels by 10%, for example, without impacting service levels?
How can I leverage all the product and customer data in our customer relationship management, enterprise resource planning, and manufacturing resource planning systems for service?
How can I reduce cost of goods sold (bottom line) and improve service revenue (top line)?
I concluded: “A good answer to these questions lies within software for inventory optimization.”
The meeting was comprised of a small group of select executives. One of them raised his hand politely and said, “Is product inventory optimization software designed to address the need for inventory forecasting and product management needs for retailers and manufacturers?”
I answered, “Yes. Inventory forecasting software provides significant financial benefits to the organization. Good inventory forecasting software should provide the user with the ability to have data integration that consolidates different data sources and business intelligence, thereby providing basic reporting for product managers as well as a sophisticated forecasting tool that can handle both large data sets and sparsely populated data collections.”
The questioner nodded knowingly and I continued outlining how the process works. “Using this system, analysts can utilize the extensive forecasting capability to help them better understand and predict the trends and activity associated with their products for both sales and replenishment.
“With their findings, product managers can better manage and optimize their current and projected inventory, providing capital savings for the chief financial officer as well as increasing customer satisfaction.”
One executive asked, “Does a product inventory optimization solution provide organizations with the ability to calculate periodic review inventory replenishment policies for distribution and/or retailer systems, thus enabling product managers to maintain adequate stock levels and improve customer satisfaction?”
I answered quickly, “Yes. The solution aids decision making by helping organizations answer three fundamental questions of inventory management. These questions are:
What is the optimal inventory range by stock-keeping unit (SKU) to achieve a specific service level based on current inventory policies?
Which items have crossed policy thresholds and therefore should be reordered to restock inventory?
How much should be ordered?”
“A solution may offer two approaches to define monthly replenishment requirements:
“The first approach can be leveraged if there is a constraint on regional stock coverage. When this is true, the engine uses historical demand data, reorder lead time, beginning inventory at the distribution center, beginning inventory at central warehouse, optimal production quantity, required stock coverage in the region, and required stock coverage at the central warehouse to define monthly replenishment requirement by region by SKU.”
The man continued, “So, when there are no constraints on regional stock coverage, should the solution still use historical demand data? What about reorder lead time and the associated inventory cost data, which includes the cost of replenishment, the cost of holding inventory, and the optional cost of backordering (stock-outs) and target service levels?”
I said, “Yes, sir. For example, the sophisticated policy calculation algorithm in the software accounts for variability in demand data and supply, thus helping organizations compare and choose from various scenarios.”
I continued. “The policies produced by a product inventory optimization solution should perform better than the standard economic order quantity policies, which do not account for variation in customer demand and replenishment order lead times. This approach helps manufacturers and retailers develop better replenishment strategies for repeatedly ordered items. The second approach helps manufacturers and retailers develop better replenishment strategies for repeatedly ordered items. The solution uses historical demand data, replenishment lead time, inventory cost, and target service levels to quickly calculate accurate replenishment policies that help manufacturers and retailers determine when each item should be ordered and in what quantity. An underlying sophisticated policy calculation algorithm helps manufacturers and retailers calculate optimal policies such that they can achieve target customer service levels while minimizing ordering and inventory holding costs.”
The businessmen all looked at each other, nodding. I said, “An effective implementation of a product inventory optimization solution consists of not only the installation of the software components but also the consulting and training services that go with it.”
WHAT THE DIFFERENT SOFTWARE COMPONENTS SHOULD DO
Forecast Software Features
Forecast software should have an easy-to-use graphical user interface (GUI). It allows the choice of forecast automation level. Users can choose the level of automation for the forecasting process: rediagnose and identify candidate models, reestimate existing model parameters, or generate forecasts using existing models and parameters. No programming should be required. Users just point and click their way to powerful forecasting capabilities.
Product Inventory Optimization Software Engine Features
The engine should help the organization to develop better replenishment strategies for repeatedly ordered items. The solution should use historical demand data and replenishment lead time to quickly calculate accurate replenishment policies that help the product manager to determine when each item should be ordered and in what quantity. The underlying policy calculation algorithm should also be used to calculate optimal policies to achieve target stock coverage. These accurate inventory policies will help determine how much of each item should be ordered and when it should be ordered for each region by SKU.
Business Intelligence Software Features
Business intelligence (BI) software provides comprehensive and centralized data, self-service reporting, ad hoc query and analysis, slice and drill-down reporting, integrated analytics, and integration with Microsoft Office products. Most reporting products integrate with Microsoft Office. This integration enables the product manager to do self-reporting and be somewhat independent of IT.
Data Integration Software Features
Data integration software is a powerful, configurable, and comprehensive software that empowers the IT department to access virtually all data sources; extract, cleanse, transform, conform, aggregate, load, and manage product data; support data warehousing, migration, synchronization, and federation initiatives; support both batch-oriented and real-time master data management solutions; and create real-time data integration services in support of service-oriented architectures. There is no wait when you have a good data integration environment.1
Data Mining Software Features
Software for data mining supports the entire data mining process with a broad set of tools. It allows the customer statistical modeling group, business managers, and the IT department to interface with other groups and create accurate business-driven data mining models in a seamless process, enabling the entire team to collaborate more efficiently. It will help create better-performing models with new innovative algorithms that enhance the stability and accuracy of predictions, which can be verified easily by visual model assessment and validation metrics.
I clasped my hands together and summed it all up. “The software platform should be easy to use and include an intuitive user interface that incorporates common design principles established for analytics software.”
An executive asked, “Shouldn’t it also have additional navigation tools for moving easily around the workspace and display the key performance indicators (KPI) that we discussed?”
I smiled in agreement. “Absolutely! The GUI can be tailored for all analysts’ needs via flexible, interactive property sheets, code editors, and display settings.”
The man grinned at me, his eyes sparkling. “Does the software also ease the model deployment and scoring process?”
I said, “Yes. Scoring—the process of applying a model to new data—is the end result of many data mining endeavors. The software automates the tedious scoring process and supplies complete scoring code for all stages of model development in SAS, C, Java, and PMML. The scoring code can be deployed in a variety of real-time or batch environments or directly in relational databases. The outcome is faster implementation of data mining results.”
Another executive looked up from his laptop and started talking. “I understand that one desirable characteristic of data mining is that it provides scalable processing for large installations with millions of SKUs. It has an architecture that scales from single-user to large enterprise solutions.”
I love when we get into discussions like this. I added, “Yes! It provides server-based processing and storage. Data mining should be done by a software component that is fully prepared to perform with grid computing, parallel and in-database processing, and multithreaded predictive algorithms. This is now called high-performance analytics!”
Somebody asked: “What consulting and training services are needed?”
I replied: “The implementation services associated with a product inventory optimization initiative can be broken down into four areas of consulting services needed. These four areas include:
Installation and configuration;
Analytic and forecast consulting;
Product inventory optimization consulting; and
Training services.”
I admitted, “It may sound like a lot, but it is worth many dollars in savings, especially for a company in China!”
An executive said to me, “I believe the main objective of the consulting services is to provide the initial setup and direction to help the customer take ownership and continue the ongoing expansion of the system capabilities. Do you see it differently?”
I agreed, “You are right. This is because these systems are extremely interactive and are developed in an iterative way to adjust to extreme competitive situations.”
The executive continued, “For the consulting engagement, would you require a dedicated team to be active participants and to receive the appropriate knowledge transfer and system training to create adequate dashboards to keep track of KPIs around optimized levels of inventory?”
I smiled knowingly. “Yes. This approach includes working with internal resources to obtain access to the existing data warehouse environment and understand the data elements that are available for analysis and reporting.”
I thought the critical takeaway from these dashboards and the KPIs they support was the need for organizations to collaborate around a single system that maintains the corporate memory and the one single truth. These feedback systems are needed for effective organizational performance. They help monitor both the generation of revenue and the maintenance of high levels of customer satisfaction.
