Perfecting the Balancing Act of Inventory Management

Out-of-balance inventory that might have amounted to a minor stumble back in the 20th century could now be catastrophic, if not fatal.

It’s time to explore moving from people doing the work with the help of machines, to machines doing the work under the guidance of people.
It’s time to explore moving from people doing the work with the help of machines, to machines doing the work under the guidance of people.
Aera Technology

Inventory management is fundamentally a balancing act. Organizations are ever balancing achieving profitable service levels against the cost of holding inventory, while facing the headwinds of variable demand. Trying to optimize one traditionally meant negatively impacting another.

That’s been the case for decades, but much has changed. Years ago, supply chain dynamics were far simpler. If you lost your balance and stocked too much or too little inventory, markets were much more forgiving. A fall was rarely fatal.

Today, inventory management is like walking a high-wire tightrope in a hurricane. Supply chain professionals are buffeted by gale-force winds of global complexity, rising customer expectations, vast volumes of fast-changing data and fierce competition from digital native companies. And, now the Coronavirus disease (COVID-19) crisis has only raised the stakes.

Out-of-balance inventory that might have amounted to a minor stumble back in the 20th century could now be catastrophic, if not fatal. And, we’re already seeing the fallout in consumers hoarding in the face of perceived scarcity and companies in financial peril from uncertain projected recovery.

Outdated tools

A fundamental problem underscores the inventory dilemma. Supply chain professionals are trying to solve modern inventory challenges with yesteryear’s tools, largely based on analytic models that aren’t well suited to the complex multi-echelon supply chains of today.

An analytic model is basically a mathematical equation that helps managers generate rules based on parameters (demand, lead time, output, various costs, etc.) in efforts to optimize inventory. While useful, analytic models require simplified assumptions that seldom mirror the real world and don’t account for critical complexities. 

Analytic models are mostly on historical data and are geared to solve known problems – not adapt to unknown problems and inevitable disruptions as they occur. The result can be inaccurate models that introduce additional problems and require workarounds.

While analytic models are most prevalent, simulation and heuristic-intuitive models offer alternatives that are more flexible and better suited to fast-moving multi-echelon supply chains.

Tools based on simulation models give managers the ability to input a range of potential constraints (as well as historical data) to forecast options for inventory management. The heuristic technique is similar, relying on human experience and insight to devise models. Human judgment is emphasized over the mathematical approach of analytic models, but of course human judgment is not flawless.

Breaking the bonds limiting inventory modeling

Despite their utility, models based on analytics, simulations and heuristics have suffered from limitations that weaken their effectiveness.

Unlocking data. The data available to inventory management tools has been typically incomplete, outdated and of questionable quality. Supply chain teams were not set up to pull information from the dozens or hundreds of software applications that run a global supply chain — systems for enterprise resource planning (ERP), procurement, production, warehouse management, materials requirement planning (MRP), logistics and more.

But, a recent revolution from cognitive automation has pushed intelligence into the data layer to self-discover and ingest needed data. Supply chain teams and their IT counterparts are relieved of the burden of months-long data integration projects. And, this intelligence translates into pre-aggregated and cleansed data that becomes available to a new cognitive processing layer.

Making simulation smart and adaptive. None of the existing tools or models can account for the full range of potential business scenarios. When the unexpected happens — demand spikes, a port shuts down, a quality control issue arises with a key supplier — supply chain professionals are forced to scramble to devise a solution.

But, simulation is now being combined with decision digitization, which means simulations can be incorporated in more variations, with more context and at a faster clock speed. A cognitive layer can simultaneously evaluate scenarios against business goals and quickly come back with optimized solutions, essentially creating continuous inventory balancing.

Unleashing people potential. The time and cost of resources needed to utilize legacy tools has been high. Worse, it’s difficult to find and retain qualified professionals eager to take on labor-intensive manual work that it forces on them. And, if an expert who devised a certain mission-critical model leaves the company, that knowledge disappears as well. Others are left to figure out the data mechanics that drive the model.

Cognitive automation has introduced a very different dynamic for leading organizations. With thousands of daily decisions being evaluated by a cognitive layer, people suddenly have the time to do what humans do well- finding new and innovative ways to satisfy the customer. People can focus on the strategic management of extremely complex and ever-changing data and supply chain ecosystems, guiding machines to move it toward new outcomes.

Cognitive automation for inventory optimization

Cognitive automation changes the game. The technology brings together Internet-scale compute power, real-time data aggregation and artificial intelligence and machine learning (AI/ML) to optimize inventory at speed and scale not humanly possible. Here’s how it works:

Cognitive automation starts with thousands of daily Google-like data crawls across enterprise applications to create a single top-level layer of virtualized data. There, the data is indexed, cleansed, normalized and enriched. For the first time, the enterprise has the real-time end-to-end visibility for inventory optimization.

Then, sophisticated AI/ML algorithms analyze the data for anomalies, risks and opportunities. AI will offer recommendations on optimal actions that teams can follow to optimize inventory, or it can be authorized to take action autonomously. It’s able to learn over time and assess the quality of inventory decisions made, driving continuous improvement.

Cognitive automation also gives teams the flexibility to utilize more effective simulation and heuristic models to identify and address disruptions quickly and decisively. It helps organizations tackle questions such as:

●       How can we respond to a stock shortage or demand spike in a given region?

●        How can we reduce working capital in excess stock?

●        How can we address chronic overstocks at one warehouse?

●        What’s our optimal minimum and maximum safety buffers?

●        How do we best manage our expiring inventory?

Cognitive automation is in use today at leading global enterprises to solve unexpected challenges faster, forecast more accurately, generate millions of dollars in bottom-line payback and better manage inventory and other aspects of the end-to-end supply chain.