16.3 One and Zeroes: Information Technology Advances

Just earlier, we talked about RFID and how it had the potential to do many wonderful things for how businesses can lower cost and improve performance. Sure, RFID’s alpha-numeric system offers infinitely more combinations than the numbers-only system of the UPC to accommodate an ever-growing assortment of products for sale. Its real value, however, lies in its ability to capture and store information automatically as it passes through RFID sensors. The information captured by these sensors is then stored in data warehouses. As Clive Humby, a British mathematician and co-founder of retail analytics firm Dunnhumby, said in 2006, “Data is the new oil.” As oil requires refining to be processed into usable products such as gasoline and plastic, data needs it too.

Analytics

So, you have captured broad swathes of data from different information technology systems. Whether they are from smaller, site-specific systems such as warehouse management systems (WMS) or larger, organization-wide systems such as enterprise resource planning (ERP) systems, all that data simply sit there—like oil deposits in the ground—waiting to be tapped for knowledge. The first tool in your data refining tool chest is analytics.

Data by itself is simply a byproduct of what is happening with your business. By itself, data does not tell you what you can or should do with your business. By applying different analytical techniques, whether it’s your basic Excel-based spreadsheet analysis or Big Data systems, you can transform data into different forms of business intelligence that will impact the way you implement lean programs. Broadly, there are three different types of analytical purposes.

1. Descriptive Analytics (DA). Like the name suggests, applying your analytical tools for descriptive purposes allows you to get an overview of how your current business process is moving. For instance, descriptive analytics allows a retailer to know sales trends for products, categories, geographic locations, over a specified time period (e.g., weekly, monthly).

2. Predictive Analytics (PA). Based on the information gleaned from DA, further application of your analytical tools would allow you to forecast your business’s future state. If a product sale is trending higher, will it continue to trend higher? Are there external factors causing this trend to go higher? PA differs from traditional demand forecasting in that the latter mostly uses sales history to forecast future demand, whereas PA applies other factors such as weather to enhance the long-term accuracy of demand forecast.

3. Prescriptive Analytics (SA). DA tells you what the past looks like. PA tells you what the future might look like. SA takes it a step further and tells you what you should do. Like the 2002 blockbuster science-fiction movie Minority Report, SA functions similarly to the crime-predicting system where police are told to arrest individuals who are predicted to commit a serious crime.

For lean management, you have learned before that process variations and defects can be attributed to either common or special factors. Of course, there is not much you can do about common variations. Where analytics can help is to increasingly identify causes or events leading up to special factors that will result in defects. For example, if climate events in China have the potential to adversely influence a supplier’s production, then an analytics-driven system would be able to identify it as a potential cause that could ultimately lead to delayed inbound shipment of components associated with that supplier.

A final warning: such a system may not be very accurate in the beginning because it might not have enough data points with which to generate a reliable prediction. That is why implementing such systems requires a transition period.

Process Digitization

If you take a look at the power meter, over the years they went from physical dials to LCD screens to display the number of kilowatt hours consumed by a household. Why the change? The new power meters are able to collect a lot more information on how you use your electricity, right down to an estimate on whether your household used electricity for lighting as opposed to heating and cooling! This is what process digitization can do for you: by converting manual, information-intensive processes into digitized, often automated processes, you can accelerate decision-making processes and free your resources to be deployed to other aspects of your company to generate greater value.

Remember earlier when you read that every organization is trying to collect more information on everything that their businesses touch? The primary purpose is to manage their business processes more intelligently! The amount of data collected through an ever-expanding array of sources not only help businesses to better understand their customers but also allow them to have a ready pool of data that would allow them to gain greater visibility to opportunities to get leaner! Of course, you can’t do it haphazardly. The same rule applies to prevent you from drowning in a sea of data. According to McKinsey, digitizing information-intensive processes can allow organizations to cut costs by up to 90% and accelerate processes by several folds.

However, technology implementation is an inherently social process (see Table 16.1). Ill-advised process digitization projects can end up generating even more waste than before. For instance, Hershey’s rushed digitization process resulted in their underestimating the scope of their implementation project. As a result, exacerbated by the fact that Hershey’s chose to not pilot test their newly digitized process before full deployment during a busy holiday season, their network was unable to handle the massive flow of Halloween and Christmas orders, and as a result the company failed to process $100 million worth of Hershey’s Kisses and Jolly Rancher orders despite having more than enough inventory.

Artificial Intelligence (AI)

When you think of AI, you are probably conjuring up images of robots ranging from cute Disney characters such as Wall-E to the villainous Terminators. While those are certainly some of the most visible and famous examples of AI, the technology is already being used in many fields. For example, one of the most widely used AI is in the form of chatbots. Bank of America uses its chatbot, Erica, to help customers with some of the most common, simple bank tasks. Anthem uses Sydney to help patients with common, simple healthcare questions. Sony uses a less-creatively named PreChat to help Playstation users to troubleshoot common issues before escalating to live agents.

Just what separates AI from the old expert systems? Expert systems were in use long before the rise of AI. Similar to AI, expert systems also queried a manually maintained database designed to answer common questions. So, AI and expert systems certainly have a lot in common: both rely on a repository of knowledge; both rely on the use of keywords, often in the form of pre-formed, choice-driven interface to interact with users. What makes AI “smarter” than expert systems is the fact that AI is designed to adapt and automatically add to the repository of knowledge based on user interactions. Moreover, AI also interprets user input without requiring users to select from pre-formed questions. In that sense, expert systems are more like choice-driven frequently asked questions (FAQs), while an AI chatbot is more like talking to a robot.

Beyond chatbots, virtual assistants such as Amazon’s Alexa, Google Assistant, Apple’s Siri, and Samsung’s Bixby all perform similar functions. They learn and adapt to their immediate environment while connecting users with information they can search automatically on the internet. However, voice commands must still be programmed into the AI, which limits their ability to respond both flexibly and on demand. Perhaps most beneficial to organizational leanness is the speed at which AI can determine an optimal course of action based on all available givens. Below are some considerations for AI application in lean:

• Quality Adjustment. Precision sensors can monitor production output for quality assurance. When output exceeds parameters, AI can automatically detect environmental changes and other factors to determine the origin of defects and make adjustments, if possible. Outcomes: Reduced defects, faster production, improved quality.

• Order Management. Orders transmitted can be managed by AI to determine criticality and prioritize accordingly. Advanced AI can potentially manage multiple systems for automated inventory staging and order fulfillment. Outcomes: Reduced order mistakes, faster order flow, and labor savings.

• Expert Support. Companies such as Volkswagen have long relied on augmented reality to assist mechanics in auto repair. Visual-based AI can examine machine parts and detect potential defects and suggest repairs to substantially accelerate the diagnostic and repairs process. Outcomes: Faster problem-solving, reduced downtime, labor savings.

As you can see, most applications of AI are designed to yield some common outcomes: faster decision-making and improved outcome quality. More importantly, application of AI is very broad and can help organizations across all industries.