Data analytics is the practice of examining raw data to find insight and even answers to a company’s select questions. For businesses like manufacturing, it’s used to figure out key information such as why plants are breaking down or identifying trends like customer purchases during hurricane season. Data analytics is usually broken down into the following types:

  • Descriptive – “What happened?”
  • Diagnostic – “Why did this happen?”
  • Prescriptive – “What should we do next?”
  • Predictive – “What might happen in the future?”

We’ll be examining the last type, predictive analytics (PA), in how it’s used in many industry segments like manufacturing.  

Predictive Analytics: Forecasting the Future

As the name suggests, predictive analytics is used to predict future trends and events. Predictions can be something as simple as the malfunction of a piece of machinery within 24-hours, to complex and far-reaching calculations such as the company’s cash flow for next year.

PA does so by using historical data like inventory logs, headcount, and accidents. Analysis of such data is done manually by data analysts, or through artificial intelligence using machine-learning algorithms. All look for key trends and patterns. A mathematical equation called a “model” is created from this analysis. Companies then use this model with their current data to hopefully predict future events. 

The industrial / manufacturing segment has turned to PA across many sectors. The ones covered here include:

  • Predictive maintenance
  • Predictive Inventory
  • Predictive Supply Chain
  • Predictive Marketing

Preventing Malfunction / Predictive Maintenance

A typical example of PA in manufacturing involves determining the likelihood of equipment breakdowns. This is unsurprising, given machine downtime can cost millions, even billions, of dollars a year. Senseye, a predictive maintenance software company, reports large manufacturing plants lose 323 production hours a year. That’s a cost of $532,000 an hour in lost profits, repair costs, and lost production time for a cumulative $172 million per plant per year. Manufacturing and industrial firms among the Fortune 500 lose an average of $864 billion annually from all this downtime. 

“Unplanned downtime is the curse of the industrial sector,” notes Alexander Hill, chief global strategist at Senseye. “When expensive production lines and machinery fall silent, organizations stop earning, and those investments start costing rather than making money. The costs can spiral to well over $100,000 per hour for large manufacturers in almost all industrial sectors.”

Manufacturers turn to PA to limit or prevent such costly impacts on their production lines. Sensors on the involved machines and even across the entire plant provide raw data like temperature, running time, power level durations, and error messages. Many would be housed in rugged mini PCs which are built to withstand the harsh environment. Features like fanless to deal with dust, as well as industrial grade parts make them all around more able to withstand much more abuse than off-the-shelf PCs. Through PA, companies can then plan ahead when to shut machines down for repair, replacement, or upgrades. This reduces losses from such downtimes.  

Predictive Inventory

Businesses like manufacturing define inventory as “any item of property held in stock by a firm, including finished goods ready for sale, goods in the process of production, raw materials, and goods that will be consumed in the process of producing goods to be sold.” It’s important they keep careful track of their stock to avoid both overstocking and running out of needed goods. 

The consequences of not doing so can be dire. Sports brand Nike has been on-and-off dealing with inventory problems since the early 2000s. And in the mid-2000s, US retailer Target pulled out of the Canadian market after two short years partially due to inventory errors

Using a subset of predictive analytics called predictive inventory, companies look to determine the optimal inventory level to meet current and future demand. Industrial tablets can record huge numbers of items thanks to built-in barcode and RFID scanners. They’re also fit for a warehouse setting since they’re portable and rugged enough to be used in the busiest warehouses. Those records are added to other points of data like customer buying habits and even season and region. Algorithms then forecast the optimal demand for various products. This allows companies to order the right amount of material to be held in their warehouses as well as avoid overstocking, understocking, or becoming burdened with unsellable goods.

Predictive Supply Chain

Predictive inventory in the previous section forecasts the right amount of goods to be held by a company. The related predictive analytics used on supply chains covers the logistics of creating such goods (if the company’s a manufacturer) to transportation of products to consumers.

The COVID-19 brought the importance of supply chains to international attention. Yet the pandemic was only one factor contributing to the four trillion US dollars in lost revenue suffered by multinational companies back in 2020. As pointed out by John Piatek, vice president of GEP, a leading provider of procurement and supply chain strategy, software, and managed services: “While COVID-19, understandably, gets all the press, it is far from the only force wreaking havoc with the world’s global supply chains that costs most companies double-digit revenue loss and, perhaps more important, immeasurable reputational damage and customer loyalty. Supply chains and procurement are a key driver of sustainable competitive advantage, but despite spending millions on ERP solutions, most global companies are ill-equipped to effectively manage complex global supply chains in face of uncertainty, global warming, tariffs and trade wars, and national governments increasing control of natural resources and strategic industries.”

Transportation of goods is one aspect companies are turning to predictive analytics to deal with supply chain woes. Data on toll roads, traffic conditions, and weather reports, for example, are pulled from their appropriate information centers. Installed fleet management hardware and software in the various vehicles provide information on fuel consumption and driving behavior. Other sensors in the vehicle monitor conditions like wear-and-tear. PA can then crunch all this data to determine the fastest and most cost-efficient routes in a given year; optimum times for drivers to rest; or when vehicles need maintenance and even replacement. 

Predictive Marketing

Unsurprisingly, companies are always looking for ways to best market their products to customers. They are constantly analyzing market trends, buying habits, and personal details. Predictive marketing, another subset of predictive analytics, can be leveraged not only to create strategies to better reach customers where they are now, but may be in the future. 

Some of the best examples of PA marketing can be found in retail like Walmart. In a 2004 interview, the retail giant revealed it makes sure its stores are fully stocked with strawberry Pop-Tarts during hurricane season. Why? “Over the course of their experience with hurricanes, Walmart has learned that Strawberry Pop-Tarts are one of the most purchased food items, especially after storms, as they require no heating, can be used at any meal, and last forever, ” says Steve Horwitz, an economist at St. Lawrence University, who studied the phenomenon. 

Amazon has taken predictive marketing even further. In 2013, the online giant submitted a patent for “anticipatory package shipping.” Briefly, the AI-driven system “anticipates” customers’ buying habits and has products shipped out and stored at the appropriate distribution center before the customer has even submitted their order. That minimizes the delay when the customer finally presses the “Purchase” button. 

“It goes beyond just being able to forecast we need a hundred blouses,” points out Jenny Freshwater, software director of Amazon’s Supply Chain Optimization Technologies group. “We need to be able to determine how many do we expect our customers to buy across the sizes, and the colors. And then … where do we actually put the product so that our customers can get it when they click ‘buy.’ “

Manufacturers don’t normally market directly to the consumer. However, the above strategies still apply to business-to-business marketing from the development of white papers to content generation.

Market Outlook 

So what’s predictive analytics own predictions about itself? It is looking very good. A report from Mordor Intelligence valued the market at 10.01 million USD back in 2021 which is expected to nearly triple to 29.97 million by 2026. 

Much of the growth will be led by e-commerce. According to the National Retail Federation, for each company closing a store, 2.7 companies are predicted to be opening up new ones. The US leads in predictive analytics, with early adoption of the technology not only in retail but manufacturing and even healthcare.  

Closing Thoughts 

Manufacturers generate a large amount of information from conditions on the factory floor to state of inventory in a warehouse. Thanks to predictive analytics, that historical data can be mined, filtered, and used to forecast future events. These can be used by companies to plan accordingly to minimize problems from machine downtime to anticipating customer needs.

To learn more about how this is done, and if predictive analytics is right for your company’s wants and needs, contact a representative from Cybernet. Also follow Cybernet on Facebook, Twitter, and Linkedin to stay up to date on this and other relevant topics.