Learn what supply chain analytics is and how your business can use it to improve the efficiency and operations of its supply chain
The supply chain is complex and difficult to understand. For any business producing a product or service, they must procure various raw materials and component parts, manage their inventories and production processes, and ultimately get a high-quality product to their end customer in a timely manner.
However, a supply chain is rarely static, and things can change fast. Therefore, it’s important to understand what’s going on and come up with actionable insights about how the supply chain can be improved. This is where supply data analytics comes into play.
In this article, we’ll explore what supply chain data analytics is and how it can be used to improve the performance of your supply chain.
Supply chain data analytics is the computational analysis of the systems and processes involved in a company’s supply chain. This involves integrating and evaluating data from all parties involved in producing a good of service — including suppliers, distributors, shipping providers, warehouses, and more.
By integrating data from various data sources such as inventory management and procurement systems, analysts can study how materials or component parts are being sourced, assembled, transported, and ultimately provided to an end consumer as a finished product or service.
With the data integrated and centralized, it can then be processed and visualized to understand the relationship of these systems and processes to show what’s happening, why it’s happening, and how they can be improved to make the overall supply chain more efficient.
Supply chains are complex, unpredictable, and frequently change. Supply chain data analytics allows companies to monitor and analyze these complex structures so that the processes and operations can be adjusted to optimize the supply and demand relationship with high quality, on time, and lean production.
For example, many manufacturing companies are dependent on raw materials that fluctuate in price and supply over time due to broader macroeconomic, environmental, and weather conditions. To further complicate things, customer demand for the end product can fluctuate independently of the raw materials due to things like seasonality, weather, or other factors that are out of the control of the company building the product.
Having supply chain data analytics systems in place can help businesses understand how these different supply and demand conditions affect factors such as inventory levels, delivery timeliness, and labor needs so that management can adjust operations in response to — or ideally in preparation for — these changes.
Supply chain data analytics involves capturing data from multiple sources, integrating that disparate data, and analyzing the results through ad hoc analyses, visualizations, and reports.
Since most supply chains involve a network of different organizations and business departments, there are typically numerous different systems that are producing relevant data. For example, a company that is ordering raw materials from multiple different providers will need to reconcile and link their internal inventory data with external vendor systems and shipping providers to track how component parts and raw materials are flowing into the company’s warehouses and production centers.
However, integrations alone are rarely sufficient for starting an analysis. Data needs to be cleaned, standardized, and integrated so that it can be easily queried and reported on. This is particularly the case when pulling data from different companies, as they likely have different standards for how data is structured, how products are identified, and how the data interfaces are exposed. Data processing pipelines allow businesses to take these varying standards and bring them into a consistent format for analysis.
Finally, with the data integrated and processed into a centralized repository, analysts can build out models, visualizations, and reports while supply chain professionals can explore the data directly with modern tools. This gives both groups the ability to start to understand what is going on in these datasets and draw inferences about the health of the supply chain.
Supply chain analytics is broad in that it attempts to understand not only what happened and why it happened, but often also attempts to use this data to make predictions about the future and what to do about it. We can actually break this into five different types of analyses that answer these various questions:
Descriptive analytics is the process of interpreting historical data. The result is often used to answer the questions “what happened?” or “what is happening?” in a business’s operations.
In the context of the supply chain, this involves understanding the historical state of everything from inventory levels to delivery times to consumer reviews. These different pieces can be viewed over time to see how they fluctuate or are changing over time.
Diagnostic analytics takes descriptive analytics one step further and tries to understand the relationship between various data points. More broadly, it answers the “why” of the historical data that descriptive analytics surfaces.
So, for example, if descriptive analytics shows that inventory levels are insufficient to meet customer demand and that deliveries from suppliers are also slow, diagnostic analytics tries to determine if there is in fact a relationship between the two and if slow deliveries caused a shortage in some key component that caused insufficient production to meet consumer demand.
Predictive analytics takes the results of both descriptive and diagnostic analytics to make projections about the future using things like statistical analysis, often computing correlations between various datasets to determine some type of causal relationship.
To continue with our supply chain example, let’s say we’ve determined with diagnostic analytics that slow deliveries from suppliers leads to an inability to meet consumer demand due to low inventory of key components of a product. Well, is there a way we can predict when this could happen?
Maybe, for example, colder months of the year cause road problems that lead to frequent delays in deliveries by suppliers. By running a statistical analysis between the delays and local weather data, analysts can determine if there’s a strong correlation between the two and decide whether it warrants a further investigation to determine causality.
Embedded analytics can also help businesses uncover opportunities for new revenue streams. By adding analytics, the data your business creates and captures can be monetized. For example, a real estate listing company might capture data around listing trends and embed analytics into their app. Their team can then analyze how real estate prices have trended over time, dig into particular zip codes, or identify how proximity to a particular location (e.g. a school or freeway) impact pricing. They can then use the resulting insights from this ‘research’ to build and sell a new product.
Prescriptive analytics tries to take the analysis of predictive analytics to determine what can be done based on the projections. Rather than just showing the plain facts, it draws broader meaning about the data so that business leaders can try to solve issues before they arise.
If we’ve determined from our prescriptive analytics that there’s a strong correlation between cold weather and delivery delays due to road problems, the next question is how we fix this. Analysts can then run simulations of various different proposed solutions to see how they affect key metrics such as delivery timeliness.
For example, maybe different shipping methods are more reliable in the winter. Or maybe the company should boost inventories before the winter in preparation for delays. Data analysis and modeling can help to automate and test these scenarios before operators decide to invest in the ultimate solution.
Cognitive analytics is the use of machine learning and AI to make sense of complex and interconnected data. It effectively tries to run predictive and prescriptive analytics models on wide ranges of datasets to try to understand relationships that would otherwise be infeasible for a human to manually experiment on.
For instance, a cognitive analytics model may try to map the delivery delay and inventory data on various other data sources to find unexpected relationships. Maybe it finds that there’s actually a stronger correlation with natural gas prices that often rise in the winter, leading analysts to believe it’s actually an issue with suppliers’ input costs that are putting seasonal strain on their businesses. Without being able to run these broad types of regression analyses, it would be very difficult for a human analyst to navigate these seemingly random relationships.
While there are so many different types of businesses and supply chain structures, there are a number of features of an effective supply chain analytics system that can be applied in all situations:
This is used by analysts to represent raw datasets in a visual format that a human can easily understand. For example, bar charts, histograms, and line graphs are all examples of visualizations that can be created to show how different parts of a supply chain are operating.
Supply chains consist of a lot of sensitive operational data about a business. Therefore, it’s important that data is stored using industry standard security practices and is access controlled to ensure that only those who are allowed to access it do.
Supply chains aren’t just about the raw materials and shipping providers. Anything involved in the process that produces relevant data should be integrated to ensure that analyses do not miss relationships and correlations that ultimately affect strategic business decisions.
The physical supply chain processes should be replicated in a digital modeling system — often referred to as the “digital twin”— so that analysts can easily iterate on and experiment with adjustments before the business makes material investments in adjusting the physical systems.
In addition to internal integrations that pull data from things like inventory management systems, analysts can and should pull information from external sources that affect their supply chain, such as public weather datasets or feeds from social media APIs.
While data needs to be secured and access controlled, it should also be easy to share amongst analysts to help with the collaborative and iterative process of improving models and visualizations. In addition to increasing creativity, it’ll reduce the duplication of efforts, particularly with data pipelines that are often generalizable.
Supply chain analytics is relevant to nearly all types of businesses, as its purpose is simply to understand and drive action for operations around producing a product or service. Here we’ll explore some common examples of how businesses can leverage supply chain analytics to optimize inventory, improve production quality, and hedge raw material costs.
Consumer demand for products can vary widely between businesses. For instance, some businesses are highly seasonal and can have big spikes in different times of year or around holidays, while others may be much more consistent over time.
At scale, there’s often sufficient data to project demand into the future using statistical analyses of sales data. By doing so, the business can understand what upcoming sales may look like to ensure there’s sufficient inventory in place while keeping costs low by not producing more than is practically needed.
End customers provide a lot of actionable data through online reviews, support phone calls, social media posts, and much more. Much of this data is available to companies to integrate and aggregate so that analysts can start to stitch together common complaints or feedback.
This customer feedback data can then be used to understand potential issues with product quality or even ways the product could be modified to better fit consumer needs. Ultimately, the end consumer’s opinion on a product or service is what matters most so understanding their experience is invaluable.
For many businesses that rely on standard commodities like oil or grain, there can be a lot of fluctuation in pricing of necessary raw materials due to macroeconomic conditions that are out of the control of the business. Many businesses use historical data about things like prices and weather to predict what may happen to their costs so that they can hedge them with financial instruments.
For instance, it’s very common for airlines to lock-in their oil prices using futures contracts — where they commit to buy at a specified price and time in the future. While it’s certainly difficult to predict the way that a commodity is going to move, this practice has the added benefit of allowing businesses to know exactly what their costs are going to be over the duration of that contract and plan accordingly rather than being exposed to market fluctuations.
Supply chain analytics allows businesses to integrate and analyze all of the data related to their supply chains to better understand what is going on and how it can be improved to drive better efficiencies. However, doing so requires a robust analytic stack that can integrate disparate data sources, aggregate them using secure data pipelines, and build reports and data visualizations in a collaborative manner.
ThoughtSpot provides all the tools necessary to build a robust data analytics stack that can power your supply chain analytics efforts. Try a free trial of ThoughtSpot to see how you can improve your supply chains today.
Join the Modern Analytics Cloud revolution and try
ThoughtSpot for free