Compared to end-customer-focused industries, the process industry faces different challenges and increasingly faces untapped potential in the use of data analytics. With increasing digitalization of central business processes and the pressure to focus more on the end customer, this could soon change. We show what companies should bear in mind on their journey.

When it comes to developing new recipes and user solutions, the chemical industry is already increasingly relying on state-of-the-art data and AI technology: For example, a German specialty chemicals manufacturer uses software to shorten the development time for plastics based on existing recipes and raw material information. Another company in the industry has also been working with IBM for some time on research into high-performance polymers using AI.

However, when it comes to analyzing data along a company's own value chain, there is a great deal of uncertainty in the industry as to which data can be used to create value. This is where enormous potential for process optimization and cost savings lies dormant. Current challenges of interrupted supply chains and rising energy prizes emphasize it concisely. Especially when looking at the background of current challenges in the chemical industry, the evaluation of data to improve business processes is becoming increasingly important.

Structural deficits prevent effective use of data

The reasons for the hesitant use of existing data in the company so far are manifold: On the one hand, structural obstacles, such as existing data silos between departments, a lack of higher-level data architecture ensure that data is neither available company-wide nor can be made usable.  A fear of losing intellectual property also has a negative impact on the availability of data due to the slow progress of cloud adoption. On the other hand, a lack of understanding about which kind of data is generated in the company, which already exists, how a data set is defined and what it can be used for.

A wide range of potential applications

There are many fields of application in which data science projects in the chemical industry already deliver measurable added value. For one thing, this concerns classic process improvement in the operations environment, for example, through the reduction of set-up times or higher overall equipment effectiveness, as well as data evaluation for optimized planning and forecasting processes and better decision making in the indirect areas.

For another, driven by the increasing digitalization of distribution channels and a stronger end-customer orientation, the focus in the chemical industry is on aligning a company's own offering with dynamic customer wishes as well as optimizing customer interactions and increasing service quality. This is because customer service, as the last stage in the digital customer journey, often determines whether a customer experience is perceived positively or negatively. In short, customer experiences are increasingly becoming the focus of attention. In the end data science projects, in combination with experience data, can help us better understand and respond to the needs of different target groups.

Understanding data as a prerequisite

In order to achieve these added values, it is first necessary to understand which data needs to be considered and what its significance is. First it is important to distinguish between "O data" and "X data". O data, or operational data, describes all types of data that can be obtained from objective, observable and measurable processes and systems, such as information generated by machines or during the order process. In addition to key production figures, this includes, for example, MES or ERP data, as well as CRM data or access figures for certain websites.

X or experience data, on the other hand, represents intentions, attitudes or other subjective feelings that are not objectively observable. This includes, for example, metrics such as Customer Satisfaction Score, Net Promoter Score or similar.

Profitable combinations of data types

While each data type can provide valuable insights on its own, the real potential lies in the ability to combine O and X data to provide even more comprehensive and meaningful insights. For example, anomalies in the operational data can be explained by additional information from a customer survey and vice versa. In addition, for example, the effects of certain production parameters on the behavior of customers can also be predicted.

Put simply, a data type by itself only shows that something is happening. It is only through the combination of different data types that it becomes clear why something is happening. For example, if customers select a certain product configuration but then do not complete the purchase, experience data help identify the reasons why.

Example: Sample process in the chemical industry

A typical example of a process from the chemical industry where this combination of O and X data can be used to implement process improvements is the production of samples in the chemical industry. Here, material properties or data from the manufacturing process, such as viscosity or processing temperatures from individual batches, can be compared with feedback from customers in order to optimize product quality in the long term. Whereas this customer feedback was previously unstructured and had to be aggregated manually, it can now be systematically recorded and correlated with operational data through the use of experience management platforms.

Iterative approach

Due to the variety of possible data sources and the associated dimensions of information, it is essential for the success of data science and data analytics initiatives to precisely describe the question that needs to be answered using the data at the beginning of the project. In a joint workshop with the client, it is important to work out the type of result desired in the interests of process improvement rather than formulating problems. This is because, often, the causes for certain deviations or abnormalities lie elsewhere than originally assumed. An agnostic approach is therefore particularly critical for success.

Based on this question, data scientists formulate hypotheses together with domain experts who contribute the necessary technical expertise, and test them using the data available to them. To do this, they consult various data sources and examine possible influencing factors and their effects on the target dimensions in an iterative procedure.

Result: Data-based recommendations for action

In this way, various insights into the issue under investigation can be gained, from which concrete recommendations for action can be derived. This ranges from identifying the causes of problems in processes and possible countermeasures, through to predictions for the future, to designing personalized customer or employee experiences.

In addition to data-based recommendations for action, data science projects typically also provide quantitative information on the financial impact of the identified problems or their countermeasures. They help to create the necessary sense of urgency for the announced measures at the management level.

Conclusion: Start smart

The use of data analytics in the chemical industry holds enormous potential for process improvements, cost savings and higher service quality. Current developments in the industry, such as the increasing digitalization of distribution channels, a stronger focus on the end customer, and, most recently, more resilient planning and supply chain processes, are important drivers that fuel the need for data-driven analyses.

Most companies in the industry still have a few milestones ahead of them on their journey. Above all, a comprehensive data strategy or data architecture for the systematic provision and utilization of existing treasure troves of data is still missing in most organizations today.

But even independently of this, significant added value can already be achieved for your business using focused individual initiatives in line with the motto "start smart, stay smart".