Zühlke – Empowering Ideas

Industrial Sector

On the way to becoming a data-driven industrial company

#ML 4 Industry

In this article, we show how industry can actively pursue the route to the data-driven enterprise both during and after the crisis.

7 minutes to read

How you can use artificial intelligence on the way out of the current crisis

Businesses have long understood that the step towards data-driven organisation represents the final stage of refinement in digitalisation. To achieve this goal, strategy roadmaps were developed, and promising pilot projects were defined. However, the current crisis is presenting businesses with challenges whose vital solutions are eclipsing strategic initiatives. In this article, we show how industry can actively pursue the route to the data-driven enterprise both during and after the crisis.

Assessing the company's situation using the 3-phase model

The economic consequences of the pandemic cannot yet be accurately assessed. The effects depend on the speed at which markets, industries and businesses can recover. The decisive factors here are the level at which an economy, an industry or a business entered the crisis, the extent of government aid packages, the containment measures taken and how quickly a vaccine can be developed. In August 2020, the industrial sector recorded negative growth of 11.1% in Europe [1].

The market research company Gartner has defined a 3-phase model that makes it easier for businesses to determine their own situation. According to the model, there are three phases:

1.    Respond / Fight for survival
2.    Recover / Revive
3.    Renew / Make plans.

Classifying your business using this model helps with formulating your own options for action and the next steps, and with achieving a better understanding of their interdependencies.

3 Phases EN
The 3-phase model from Gartner [2]

Are strategy roadmaps to the data-driven enterprise now obsolete?

The question that currently arises is that of prioritising measures. Gartner's 3-phase model can help to set priorities and reposition strategic roadmaps in medium- and long-term planning.
As a data-driven business, you want to use data, analytics and artificial intelligence (AI) in all functions and areas: you make better decisions based on data, you use AI and machine learning (ML) technologies to make processes leaner, faster and more customer-oriented, and you use the possibilities of these technologies to develop radically new products and services and thus open up new sources of revenue. This endeavour remains justified during the crisis, and the roadmaps already drawn up remain valid.
In the following, we look at each of the 3 phases and give our recommendation for action with regard to the route to a data-driven company.

Responding – Securing resources

The main activity in this phase is securing your own resources. You get an overview, take immediate measures and ensure survival. This is done by negotiating with partners, freezing projects and radically reducing costs. The focus is on the development of a cash flow strategy. If no systems have been established for this purpose in the past – for example, for analysing the impact of decisions – there are no readily available technological solutions that can provide automated support in this area. At best, ad hoc business intelligence and manual data analysis contribute to the assessment of the situation.
As the illustration shows, at the beginning of a crisis you are in all three phases simultaneously (with different intensities). So despite the acute challenge, it's worth looking ahead as soon as the hectic pace has eased somewhat.

Recovering – Targeted implementation of individual ML use cases, e.g. to reduce costs

In the second phase, the aims are to revive the business and to restructure the investment plans. The goal of this phase remains cost optimisation. For example, quality, efficiency or productivity can be increased. Initiatives for automation and digitalisation will be reviewed and measures reprioritised. In this phase, it is worthwhile to identify specific Machine Learning use cases that make a particular contribution to reducing costs. It is best to start with a lean process analysis in an interdisciplinary team.

Take the example of automated quality control using computer vision: with automated image control and new unsupervised ML approaches, defects down to the smallest scratches in product surfaces can be detected quickly and reliably, even if a system has never seen them before. Alternatively, semantic segmentation can be used to map image elements with pixel accuracy, which in a second step contributes to the measurement of objects. The dimensions of gaps, for example, or the rates at which liquids spread out can thus be measured and assessed in real time. Today, these possibilities compete with laser-based processes, but are often more cost-effective, faster and can be applied more flexibly.

In the analysis, it pays to think holistically. Potential use cases can be found in the reduction of production rejects, in improved service management through an increased first-fix rate or in areas completely unrelated to production processes.

It can also be worthwhile to automate the preparation of quotations. Significant cost reductions are possible here, if individual quotations have to be prepared again and again even for the smallest production parts. An algorithm can translate the customer enquiry directly into a quotation, which an employee then only has to approve. This is a win-win situation: for the end customer, because he finds out more quickly whether he will get his tailor-made product and how much it will cost, and for the supplier, because the sales team can concentrate on the really interesting cases. The throughput of quotations is thus increased many times over.

It is therefore worthwhile in this phase to examine the entire value-added chain for weak points and to rectify these in a purposeful and targeted manner.

Renewing – Agile establishment of the data-driven company

In the third phase, strategic considerations towards digital transformation predominate. Developing new business models has always been difficult, but digital transformation and data-driven strategies give you a new perspective. This way often generates ideas that bring real competitive advantages.

Instead of implementing isolated Machine Learning projects, it is therefore worth taking a strategic approach. Zühlke has developed a process model for this purpose; a model that has proven its worth. Based on a clear vision, a data strategy is developed and a project pipeline for data & ML is set up.

The Zühlke process model for data-driven businesses
The Zühlke process model for data-driven companies

In the first step, businesses prepare themselves for the transformation. A core team, including top management, first develops a vision for the data-driven future. It is important that this vision also has motivating power and that possible worries in the workforce are noted and taken into account.

The second step is about the "What" Based on the business strategy, a data strategy is then developed. For this purpose, an initial portfolio of specific projects and initiatives is developed.

Finally, in the third step, the foundations (capabilities, technical data platforms, structures and processes, etc.) are created, and projects and initiatives are implemented in parallel. This is the central functional principle of our process model: foundations are created step by step and always in relation to a specific use case. This ensures that we establish practical and lean structures that can be tested and, if necessary, adapted on the basis of a specific use case. For the first implementation project, we choose a use case that has a high chance of success. This "lighthouse project" should have a positive impact on the business and thus further strengthen the internal acceptance of the business transformation.

The approach illustrated represents an agile, practical and proven way to establish the data strategy. The foundation for a data-driven company and thus for an innovative, future-proof business can be laid in 4-6 months (depending on the scope of the first implementation projects).

Zühlke as a long-term partner on the way to becoming a data-driven enterprise

The current crisis may have slowed down or completely stopped the drive for further digitalisation and the implementation of strategic projects in the short term, because first and foremost you have to push ahead with digital optimisation in order to reduce your own costs. However, we at Zühlke are convinced that the crisis is a catalyst for prioritising the strategic project roadmap. In this sense, the crisis does not change the goal, but ensures that businesses on the way to their digitalised future are offered an intermediate step in which to focus more strongly on process optimisation.

We would be very happy to accompany you on this journey.

[1] Source: Eurostat, Impact of Covid-19 crisis on industrial production, August 2020 [2] Source Gartner phase model
Tobias Joppe
Contact person for Germany

Tobias Joppe

Director Customers Solutions

Tobias Joppe studied automation and control engineering at the TU Braunschweig and was most recently head of a innovation team at Siemens AG. He has been with Zühlke since 2008, is a partner and, as Director Customers Solutions, is responsible for the Trend Lead Data Science in Germany. In his role, he builds the bridge between cutting-edge technology and current customer needs. Together with customers, he translates visions and goals into a strategic roadmap and concrete project procedures. As Director Customers Solutions, many completed interdisciplinary projects form the basis of his experience.