Value Creation in the Self-Driving Company, Part 1

The economic core of the self-driving company is its value creation: from product idea through development to market launch, fully networked systems relieve people, cut costs continuously, and accelerate every step at consistent precision. This article starts from Michael E. Porter's classic notion of value creation, shows how digitalization changes it, and introduces the five drivers of value creation in the self-driving company. The first two, research and development as well as production forecasting and planning, are covered in this first part; drivers three to five follow in the second part.

From Porter's value chain to networked value creation

The classic notion of value creation comes from Michael E. Porter. In "Competitive Advantage" (1985) he describes the company as a chain of activities that add value to a product: five primary activities, namely inbound logistics, operations, outbound logistics, marketing and sales, and service, and four support activities, namely firm infrastructure, human resource management, technology development, and procurement. Value creation is the difference between what customers are willing to pay and the cost of these activities. To this day, the model is the language in which executive management and controlling talk about margin.

It does, however, date from a time when industrial goods were the main source of value creation and the activities actually took place one after the other. Today, globally successful apps and platforms generate substantial value with extremely few resources, without inbound logistics and without a production hall. And in traditional companies, too, digitalization makes effects possible that Porter's model does not anticipate, because the activities no longer happen in sequence but simultaneously and in a networked way.

In my book "Das selbstfahrende Unternehmen" (Springer Gabler 2021, published in German) I therefore chose a more broadly defined notion of value creation: value creation is everything that contributes to generating revenue. According to the Business Model Canvas in the extended version by Joyce and Paquin (2016), in most companies that is the unique selling proposition, the key resources, and the key activities. Whether the product is physical like a car, a service like physiotherapy, or virtual like a loan, where no money is transferred in physical form but only rights and obligations are agreed, makes no difference. What all products have in common is that their creation rests on processes. And these processes will sooner or later be digitalized and automated, either by the company itself or by its competitors.

Value creation drove every industrial revolution

Every industrial revolution was driven by an increase in value creation, and every one assigned new tasks to people. The steam engine took over transport, mass production the hand movements on the assembly line, digitalization for the first time intellectual work. Since the 2010s, Industry 4.0 has been automating in real time the coordinating and communicating processes that were previously handled slowly and with errors by clerical staff and middle management. This accelerates all processes and at the same time creates transparency in the company. Why Industry 4.0 is nevertheless not the destination is something I explained in the article Industry 4.0 Is Over, Now Comes the Self-Driving Factory.

The self-driving company goes one step further. Products are manufactured in a fully automated way, and isolated, linear processes become holistically networked, multidimensional functions that optimize one another in real time. Manufacturing costs then rest mainly on the one-time investment and on ongoing energy and material expenses, while personnel costs fall drastically. The real driver of this development is therefore an improvement in value creation combined with a reduction in coordination costs. Just as the steam locomotive, mass production, and the PC rewarded those companies that were willing to invest in new technologies, cognitive software will become the success driver of those companies that embrace it and allow the corresponding changes.

In the self-driving company, value is no longer created along a chain but in a control loop in which every function learns in real time from the data of all the others.

The five drivers of value creation

When the individual systems of a company merge, learn to communicate with one another, and optimize each other, the traditional business turns into the high-performance organism of the self-driving company. On closer inspection, five essential drivers of value creation emerge. They can be mapped to the activities in Porter's model, but they fundamentally change their character.

Drivers 1 and 2 stand at the beginning of value creation and decide what is produced at all and in what quantity. In my experience, they are also the ones where many companies are still furthest from the vision today, because they do not take place in the factory but in the minds of management and controlling.

Driver 1: Research and development

While value creation in the self-driving company becomes highly automated, the research and development of new products remains a field of work for people. Research and development are driven by a wealth of information that is available from all areas of the company, the customer channels, the internet, and all networked partner organizations in previously unknown quantity and quality. Simulations and models that today are created in laborious manual work are produced by artificial intelligence on the basis of this data at a speed that manual work cannot match.

The creative spark nevertheless continues to come from people. Their task shifts to where data alone is not enough: bringing complex and hard-to-grasp phenomena such as the zeitgeist, trends, latent needs, or possible risk scenarios into the product ideas. Good market research alone does not produce a good product. Henry Ford is credited with the remark that, had he asked people what they wanted, they would have asked for faster horses. So the point is not to extrapolate an existing development linearly, but to design entirely new approaches and then test how well they are accepted and at what cost they can be manufactured. The coffee capsule is one example: the linear extrapolation would have been an even more complex fully automatic machine; instead, the process of making coffee was radically redefined.

Conversely, the 1957 Ford Edsel shows that elaborate planning, lengthy market surveys, and a large investment of expertise do not guarantee success: instead of the expected 200,000 vehicles, 63,000 were built in the first production year, and after three model years the line was discontinued (example from "Das selbstfahrende Unternehmen", Springer Gabler 2021). This rule also applies in the self-driving company; the difference lies in how uncertainty is handled.

There, product innovations are developed evolutionarily according to the "fail fast" principle. In concrete terms, this means:

• Many ideas in parallel. Instead of one product idea that matures over years, countless parallel ideas are generated and tracked.
• Early market launch. Very early development stages are tried out with individual customer groups; several variants run simultaneously with different reference groups.
• Feedback as the data basis. This quickly produces a broad data basis on the individual variants, which is evaluated in the lab and condensed into the ideal product variant. An expensive market research institute is not needed for this, because all customer reactions are captured anyway.
• Consistent selection. An idea is pursued further only if the desired success is foreseeable. All others are quickly evaluated and ended instead of being kept alive artificially.

A product idea developed in this way is comparatively inexpensive, because the effort only rises once demand has been proven. For digital companies this approach has been established practice for years; in the self-driving company it becomes the standard for physical products and services alike.

Driver 2: Production forecasting and planning

Through the targeted collection, analysis, and preparation of market data, Michael Bloomberg became one of the richest people in the world. The value of his data for customers lies in the improved plannability of their investment and production decisions: they reconcile the external analyses with their internal data and the situation of their own company. Because global markets are changing ever faster, this kind of data analysis will continue to gain importance in the coming years.

In most companies, this process is still carried out by people. Controlling prepares the figures for the chief financial officer, who discusses them with the chief executive; purchasing and staffing decisions follow from a variety of planning meetings and contract negotiations. On the way to the self-driving company, this process is increasingly automated with artificial intelligence and the analysis of large volumes of data. The result is a planning basis with which the activities in production and in the collaboration with suppliers are coordinated in the interest of value creation. Integrated planning of sales and production can then take place largely without human intervention.

One of the most important focus topics of the coming years will be forecasting future production. In addition to market data, sales planning and current sales figures must flow into the forward-looking planning of production capacity. In companies as they have existed so far, this was practically impossible because of departmental silos and separate data pools: what the field sales force knows about the next period never reaches manufacturing. In the self-driving company, all company data generated in real time is accessible to all authorized people and all software systems. Sales processes are standardized, their data is available digitally, and every opportunity in the pipeline carries a probability.

Production can thus access the probability-weighted sales pipeline directly and plan production capacities and inventories using intelligent algorithms. Idle times or fluctuations in production are detected early and automatically cushioned through feedback into sales, for example by stepping up sales activities for products with free capacity. The result is a control loop that handles volumes of data at a speed no planning meeting can match. Management defines the limits within which this control loop operates and the deviation at which it wants to be informed. How this form of decision-making works is described in the article How Self-Driving Companies Decide.

What this means for your company today

The first two drivers do not require a new factory, but a different data basis and a different attitude toward uncertainty. From our projects we see three prerequisites that can be created right now:

1. One shared data basis from sales to production. As long as sales planning, pipeline, and capacity planning sit in separate systems, every forecast remains manual work. The first step is to bring this data together in an architecture as described in the article Enterprise Architecture for Self-Driving Companies.
2. Sales data of forecasting quality. A standardized sales process with probabilities per stage is the prerequisite for algorithms to derive production requirements from it.
3. Experiments as routine operations. Anyone who wants to test many product ideas in parallel needs the ability to create variants cheaply, capture customer reactions, and end ideas consistently. That is a question of software architecture as much as of corporate culture.

Drivers 3 to 5, production automation, robotics and digital twins, and automated warehouse systems, follow in the second part of this article.

The next step

Value creation in the self-driving company is Section 6.3 of the book "Das selbstfahrende Unternehmen" (Springer Gabler 2021, published in German). You will find an overview here: About the book. If you would like to assess how far your planning is from a closed control loop between sales and production, book an expert meeting.