Why Acceptance is More Important Than Accuracy in Digital Twins

[Erkan Teskancan]

## Why Adoption is More Important Than Accuracy in Digital Twins

When industrial companies implement digital twins, the goal is to achieve optimal results more consistently. An optimal result is defined as one that reduces costs, increases production, and positively impacts operations. Digital twins enable organizations to reach such optimal results more regularly by making multiple variables visible in decision-making processes.

Allowing all employees access to a tool that contains the collective corporate knowledge enables them to do their jobs better. However, there's a big "if" here: Digital twins and other smart products can only yield optimal results if employees consistently use them. This is why adoption is crucial.

### Employee Tolerance for Incorrect Results

A digital twin doesn't have to be perfect; what matters is that it behaves consistently and can explain its decisions. Adoption decreases when recommendations don't align with employees' experience and the reasons aren't understood. The solution is to involve end-users from the beginning of the process. Their opinions should be considered when deciding how the digital twin or smart product will be developed.

It should be explained that this process is iterative and that early versions are developed through real-world testing. When employees understand the components of the current tool and how these components can be adjusted based on performance, they become more tolerant of early errors, and the improved product is adopted.

### Prioritize Return on Interaction

A common mistake when developing digital twins is to overemphasize real-world resemblance, especially visuals. Creating visual replicas of physical systems with unnecessary detail requires significant resources. These resources could be allocated to other priorities, such as more frequent data updates or the effective inclusion of more parameters.

When deciding on the realism of a digital twin, the key question is: How much realism do end-users need, and can they comfortably use the tool? For example, a digital twin for equipment maintenance scheduling doesn't need to be an exact replica of the physical system. However, using similar visuals can be beneficial to make it easier for users to get accustomed to it.

Easy and fast user adaptation means more interaction, and increased interaction enhances the opportunity to optimize results. Furthermore, the environmental conditions in which users work (e.g., use of personal protective equipment like gloves and glasses, hearing conditions in noisy environments) also affect their capacity to interact with the digital twin, which directly impacts the return on investment.

### Making the Invisible Visible

Prototype testing is particularly exciting when developing a smart product. While a group of employees makes real-time decisions, the prototype's performance is run in parallel with human decisions. This reveals unseen factors not yet incorporated into the model.

For example, a maintenance planning team might consider the speed at which different customers approve work orders or the working hours of technicians. When the model doesn't yet include these realities, its recommendations might differ from those of employees, leading to a loss of trust. However, if employees are involved in the process and understand that this stage is about revealing the model's shortcomings, trust is strengthened.

### Start with the End-User for Return on Investment

Digital twins and smart products, when well-developed, enable teams to make better decisions. This, in turn, improves business performance. However, this process takes time. The return on investment is possible with a high level of adoption, and adoption is directly related to user trust.

User understanding of how the model works, what data its outputs are based on, and how the model was developed builds trust. Industrial companies can maximize the potential of digital twins by involving end-users in the development process from the outset. This can lead to high returns, both financial and interaction-based.