By Samah Chowdhury
So-called “digital twins” are dynamic, virtual replicas of complex systems. Organizations often use them for scenario planning because they blend real-world elements with simulations and a constant flow of data, helping evaluate the consequences of different decisions. For example, when BMO acquired 503 Bank of the West branches in 2023, it used Matterport’s capture services to create dimensionally accurate 3D digital twins of all the branch locations within three months.
This technology enabled BMO to complete remote site assessments and operational tests within the model without service disruptions. The results: Over $500,000 saved in 15 months, 6,000 survey hours recouped across 503 locations, and branch resources and documentation centralized.
The use of digital twins began in the 1960s when NASA used twin models to monitor and adjust spacecraft during space missions. Recently, the Biden administration announced a $285 million investment in digital twin technology for semiconductor manufacturing based on its potential to enhance efficiency, innovation, and resilience in the U.S.
How do they work?
A digital twin comprises three core elements: the physical system (product, process, network), a virtual model representing it and a data connection that updates the model in real time. The virtual model mirrors the physical system’s current state and behavior, continuously synchronized with data from sensors and internet of things devices. This setup allows the digital twin to simulate and predict the physical system’s performance under various conditions. Bringing all three components together requires several key technologies. First, the collection and use of data involves cloud computing and platforms for storage and processing. Second, AI and machine learning are needed to enable simulation models that provide advanced analytics and accurate virtual models. Lastly, augmented reality and virtual reality enable advanced visualization and interactions between the digital model and the physical system.
What problem do digital twins solve?
While data is the mantra of our modern age, data sets taken in isolation are of limited value because they tend to be sparse, noisy, and often indirect. Because systems exist across a web of components, any micro change results in a ripple effect, making accurately replicating a system extremely difficult. In banking, digital twin technology’s true potential is harnessed when integrated with a bank’s proprietary knowledge along with an inflow of external stimuli into decision-making models. With data flowing from multiple channels, using a mirrored environment enables precise contingency and incident response plans. When changes are made, other parts can adapt accordingly, simplifying coordination with business units and third parties. For example, a digital twin of a bank’s technology stack can predict outcomes of certain technology changes with the potential to evolve based on results from prior simulation runs. Digital twins can also mitigate risk across evolving fraud vectors through intelligent, comprehensive, data-driven strategic planning.
In the banking industry, digital twins may seem like enhanced scenario analysis. And if this is what you’re thinking, we don’t blame you. But here is where the key difference lies: data. Traditional scenario analysis relies on static data while digital twins use real-time dynamic data and facilitate bidirectional data flow. This means that a digital twin can take insights it produced and trigger changes to optimize the physical system it replicates, whereas scenario analysis merely provides an output that must be reviewed and acted upon separately.
Let’s look at a few potential use cases for banks:
- Stress testing. A digital twin could enable banks to simulate various scenarios, such as economic downturns, market fluctuations, or operational disruptions, to assess their resilience and performance under stress. Banks could identify weaknesses and mitigate risks preemptively by inputting diverse parameters to the digital twin. Add real-time insights and your bank can continuously adjust strategies that bolster resilience and stability.
- Digital financial twin. This is an approach where digital twins could be used to precisely map financial and nonfinancial metrics across the life cycle of a bank product. The digital twin would be set up to link metrics related to the product’s service, partners, customers, and employees, resulting in efficient and quality decision-making. To go further, the digital twin would combine with real-time data from an enterprise resource planning system to ensure the highest level of resource optimization, drive sustainability and accelerate product development.
- Predictive transformation. Digital twins could be developed to replicate a bank’s entire operation. Getting an enterprise view can offer banks the ability to simulate and evaluate the effect of technology transformations. For instance, a digital twin could offer the most favorable and lowest-risk path to cloud transition from on-premise technology. Or a digital twin could help a bank manage increasing transactions and unify customer experiences across channels based on current interactions and historical performance. A similar strategy could be applied for new product rollouts, comparable to an intelligent pre-production sandbox, where a bank is empowered to troubleshoot and fix problems before going to production.
Are digital twins a must-have for your bank?
To answer this, consider whether the investment is appropriately weighed against the economic return of developing digital twins. Not every product, service, or process is complex enough to warrant the intense sensor data flow digital twins demand. If your bank decides to explore digital twin implementation, you might consider starting by identifying a complex problem. To maximize the value and effect of this technology, consider reserving it for problems characterized by a high degree of variability, situated within intricate systems, and involving outcomes that rely on accurate predictions. Currently, there are no purpose-built digital twin developers specific to the banking industry. Until there is significant demand, we recommend focusing on foundational elements, such as data readiness, to prepare for future applications of this technology.
Samah Chowdhury is senior director of innovation strategy in ABA’s Office of Innovation.