Digital Twins: A Marketer's Guide
You’ve probably heard of digital twins and may be wondering, what is a digital twin and is it applicable to marketing? The concept of digital twins first came about in manufacturing in the early 2000s, where simulations of manufacturing systems, machines, and processes are created for the purposes of predictive maintenance or quality control. As with most breakthrough concepts, the idea of digital twins has made it out of manufacturing and into marketing and customer engagement. So the question becomes, if we can make digital twins of machines, can we make them for customers too, and what are the possibilities?
Well, the answer is yes, we can make digital twins of the customer for marketing purposes. A digital twin of the customer is a virtual representation or digital avatar of a customer. It can be used to learn potential customer behaviors and simulate or anticipate customer outcomes. Customers are most often individuals—but can also be personas, groups of people, or even machines.
But how will we use digital twins of a customer, you ask? I believe the biggest boon to marketing will be in what is called customer journey optimization (CJO). CJO relies on reinforcement learning to guide customers to a conversion event based on historical data merged with current data as well as similar marketing action behavior patterns. Over time, the system learns what works best for certain situations and responds accordingly. CJO frequently applies learning outcomes to digital representations of a customer before delivering the next best experience in a customer journey to an individual customer.
By understanding customers and their potential behaviors and simulating outcomes, marketers can help guide customers to an end conversion event versus forcing them down a predefined, brand-initiated path. Digital twins of the customer can be inserted into these CJO scenarios to present actual customers with the best course of action based on a mix of other customers/digital twins of the customer profiles and behaviors.
How Do We Create a Digital Twin of the Customer?
Perhaps the most important step is to devise a plan for how (a use case) a digital twin of the customer will be used, what data is needed (organic and synthetic), and what the end goal is (it should be to support the two main goals of digital transformation—increased business efficiency and an improved customer experience).
Once a plan is in place, data collection becomes important to build customer profiles. Zero- and first-party data collected off of branded and owned sites will become the foundation for this profile. Second-party data can augment the customer profile through data exchanges and data clean rooms.
As third-party data depreciates and becomes more difficult to obtain, organizations must become competent at simulating data (synthetic data generation) to build and augment virtual customer profiles. Synthetic data generation is used to fill in gaps where data is incomplete or missing. By simulating what one complete customer profile looks like based on data from a similar customer profile, synthetic data generation can use lookalike modeling to aid personalization and targeting efforts.
These digital customer profiles, or digital twins of a customer, must be stored in an easily accessible data environment (such as a customer data platform) for use by the broader business. Additionally, brands must account for and understand data privacy and cyber-risk concerns as simulating data and environments will increase legal and regulatory risk.
Ensure you have the right people on staff with the ability to create all different types of analytical models, ranging from prescriptive to artificial intelligence/machine learning-based reinforcement learning style models. These will be needed for customer journey optimization work.
Engagement methods (especially that those that collect additional zero- and first-party data to build these digital customer profiles) must be clear and transparent. Value exchanges must be created that encourage participation with a customer benefit involved.
What Are Some of the Biggest Challenges to Doing This Well?
- Augmenting the customer profile with synthetic data that is realistic and useful. Gartner predicts that by 2024 60 percent of the data used to develop AI and analytics applications will be synthetically generated. Generating this data natively inside of MarTech applications is new to the market.
- Organizations must be transparent. Initially, it will be hard to convince customers that they need to help your brand create a digital twin of themselves. They will want to know why you are creating, where and how you are storing it, and why they should help you create it to help improve their overall experience.
- Getting the privacy, security, and compliance aspects right. If consumers feel that digital twin data could be used against them, they will not participate. Thus, this aspect must be clear and well-documented.
As digital twins of the customer become more popular over the next five to 10 years, we will certainly see them move into marketing and customer experience applications. I expect that software solutions will have simulation capabilities built in—where brands can simulate audience behaviors, actions, and outcomes directly in their customer engagement platform.
Digital twins will help digital marketing teams provide better customer experiences, by testing and learning about optimal outcomes. Additionally, any type of marketing KPI and expected/needed results will be “simulated” within the tool. For instance, if a brand desires to get to a needed threshold from a conversion or engagement perspective, it will be able to simulate and understand exactly the type and size of audience it should target.
If you are interested in further exploring digital twins of the customer, many vendors already have some simulation capabilities (optimization, RL, predictions/projections, etc.) embedded in their software. But this is only the beginning, as digital twins will be necessary for all the new channels (in-app, in-game, in-stream) and environments (metaverse) of the future.
Jonathan Moran is head of MarTech marketing at SAS, with a focus on customer experience and marketing technologies. Moran has more than 20 years of marketing and analytics industry experience, including roles at Earnix and the Teradata Corporation in presales, consulting, and marketing.