Tracking the Evolution of Predictive Routing Using Scalar Function Models

Theoretical physicist Geoffrey West is well-known for his research on physiological network scaling. Recently, his work has been applied to systems like cities and socioeconomic models. And we can apply his theories on scalar functions to almost any system—even the contact center. 

Scalar functions are a measure of how a system’s characteristics change as the size of the system changes. With these functions, not all characteristics change uniformly with a change in general size; however, these differences follow predictable models that can be defined as algorithmic constants.

To understand how this relates to the customer experience routing in the contact center, it’s important to differentiate between two distinct scaling models—isometric scaling and allometric scaling.

Isometric scaling occurs when proportional relationships between features are preserved as size changes during a system’s growth or evolution. For example, as surface area increases, the volume equals the square of the relative increase. Allometric scaling occurs when there is a deviation, either positive or negative, from an isometric model. However, these deviations aren’t random, and they comply with and can be quantified through derivable functions.

In his research, West found that a key factor in deriving scalar functions was the dimensionality of the system. As contact centers evolve into customer experience centers, we see growth along several axes because customer experience is a key indicator of success to many companies.  

Each axes is seen as a dimension within the system; each of these dimensions contributes to the increase of routing complexity as the size of the system increases. In customer experience interaction routing strategies, dimensions correlate with:

  • Agent count  ̶  total number of agents in the system
  • Agent skills  ̶  granularity and volume of skill types and skill gradations
  • Client persona  ̶  potential granularity and number of client characteristics and classifiers
  • Channels  ̶  number of channels, endpoints or modalities supported
  • Interaction leg quanta  ̶  number of legs in an omnichannel interaction or customer journey
  • Competing business/operational objectives

By viewing the evolution of “customer experience routing strategies” as “growing networks,” we see that complexity grows isometrically with the growth of the network itself. And, considering the number of dimensions, this complexity quickly affects the opportunities and challenges of achieving business objectives and creating great customer experience.

How AI addresses increasing contact center complexities  

Many organizations grapple with this challenge of ever-increasing complexity. Traditional routing tools that are based on prebuilt state machines and Boolean expressions leave companies drowning in an abyss of expanding targets, next-best-action algorithms, millions of skill-set permutations, and databases that are overflowing with client information. In addition, many companies are expected to leverage all of the above factors to achieve business objectives and create great customer experiences.

Modern tools like artificial intelligence (AI) let us abstract away the complexity and automate the building of a customer experience-decision engine, allowing for continual improvement that’s tested and validated—in real time. Machine learning algorithms are designed to solve these complex problems. With AI-based routing, you can reduce the allometric scaling of adverse properties involved in routing strategy complexity.

Genesys Predictive Routing is an AI-based solution that optimizes business outcomes. Using the treasure trove of data you collect, predictive routing applies machine learning algorithms to identify the factors that drive outcomes. Predictive models are built based on these attributes and are then used—in real time—to predict the optimal agent-to-customer match to obtain the desired business outcome.

Although these are still early days for AI and machine learning, the inevitable reality of increasing complexity—even in the contact center—qualifies that the adoption of solutions like Genesys Predictive Routing is not a matter of if, but when. Check out the 30-minute Tuesday Teachings on predictive routing and learn how to optimize routing decisions by applying machine learning for the desired business outcome.