Bond Builds a Model Factory to Power Personalization, CRM and Loyalty

Bond Brand Loyalty

 

In today’s fast-moving world, speed to insights is critical to the success of a brands’ personalization and customer engagement initiatives. But how can marketers win the race?

We sat down with Francis Silva, our VP of Data & Analytics, to learn more about the challenges with today’s personalization models, and how Bond’s innovative new Model Factory helps businesses solve complex challenges and get to the finish line by using insights faster and more efficiently. 

GettyImages-528068934

What challenges do typical personalization models present for businesses?

One key challenge for many organizations is that personalization can be extremely time-consuming. Take propensity models, for example, which predict consumer behaviours such as, ‘What is the probability that this customer will respond to this offer?’ and infer things like, ‘How does redemption behaviour influence churn?’ These are vital inputs to not only personalization, but marketing strategy and campaign design as well. To make those predictions effectively, an organization would need to build thousands of models—it’s not just ‘one and done.’ 

For example, at the most basic level, you might build a churn model for three customer value segments. Next, you’ll want to create a time-to-next-purchase model and a customer lifetime value model, and so on, and the number of models you have to build and manage explodes. In order to even build these models, you’ll need to clean and transform your data and perform feature selection, which both can be tedious and time consuming. Once you build those models, you have to manage them. So, for many businesses, building the data set and managing the models is a very manual and time-intensive process. 

Another challenge is that it’s difficult to scale. The time and effort to create and manage propensity models is mostly spent on data wrangling, model selection and optimization instead of interpretation and validation. That limits scale and the ROI for each model. 

Bond recently launched its new proprietary application, Model Factory. How does it solve clients’ challenges and what are its main advantages? 

Think of Model Factory as a way of automating and rapidly accelerating speed to insights. As a module of our suite of technology solutions, Model Factory allows us to rapidly create, deploy and manage models used for personalization, CRM and loyalty. While we started with propensity models, Model Factory supports many functions including customer segmentation, customer lifetime value, time to next purchase, regression modelling, and forecasting. 

In terms of specific advantages, there are three main ones. The first is optimal model selection, which allows users to test thousands of scenarios and parameters through a ton of different evaluation metrics that we’ve pre-configured into Model Factory. The second is fast time to insight. The platform uses our existing Customer Genome, another proprietary Bond tool that allows us to take a client’s data, clean it and get it ready for modelling. That means no time is spent on building a new data set. And thirdly: easy experimentation. Model Factory allows our clients to test numerous hypotheses in a time- and cost-effective way—even with really large enterprise level data sets. 

What technology partners is Bond working with? 

factory bond partners techModel Factory exists on Microsoft Azure and we’ve partnered with three other best-in-class technology providers. Databricks gives us a scalable compute and analytics development environment and does the heavy lifting in the machine learning process; Snowflake allows us to easily serve up the outputs to our analytics users for further analysis, and it powers the data cubes behind PowerBI; and PowerBI provides slick visualizations and dashboards that allow users to see the outputs of models and how well they are doing. 

Was accessibility a big consideration when developing Model Factory?

Absolutely. What’s really exciting is that Model Factory democratizes data and is a big step into self-serve tools. You don’t have to be a data scientist (or work with one) to understand the results. It truly empowers clients and our own client teams to be closer to data, which in turn helps them solve complex business challenges and deliver improved bottom line results. Marketers want to get more personalized with their customers and be more data-driven with their decisions. Model Factory creates an infinitely larger toolbox to help them get there faster and more efficiently.