We’re living in the experience economy. And like it or not, the standard is being set by firms that excel at hyper-personalization: Amazon, Starbucks, Netflix, Stitch Fix, Google, Sephora, Facebook are just a few of the everyday brands that we all know.
What do they have in common? They each have moved well beyond “personalization” protocols that are built with rules-based segmentation and behavioral triggers. Instead, they use the power of predictive personalization to anticipate what their customers need and want, presenting them with tailored and creative options delivered in a channel that makes them more likely to engage.
Amazon’s Jeff Bezos credits the company’s success to the velocity of its test-and-learn agenda. The more tests that they run – every day, every week, every month – the more growth they unlock. That’s a simple enough recipe for a digital-first, direct-to-consumer company.
But it’s a lot more complicated for banks that struggle with staggering levels of tech debt, siloed and disorganized data and ossified organizational structures that make cross-functional collaboration and customer-centric agility difficult. For them, achieving a test-and-learn pace that enables predictive personalization may seem out of reach. But waving the white flag simply isn’t an option.
This is especially important for banks because the center of gravity for the customer relationship continues to shift to digital channels even as the quality of those relationships has fallen dramatically. Curinos’ SalesScape benchmark shows that customers acquired through digital channels bring in 10 times fewer deposits, churn at a much higher rate and buy fewer additional products. If that trajectory continues unabated, a critical component of the business model is at risk.
Like at Amazon, adopting marketing technology with AI at the core is essential to unleash predictive personalization for banks. But what exactly does that mean? The market is awash with narrowly-focused, AI-driven point solutions (optimize text! optimize images!) and broad-sounding AI capabilities from core providers.
Because banking is a regulated industry, financial services marketers must pay even closer attention than typical retailers to the way they reach out to customers. Generic “Marketing AI” technology must be finetuned to operate in financial services, a process that can take many quarters to define and a veritable army of engineers to develop.
Here are five things bank marketers should consider when contemplating a marketing AI platform.
- Familiarity with financial services data. Platforms that are tailored to the world of financial services will have clearly-defined data models that lighten the lift for overstretched data engineering resources and speed up time to deployment. Even better is a familiarity with core data systems and implementation partners who can pull the necessary data directly.
- A robust financial services metrics library. Algorithms that clean and enrich customer data — in effect generating new and useful attributes that drive differential marketing interaction decisioning — are some of the most important determinants of quality. The extent to which your marketing provider has a robust library of battle-tested metrics for financial services data will dramatically speed up your launch dates – and your race to ROI.
- Orientation to relevant customer applications. Marketing technology that is built for financial services will have out-of-the-box configuration (product definitions, eligibility rules, optimization metrics) so that marketers can be productive immediately without needing to expand their team, agency or consultant rosters.
- Optimization against downstream KPIs. Unlike general retail, the way in which a customer ends up using a financial product has a direct impact on profitability. That’s why it’s important for marketers in this vertical to be able to move beyond clicks and opens to optimize interactions on the true drivers of business value (primacy scores, trailing deposits, trailing loan utilization).
- Strong model guardrails, ability to audit and compliance controls. Ensuring no bias becomes embedded in AI software is absolutely essential to avoid the possibility that certain groups are unfairly rewarded over others. Your AI marketing engine cannot be a black box. It must be easily configured on the front end to ensure decisioning is never driven by potentially bias-reinforcing data variables (e.g., gender). In addition, the platform should be fully auditable on the back end – with clear visibility into the decision logic behind the delivery of a particular marketing experience.
Today’s consumers expect the companies they deal with to be “learning” about them as they move around digital environments, adjusting their product and marketing interactions to better meet their needs on the fly. The hard reality is that financial institutions have a long way to go to make this kind of predictive personalization a reality. Marketing technology holds significant promise, but the market is awash in solutions claiming personalization. Knowing how to evaluate the sea of options and identify the right provider is critical.