Banks and credit unions must integrate machine learning into their digital marketing strategies, but where and how should they get started?
Machine learning is responsible for a slew of radical new advancements across many industries. In just the past couple years, robust machine learning programs have been developed by the likes of Google to improve eye care for diabetics, NASA to analyze written text and learn subjects more quickly, Uber to predict how long it will take to bring you your favorite meal, and Stripe to better detect credit card fraud.
Sophisticated data analytics solutions are evolving quickly and becoming more accessible. Advancements in machine learning, predictive analytics, big data and artificial intelligence aren’t limited to Fortune 500 companies. Now organizations of all sizes can capitalize on the power, promise and potential of the data-driven revolution.
Using data to analyze consumer behavior is nothing new to banks and credit unions. Long before phrases like “big data” and “data mining” were coined, financial institutions were collecting massive amounts of data on people’s saving, borrowing and spending behaviors. And early adopters have been using machine learning to develop better credit scoring, combat fraud, offer faster customer service and automate chatbots. But there is one critical area of the financial industry that has yet to dive deeply into the opportunities of machine learning: Marketing.
Applying Machine Learning to Online Marketing
The homepage on a bank or credit union website is one of the most powerful messaging channels in the financial marketer’s arsenal.
The largest and most valuable piece of real estate on the website homepage is called the marquee banner — the big marketing panel that dominates the screen when a user first loads the site. We can use machine learning to determine the optimial combination of messaging and imagery on marque banners, giving marketers a real opportunity to improve acquisition rates.
To understand the opportunity of the homepage marquee banner as it pertains to marketing, we analyzed the web stats for ten multi-billion dollar financial institutions in the United States. Here are some important findings:
- 47% of a financial institution’s customers see the website homepage each month
- An average of 51% of users are current customers, and they account for more than 79% of the total sessions. The remaining 49% of the visitors are likely prospects, career seekers, or other audiences.
- The average click-through-rate (CTR) on marquee banners across all audiences averaged less than 4%.
- The best performing marquee banners had a CTR of 19% and the worst had a CTR of less than 1%. The marquee banners with listed rates have 2X the click-through-rates of those without rates.
- 87% of the website visitors see the homepage marquee banner at least 1 time each month.
Key Fact: Research reveals that high-performing marquee banners have almost 20X the click-through-rate of poor performing banners.
Currently banks and credit unions use a fairly formulaic and static approach to their homepage marquee banners, falling into the following categories:
(1) Product Promotions. These banners often focus on happy people, low rates (on loans) or high rates (on deposit accounts), and in some cases the context of the product such as a car for auto loans or a house for mortgages. Viewing industry leaders we find that the homepage marquees advertise checking products more than other types of products.
(2) Service Promotions. These banners tout services such as mobile apps, online banking, and personal financial tools.
(3) Testimonials/Case Studies. Institutions offering high-end customer services will sometimes utilize customer case studies to position the brand as high service on the homepage marquee.
Applying Machine Learning
This high value piece of digital real-estate is a prime location to apply machine learning. Algorithms can pair large variations of visitor segments with large variations of marquee banners to find the perfect match. The first step is to create a large variation of visitor segments by attributes such as product preference, customers/prospects, age, gender, and affluence.
The second step is to create variation between marquee banners. This step requires more work, as the marketing team will need to provide assets for these variations — e.g., different background photos or illustrations, a varied set of messages, and different calls-to-action.
A machine learning program can then run thousands of variant tests on the live site between visitor segments and marquee banners to determine which banners get the best click-throughs and product applications for different visitors. This type of machine learning program can be run on a bank homepage indefinitely, and will likely yield better results as more visitor attributes and more marquee variations are devised.
As the banking industry moves to an environment in which all customers are attained and serviced through web and mobile properties, digital marketing is getting more competitive. Finding the right marketing messaging for all customer segments is a highly challenging and highly rewarding endeavor. Machine learning pilot programs are providing leading banks with a competitive edge while helping marketers better understand what types of marketing messages are preferred with various audiences.
The latest digital advertising platforms and c0ntent management tools are making machine learning accessible and affordable to non-technical marketing teams in a variety of industries. With the adoption of these advanced marketing technologies, we are moving into a world in which the most visible marketing asset to digital marketers, the homepage marquee banner, will deliver more personalized messaging, better marketing results, and an overall better experience to all prospects and customers.