Optimising Digital Outreach to Consumers
Challenge
A large US-based financial institution wanted to increase loan applications and was investing about $150,000 a month on branded paid search.
The organisation wanted to learn how branded paid search was impacting loan applications considering changing consumer behaviours due to the pandemic.
Test
Two sets of people with similar attributes in different postal codes
• how does one group react to branded search ads?
• how does the other react to no branded search ads?
Results
• Removing branded paid search resulted in a 2.1% decline in loan applications and no impact on booked or approved loans.
• Volume of loans driven by branded search was insignificant.
• Branded paid search performed lower than expected and the bank’s website was more effective for driving loan applications
Next best actions
The organisation moved its marketing dollars to more effective channels.
Expanding the Customer Relationship
Challenge
A bank had many customers who only had one or two products with the bank, such as a checking account or a credit card. They wanted to find ways to incentivize their customers to expand their business with the bank to its other products and services. The bank tested whether its 2 million credit card customers would add wealth management to their relationship with the bank.
Test
The analysis included identifying which customers were the most likely to expand the relationship. They analyzed two of their customer segments:
• Single-product customers
• Multiple-product customers
Results
Through propensity modeling, the bank narrowed down those customers who would most likely be interested in other products within the bank.
Next best actions
The bank identified several customer segments to focus their outreach for expanding the relationship. The bank then developed a rebate offer for customers who signed on for wealth management services as an incentive
Adjusting Operating Hours Across Branches
Challenge
A Brazilian bank (+500 branches) is looking to optimise its branch hours to increase sales and transactions. They want to find the best strategy for adjusting branch hours across their network of branches, and need to determine the impact of these adjustments on sales and transactions.
Implementation
The business will conduct a randomised experiment across a subset of its branch network to determine the impact of adjusted branch hours on sales and transactions. To do this, they will randomly select a group of branches to adjust their hours and compare their sales and transaction data to a control group of branches that maintain their regular hours.
Test
– 20 branches: close them two hours early
– 20 branches: keep them open two hours later
The experiment will run for a period of three months, with the adjusted hours in effect for the duration of the experiment. The business will collect data on sales and transactions for both the experimental and control groups, and will also collect data on customer feedback regarding the adjusted hours.
The business will then analyse the data to determine the impact of the adjusted hours on sales and transactions, as well as any trends in customer feedback.
Result
The experiment revealed that adjusted branch hours had a positive impact on sales and transactions. Adjusting by one hour made the most sense:
• Opening earlier in the morning led to a 5% increase in sales and closing earlier in the evening had no impact
• Certain branches, such as those in less competitive areas, were less sensitive to scaling back hours
Next Best Actions
Adjust branch hours in a targeted approach across the network.
