How AI helped boost interest in Krungsri Credit Card

Background and aim

Krungsri Credit Card (KCC) wished to increase awareness of and traffic to its credit-card applications, and selected Appier as its technology partner to reduce its customer acquisition costs and improve the quality of customer leads. 

Execution

KCC committed 60% of its lead-generation budget to Appier’s AI-powered intelligent audience solution (Gillnet), to segment audiences and find the most valuable prospective customers. Appier’s AI-powered segmentation works by first defining specific segments based on massive audience behaviors such as frequency and time of visit, and products viewed.

Next, predictive AI technology combines the best segments based on rules such as ‘two visits in last five days’ or ‘item added to card in last 14 days’. Following that, the AI ranks the segment combinations based on their value. For example, customers who visited twice in recent days and added an item are most important. Finally, KCC can reach its top performing segments through social-media platforms such as Facebook.

Results

• AI-powered segmentation reduced the number of overlapping audiences, allowing KCC to better identify and target customers.
• The clickthrough-rate to credit-card applications increased by 138% over the course of the campaign.
• The average cost-per-acquisition exceeded the set KPI by 72%.

KCC was also able to understand the profiles of the best-performing audience segments. These were mobile and desktop users who visited credit-card product pages eight times in one day, and mobile users who clicked ‘apply online’ once a day. These behaviour patterns were captured, and KCC was able to deliver relevant advertisements at critical moments in the user journey.

Comments

Somwang Toraktrakul, KCC managing director:

At Krungsri, we foster an innovative culture and aim to become a financial technology pioneer and provider. Part of this involves leveraging data more efficiently using AI to find and engage with customers, so partnering with Appier to do this was an obvious choice. We have been exceedingly pleased with the results, and we see real potential for further adoption of AI.