Case Study: Capital One achieves keyword-level bidding accuracy with Blue Orange Digital

Capital One improves keyword conversion value prediction 85% with Blue Orange Digital

Capital One faced a challenge with its Google Ads strategy, as it relied on third-party tools and manual judgment with no clear view of which keywords drove account conversions. This last-click attribution model missed most of the customer journey. Blue Orange Digital was engaged to address this sales and pricing inefficiency.

Blue Orange Digital built a neural network attribution model in TensorFlow that predicts the conversion value of each keyword. This model feeds an automated bid adjustment engine through the Google Ads API, allowing high-performing keywords to receive higher bids and underperformers to get lower ones. The solution achieved 85% accuracy in predicting conversion value per keyword and automated the previously manual bid change process.

Capital One