Capital One

Deep social listening AI


Capital One is one of the largest banks in the US ranked #145 on the Fortune 500 and serves approximately 45 million customer accounts. Headquartered in McLean, Virginia, Capital One offers a broad array of financial products and services to consumers, small businesses and commercial clients in the U.S., Canada, and the UK.

As the business was shifting focus on the way they managed brick and mortar locations with the Capital One Café. Capital One asked IV.AI to develop an AI model to help tackle various elements of their communication strategy and Café growth to better quantify quality. The team looked at the challenges and barriers which surround communication strategy for businesses and the way value is assigned to the methods companies can use to connect with customers.

IV.AI's solution started with an AI strategy and a white paper

IV.AI worked with Capital One to understanding market needs, messaging, and analytics:

  1. Collected and mapped all the banking and finance industries messaging and use of language
  2. Built an ontology specifically based on Capital One’s previous content and product messaging
  3. Tested and assigned confidence to the AI model to understand how the language resonates
  4. Compared Capital One’s messaging to messaging from competition with similar product offerings for target markets
  5. Showcased results with the Capital One team to asses retraining and final model for sign-off

Creating neural networks and supervised learning models that processed large corpora of data sets:

  1. IV.AI built an AI index that was validated against the messaging of the brand to help find new data from public mediums
  2. Customized AI NLP assigned tone and sentiment analysis to the various types of language used by the brand to gauge effectiveness
  3. This resulted in a unique Quality Index that showcased methods Capital One could use to validate how they communicate with customers and spend on media to connect with new audiences
  4. Developed an AI dashboard designed to showcase insights into the market, helping all parties make more informed decisions throughout their normal workflow

The results gained traction in the organization and Capital One expanded the workflow from the communications team to include CRM, Social, and Marketing.

Capital One Quality Index

With a working knowledge of how quality could be assessed by teams, IV productized the Capital One Quality Index with reinforcement learning to test and monitor various unique locations across the US. The AI was trained to focus on locations relevant to the national Capital One Café and map out the shape of each market at a local level.

The result was an advanced AI system that learns about a conversational ecosystem at a local level and finds relevant concepts and influencers in the various markets. The first of its kind AI model enables a national team to communicate the local level using targeted messaging and influencers to insure brand safety and efficient marketing.

IV.AI’s solution was a Deep Social Listening AI

The team took concepts from the strategy and built a product that solved the biggest pain point:

  1. A data lake was created by IV.AI with all the data necessary to build the model
  2. The data was cleaned and filtered to ensure clean signals without human bias
  3. Linguistic trait ensemble models designed to group like concepts were applied
  4. Application of filters based on personality and speech similarities across the group
  5. Iterations shared with the business team to gauge quality of the output to KPIs

Deep learning model designed to focus on KPIs and business objectives:

  1. The discovery and training phase resulted in an AI NLP that was able to quantify messaging in local regions to inform team strategy
  2. Customized weights were applied to identify tone and sentiment analysis
  3. The Capital One Quality Index was tuned to validate the output of the brand messages with the best audiences in the target geographies
  4. Modules were added to the AI dashboard designed in the discovery phase to showcase insights
  5. A product funnel designed so the model can learn from success metrics and drive growth was created

The first of it’s kind deep social listening model is able to streamline research into new customer markets and simplify complex data signals to help teams make more informed decisions

Get in touch

IV.AI makes it easy for every company to become an AI company. Get in touch to see success metrics, live AI you can test, and let us help guide you through your AI engagement.

Francisco Brito
Chief Product Officer