AI’s use in marketing will continue to skyrocket in the years to come. With the wave of hype surrounding the release of OpenAI’s ChatGPT, many marketers are thinking through how to best utilize it. Learn how it can help serve the next best action to customers from Allister Rebeiro, vice president of customer strategy at Merkle.
OpenAI’s recent public launch of its generative chat application, ChatGPT, has been making waves over the past few months. When prompted about the top use cases for artificial intelligence (AI) in marketing, ChatGPT responded with a list that included personalization, predictive analytics, and chatbots.Â
Let’s start with a quick refresher on the concepts used here. AI is a technique that enables a computer to mimic a human cognitive function. Machine learning, developed after AI, is the ability to learn without being explicitly programmed. The latest frontier is the development of neural networks and deep learning, which develops algorithms based on the way the human brain processes and distributes information.Â
Next Best Action and AI
Next best action allows marketers to generate a decision, such as an offer to speak to a sales rep or a product recommendation at checkout for customers or prospects in near-real time. The real-time aspect differentiates next best action from other marketing decisions because it is based on a specific event that is happening at that moment rather than being pre-planned like a seasonal direct mail campaign.
A key driver of the next best action is to create the most optimal experience for a customer. A retailer wants to create the best shopping experience for its customers, regardless of the platform they choose to shop on. As a customer engages with the brand, the retailer will want the ability to respond to those interactions in ways that support the customer’s preferences and needs.Â
While most marketers can do this successfully at some level for specific customers and action combinations, doing so at scale poses a challenge. This is where AI can change the game by bringing in the ability to understand context, analyze behavior, and generate a decision dynamically and in near-real time. At the most basic level, the next best action is powered by rule-based algorithms. Both A/B testing and multivariate testing are evolutions of rule-based algorithms and precursors to machine learning. Machine learning can identify behavioral cues at scale, picking up the right action when there is more than one option available.Â
Screen grab of question asked to AI
Source: OpenAIOpens a new window
Â
How Can AI Improve Next Best Action
How exactly can AI take the existing next best action further? And how should market leaders leverage AI in their strategies today and going forward?Â
See More: How AI-Based Marketing Is Transforming Telecom Companies
Improved acquisition in media
For many industries, the cost of prospecting and sourcing leads can be expensive. Most marketing teams aiming for 1:1 personalization work with limited data on an individual prospect’s needs and preferences, and the cost to find and engage this prospect could be high. AI can make this acquisition effort more efficient by improving ad targeting and placement.Â
Most segmentation models are refreshed infrequently, given the time and effort required. By analyzing and learning from prospect behavior at scale, AI can identify themes and automatically add or remove prospects from defined segments to allow for more precise targeting. Once a segment has been assigned, AI can select relevant messaging by inferring intent based on content and behavior. An AI-powered recommendation engine analyzes browsing and buying behaviors and generates unique product matches using their interests, demographics, and look-alike models.
Deeper engagement
Having existing customers usually means you have data that can be used to drive personalized decision-making to deepen your customer relationships. One of Merkle’s clients, a leading national bank, uses next best action technology to enhance the customer journey and experience across different channels. Call center support agents see the next best product recommendations during all inbound calls, and these recommendations are based on what has or has not worked in the past. Intelligent/predictive call routing (IVR) is used to match the customer profile and topic of the call with a care agent’s specific skills. And finally, every time a customer completes an ATM transaction, a product recommendation is made and is followed up on through the customer’s preferred channels.
Stronger retention
By continuously monitoring customer behavior, AI can be deployed for proactive churn management. Many organizations use the next best action to reduce churn. They do this by studying customers’ behavior and giving suggestions to the call centers and stores on how to keep them. By identifying and focusing on higher-value customers, AI can help brands identify drivers of customer attrition, understand factors that influence retention, and optimize their retention budgets. ​
See More: How AI Is Gaining Rapid Acceptance and Adoption in FinTech
Bringing it Together: A Case Study from the Insurance Industry
An auto insurance provider wanted to create a better experience for its customers who were starting a new quote. The company had already placed a significant amount of effort into driving these customers to the home page and wanted to make the experience as easy as possible. An analysis of traffic on the quote process revealed that drop-off rates were significant when customers were unable to fill out the form completely, either because a question was not relevant or because they did not have the information ready at that moment. This presented a challenge for the insurer because it was unable to price out the quote with an incomplete questionnaire.Â
To address this, the client leveraged AI at three steps of the process:Â
1.The first was to develop a set of next best questions that were dynamically selected based on whether previous questions were answered. The questions were created using available data points on that customer. Based on the level of completed responses to these dynamic questions, the insurer was able to present the customer with a quote price range instead of a targeted quote.
2. The insurer also leveraged AI to determine the next best channel for a follow-up (email vs. a phone call, depending on the nature of the quote and linear TV).
3. In the last step, where the quote was presented, the insurer was also able to analyze the many possible product combinations and propensities to develop a next best product recommendation based on the initial quote presented to the customer.Â
Risks and Opportunities
While AI can generate the next best actions in near-real time, some of these actions will require time to deploy. Content that is generated will need multiple layers of internal review (brand, marketing, legal) before it is seen by customers. As AI continues to evolve, machine and deep learning techniques will improve the output and the strength of the recommendations and the next best actions that they generate.Â
Improved marketing outcomes from implementing an AI-driven next best action include better response times, the accuracy of content in those responses, and lower costs while sending them out. To make this marketing end state a reality, AI will need a strategy, data to learn from, executive sponsorship, and buy-in from the broader organization.Â
Are you leveraging AI to improve the next best action for your customers? Tell us how on FacebookOpens a new window , TwitterOpens a new window , and LinkedInOpens a new window .
Image Source: Shutterstock