Zehra Cataltepe is the CEO of TAZI.AI, a Predictive AI platform for Business Users. She has more than 100 AI papers and patents.

As of 2022, financial institutions have, on average, a 3.65% marketing ratio (their marketing spend as a percentage of net revenues), but how much of that marketing budget is spent on “listening” to their customers? Contrary to the wisdom that “the art of conversation lies in listening,” it’s likely only a negligible amount.

In a recent article that I co-wrote for Nasdaq, I explored the various challenges that boards face with AI risk management, especially when trying to understand how to unlock growth and shareholder value. The last item on the checklist was: “Continuously listening to your customer’s voice and making necessary updates on your systems to ensure your solutions serve their best interests.”

In this article, building on that idea, I will share how to “listen” to the voice and behavior of your customers, your competitors’ customers and your team. This three-pronged approach ensures that you stay up to date on customer needs to reduce churn while increasing loyalty and acquisition growth.

Connected AI To Listen And Serve Your Customers

AI and machine learning (ML) are already being used to understand customer communications data (e.g., Voice of Customer), customer behavioral data (e.g., segmentation) and structured data (e.g., churn prediction). However, most approaches process only one data type.

Since this data remains siloed, these approaches lack the flexibility to adapt to changing business needs.

Connected AI bridges data, business expertise and AI models across silos. Connected AI can help you “connect” generative AI and ML outputs, bridging the gap in between to serve your customers.

Start by listening to your customers’ and competitors’ voices (VoC). To do this, an AI solution can connect to public data sources—such as Google Maps, Yelp, BBB or Trustpilot—and create real-time dashboards. You can detect crucial insights in communications, such as topics, sentiment, complaint severity or compliance.

Once implemented, business teams should have tools to reconfigure classifications and provide feedback for prompt updates or model fine-tuning. This customized sorter can then enable service improvements (e.g., fraud detection, customer support), better complaint handling, competitor analysis, employee training and even complaint prediction for new product updates.

Churn prediction and segmentation models are often ML-based and rely on customer data, product usage, behavioral changes and external factors. These models can benefit from VoC-derived signals. Additionally, retrieval-augmented generation (RAG) AI solutions can reduce churn by tailoring personalized communication for each customer (such as investment news bulletins) based on their segment, market conditions and needs.

Connecting Feedback To Retention And Acquisition

Negative customer and employee feedback often correlates with future churn. However, feedback systems are rarely linked to customer accounts, complicating churn prediction.

AI can extract information from customer communication data, enrich it with additional details (e.g., geolocation or product usage) and link complaints to churn predictions. Real-time VoC solutions can also enable quicker issue resolution and deeper insights into customer needs to help you connect customer communications with retention and acquisition.

For most markets, continuous competition analysis drives growth. Customer reviews provide opportunities to evaluate competitors’ weaknesses and strengths. By linking customer communication and churn data, you can predict competitor churn and, if prepared, enable customer acquisition.

Acquisition models typically have less data to work with, so taking into account signals used by churn and segmentation models—and also utilizing external data—is helpful.

Employees Owning AI to Achieve Company Goals

Your employees serve customers through the tools provided to them. If they don’t lead, own and update AI-driven customer communication strategies, the results will not be optimal. Large companies often rely on in-house AI teams, but the changing needs frequently result in updates that are “lost in translation.”

Mid-sized companies should look to partner with AI vendors that provide configurable solutions, rather than struggle with inflexible pre-built solutions. You must ensure that solutions include data management, audits and guardrails to mitigate risks. Solutions should also be easy to monitor, manage and update by your business intelligence or business domain experts.

Employee KPIs should be a part of the AI strategy, not only to measure the effectiveness—such as campaign success and retention rates—but also to help employees evaluate, prioritize and give feedback on AI-suggested actions.

Finally, your employees have to understand how your AI solutions work and be able to provide feedback to improve them. They need to act as humans-in-the-loop, humans-on-the-loop and humans-in-command.

How you prepare them and your organization depends on your priorities for the next few quarters. For example, instead of letting yourself completely drown in the endless core system or acquisition tasks, think about how your current and future customers—and your employees—can be best served given this new wave of AI.

Zehra Cataltepe, Forbes Council Member, Forbes Technology Council

Also Published on Forbes:
https://www.forbes.com/councils/forbestechcouncil/2025/01/27/connected-ai-using-customer-behavior-to-predict-and-improve-retention/