Jimmy KimCEO & Founder, Commerce Roundtable · Marketer of 17+ Years, 4x Founder. Former DTC/Retailer & SaaS Founder. Newsletter. Host of ASOM & Send it! Podcast. DTC Event: Commerce Roundtable

How to Use AI to Analyze Customer Complaints and Fix Retention

11 Mar 2026 · 1 min

Watch on YouTube

0:50

In this segment, Jimmy Kim explains a practical approach for using AI to improve customer retention by analyzing recent negative feedback. He suggests gathering the latest 200 refund requests, 200 negative reviews, and 200 negative email or SMS interactions, then feeding this data into an AI model to surface patterns in messaging and education that may be causing churn. He highlights that the AI can reveal whether the company is overpromising, under‑educating, or miscommunicating throughout the customer journey. The second part of his framework focuses on learning the preferred communication channels of customers—such as WhatsApp and push notifications—so that outreach can be tailored to the mediums customers actually use, addressing a common gap where businesses have the tools but lack the know‑how to run them effectively.

Chapters

If you go and pull your last like 200 refund requests and then like your last 200 negative reviews and then, you know, maybe your last 200 negative emails or SMS responses and you go ingest it into AI and you can kind of find the pattern, basically the pattern of what you're not saying right or you're saying wrong, which is causing this negative effect. To your point, maybe I'm overselling them at the front and I'm giving them too much to expect, or I'm doing a poor job educating them through the entire scenario. You don't know. And I think what's really important is like realizing that. And then that second part comes into like, the channel, right? Like, you'll quickly also learn how they like to communicate, what they want to learn about, and stuff like that. I mean, dude, in this world today, if you think about WhatsApp and, you know, push notifications, these are obvious. Like, yes, there's obviously people on the other end, but like nobody really knows how to really run those things.