Merchant attrition is one of the biggest (and most expensive) problems in the payments industry. Processors manage thousands of merchants, and 30% of total revenue churns every year. Losing high-value merchants hurts, and once they’re gone, winning them back is a costly uphill battle.
Merchant Attrition
At Arcum, we use AI-driven predictive analytics to help payment processors get ahead of churn before it happens. By analyzing transaction data, pricing structures, and merchant behaviors, we give teams the insights they need to retain merchants before they even think about leaving.
Why Tree-Based Machine Learning Works Best
We’ve tested every machine-learning method out there—from logistic regression to deep learning models—but when it comes to predicting churn in structured, tabular data, tree-based models like Random Forest consistently deliver the best results. Here’s why:
The Nature of Churn: Why Merchants Leave
Churn isn’t random. It’s a decision-based process—merchants leave when something isn’t working for them. Their logic is simple:
✅ Happy with pricing & service? Stay.
❌ Frustrated with fees, service, or competitor offers? Leave.
Tree-based models naturally reflect this decision-making process, which is why they work so well in predicting churn risk.
Why Logistic Regression Falls Short
Traditional logistic regression focuses on averages, making it less effective at capturing extreme cases—which is exactly where churn happens. Most merchants stay put, but the ones who churn often do so under specific, high-impact conditions.
For logistic regression to be useful, analysts have to manually define churn indicators like:
The problem? That takes time and expertise—and it’s never 100% accurate.
Tree-Based AI Achieves Higher Accuracy
Tree-based models like Random Forest and Gradient Boosting take the guesswork out of the process. Instead of relying on pre-set indicators, they learn from real merchant behavior to identify risk signals automatically. That’s why our AI models achieve 80% precision in churn prediction—compared to only 40% for logistic regression at the same recall rate. The data confirms it: tree-based methods actually mimic how merchants make retention decisions.
The Next Step: Generative AI & AI Agents for Automated Outreach
Predicting churn is only half the battle—the real value comes from acting on that insight before it’s too late. That’s where Generative AI (Gen AI) and AI-powered outreach agents come into play.
Right now, Arcum gives processors the data they need to proactively engage merchants at risk of leaving. But the next evolution of our platform will take things even further with AI-driven outreach automation.
AI Agents: Automating Retention at Scale
We’re building AI agents that will handle merchant outreach automatically, so teams can:
🤖 Engage at-risk merchants at scale – AI agents will send personalized emails, make calls, and even handle SMS-based outreach on behalf of customer success teams.
📈 Optimize messaging in real time – Instead of generic retention emails, AI will generate merchant-specific outreach based on processing history, pricing concerns, and competitor rates.
⏳ Speed up response times – AI-powered virtual assistants will handle initial merchant questions and flag high-risk cases for human follow-up.
For example, if a merchant is flagged as high-risk due to pricing concerns, an AI agent will automatically send them a customized retention email or call script, factoring in transaction volume, processing history, and even competitor pricing to offer a compelling reason to stay.
Hyper-Personalized Retention with Gen AI
Generative AI will help fine-tune every interaction, so teams aren’t just reaching out faster—they’re reaching out smarter.
✅ AI-crafted emails and call scripts that adapt to each merchant’s situation.
✅ Smart recommendations for what to offer based on risk level and competitive pressure.
✅ Automated campaign execution, ensuring high-risk merchants are engaged at the right time, with the right message.
By integrating AI agents with Gen AI, Arcum will not only predict churn but also prevent it—without requiring manual outreach from customer success teams.
AI-Driven Retention: The Future of Merchant Services
We’re already helping processors predict churn with cutting-edge AI models, but we’re not stopping there. Our next step is to help them act on that data faster and more effectively with AI-powered retention agents and Gen AI-driven outreach.
The end goal? A fully automated, AI-powered retention engine that helps processors retain more merchants, reduce churn, and maximize revenue—all with minimal manual effort.
If you're looking to take a smarter, AI-driven approach to merchant retention, Arcum is here to help. Let’s talk!