Brian Polackoff has founded multiple companies throughout his career. While working with one of these previous companies, Vairkko, he realized just how frustrating client cancellations can be.
“I took it personally, like, really personally,” he said. “Anytime a customer decided to cancel, I genuinely did not like it. I really didn’t like it when it took me by surprise.”
After selling Vairkko in 2022, Polackoff set out to create a solution to customer cancellations. In 2024, he founded ChurnAssassin, a customer health-monitoring service for business-to-business SaaS teams.
Polackoff explained that “churn” is an industry slang term for cancellation. SaaS companies want to maintain as many customers as possible; they can’t survive as a business if they have more cancellations than they do new customers.
“Churn is a massive, massive problem for every revenue subscription-based company out there,” Polackoff said.
Polackoff said that customers cancel their subscriptions for a variety of reasons, but the most important factor for a business is being able to predict if the customer is going to cancel ahead of time. He likened ChurnAssassin to a tornado siren: The software identifies customers who are likely to cancel their subscriptions—if left to their own devices—and notifies the company 6-10 months ahead of time, on average.
ChurnAssasin can be deployed in a matter of minutes, just by copying and pasting a few lines of code into your B2B SaaS company’s own application.
“Once our code is in their software application, that’s it. That’s all they have to do. There are no additional configurations, there are no setups, there’s no training, there’s no anything that they have to do from that moment on, ChurnAssassin runs on autopilot in the background,” Polackoff said.
How ChurnAssassin works
To be able to predict cancellations, the creators of ChurnAssassin built five different machine-learning algorithms, each looking at proprietary and evidence-based churn behaviors that they’ve identified. These algorithms have been trained on the behavioral records and activities of about 260,000 real SaaS users, and they’ve identified 313 churn behaviors that indicate a customer might be considering cancellation.
One behavior that could indicate a customer is going to cancel, for example, is “top user ghosting.” This occurs when an account which frequently uses an application suddenly stops using the application entirely, and it is a major indicator of potential churn.
“This goes unnoticed by many, many customers out there, and it usually is only found in the quarterly business review,” Polackoff said. “So three months after the person has left, that’s when it usually gets identified, and usually by then it’s already too late.”
After potential cancellation behaviors are identified, ChurnAssasin’s five algorithms generate reports which are passed along to the company’s proprietary AI model. The model interprets the data and assigns each individual account a renewal probability score, which determines if the account is at risk of cancellation.
Once an account with a high risk of cancellation is identified, the customer (the B2B Saas company) can take action to retain the account—or ChurnAssassin can take action for them. Polackoff said this could involve an AI-driven customer satisfaction phone call, which is often more effective than a personal phone call with a real customer service representative.
“What we have found is that people are very willing to talk to an AI agent as long as they know it’s AI,” he said. “And what this does [is], this allows the person to be very truthful.”
QUICK BITS
Startup: ChurnAssassin
Founder: Brian Polackoff
Founded: 2024
Team size: 3
Location: Raleigh
Website: www.churnassassin.com
Funding: Bootstrapped
ChurnAssassin charges $2 for each account they protect. For example, if a B2B SaaS company has 100 customers, ChurnAssassin would monitor those accounts for $200 a month. They currently have a handful of pilot customers and will be looking for additional pilot customers beginning in May.
The startup is currently working on refining the accuracy of its prediction models. Right now, ChurnAssassin’s accuracy is about 89%, but Polackoff hopes to increase this number to about 90-93% by the end of the summer. By the end of 2025, Polackoff aims to be earning about $100,000 in revenue.
