The AI That Sees Customers Leaving Before They Do

Every subscription business runs on a quiet leak. Customers sign up, use the product for a while, then drift. By the time a cancellation email lands, the decision was made weeks ago - in a series of small signals nobody was watching. Churned, an Amsterdam software company, built its entire business around catching those signals early enough to do something about them.

The pitch is unusually blunt for enterprise software: tell a company which subscribers are about to leave, tell them what to offer, and fire off the campaign without waiting for a human to notice. Churned calls itself a predictive AI platform for high-volume subscriptions. In practice, it is trying to replace the spreadsheet-and-gut-feel version of customer retention with something closer to a forecast.

01 / THE PROBLEMRetention as a guessing game

Most retention tools react. A customer's usage drops, a support ticket goes unanswered, a payment fails - and a rule somewhere eventually flags it. Rules are easy to reason about and slow to catch anything subtle. They also multiply: every edge case becomes another "if this, then that," until a Customer Success team is maintaining a brittle machine that still misses the customers who churn quietly, without complaint.

Churned's founders argue the whole approach is backwards. "Many existing approaches are either too slow, too expensive, or require extensive manual input," says co-founder and CEO Michiel Doornenbal. The company's bet is that churn is less a relationship mystery than a data problem - one that a model, trained on a company's own history, can learn to see coming.

02 / HOW IT WORKSPredict, recommend, act

The platform starts by doing the unglamorous work: pulling customer data out of the places it hides. CRM records, support tickets, product usage logs, billing history, marketing touchpoints - Churned unifies them into a single view of each customer's health. From there its models generate a score, surface the drivers behind it, and, crucially, suggest what to do next.

That last step is where Churned's Co-Pilot AI earns its name. Rather than handing a Customer Success manager a dashboard and wishing them luck, Co-Pilot picks the next-best-action for each subscriber and triggers the campaign through tools the business already uses. A retention offer, a reactivation nudge, an upsell prompt - chosen, timed, and sent. The human sets the strategy; the model handles the thousands of individual decisions underneath it.

03 / THE FOUNDERSA professor and his students

Churned started in 2020 with an unusual founding pedigree. Francisco Blasques, a data science professor, co-founded the company with Michiel Doornenbal and Maarten Doornenbal - Michiel having been his former master's student. It is the kind of origin that shows up in the product: churn modeling here is treated as a discipline with academic roots, not a marketing feature bolted onto a dashboard.

04 / WHO USES ITWherever subscriptions leak

Churned aims squarely at high-volume, business-to-consumer subscriptions - the businesses where a small percentage change in retention moves real money. That spans digital media and publishing, direct-to-consumer brands, subscription boxes, energy and utilities, telecom, and fitness clubs. It even reaches into the nonprofit world, where "churn" means a lapsed donor rather than a canceled plan.

05 / THE MARKETAgainst the incumbents

The category Churned is walking into is not empty. Customer Success platforms like Gainsight, ChurnZero, Catalyst and Totango have spent years building the rule-based systems Churned is positioning against, and plenty of larger companies keep in-house data teams to build churn models themselves. Churned's argument is one of speed and automation: get a predictive model running on your data in days rather than staffing a project for months, and let it act rather than just report.

06 / THE MONEYA EUR 2.5M vote of confidence

In 2024 Churned raised EUR 2.5 million - roughly $2.76 million in total funding - in a round led by Newion and Volta Ventures, with existing backer Goldfish participating. The stated plan is familiar for a company at this stage: sharpen the AI, and expand beyond the Netherlands into new markets. The business model underneath is straightforward B2B SaaS - subscription access to the platform, sold to the very subscription businesses it helps defend.

There is a neat symmetry to that. A company named Churned, selling software to stop churn, on a subscription. If its own retention numbers are good, they are also the product demo.