How awork automates churn prevention across 4,500 customers using AI scoring and zero manual monitoring

awork had ~4,500 paying customers, but roughly two-thirds fell into a low-touch segment (under €150 MRR) where no one was actively monitoring retention. The CS team focused on high-touch accounts while low-touch customers churned silently — no alerts, no signals, no intervention.

The data to prevent this existed, but it was scattered across four disconnected tools: Chargebee for billing, Mixpanel for product usage, Intercom for messaging, and HubSpot for CRM. No one was connecting billing patterns with usage behavior to spot customers dropping off.

Fabian, who leads retention at awork, knew the critical signal: when less than 30% of users in a workspace are active on a weekly basis, the account is at risk. But there was no scalable way to monitor this across thousands of accounts — let alone act on it automatically.

awork evaluated Churned.io and other retention tools. None delivered what they needed: AI-powered scoring that pushes actionable data back into the tools where their team already works— not another dashboard to check.

After connecting Chargebee billing data, Mixpanel product usage, and HubSpot CRM, Customerscore.io matched billing identities with real usage behavior and began scoring every customer by churn risk. Within weeks, awork had a retention system that transformed how they manage their entire customer base.

Key features:

AI Churn Risk Scoring

The AI model scores each of awork's ~4,500 active customers on a 1–5 scale, with explainable churn drivers and churn brakes for every account. It analyzes hundreds of features — admin vs. IC activity ratios, weekly engagement trends, billing patterns, team size, and feature adoption. Combining traditional ML with AI-powered reasoning significantly improved scoring precision: a workspace with low activity but 25 consecutive renewals now gets a different risk assessment than one with the same activity pattern but only 2 months of history.

Two-Way Data Sync

Customerscore.io pushes health scores and AI churn risk directly into HubSpot, Intercom, and Mixpanel via 2-way sync. 4,500+ customers are synced with live health data across all three tools. The CS team never needs to open another tool — scores, trends, and risk signals appear where they already work.

Segmented Retention Strategy

Based on AI scoring, awork built a two-tier retention approach. For the low-touch segment (under €150 MRR), Fabian runs automated Intercom campaigns triggered by churn risk — reaching at-risk customers with re-engagement sequences at scale, without manual effort. For the high-touch segment (over €150 MRR), the CS team uses HubSpot to identify and personally reach out to high-risk accounts flagged by the AI.

Upsell Scoring (Coming Next)

The next step is expanding from churn prevention to revenue expansion. We're working on an AI Expansion Score that will identify which low-touch customers are ready for an upgrade — and when to approach them — based on usage patterns invisible in billing data alone.