North Metric organizes revenue work into three domains and ships ten AI agents that act within them. Five agents are currently live; five more are scheduled for release.

## What this page covers
The difference between the three revenue domains and the ten agents that work inside them, with each agent's scope and scoring logic.

## Domains vs agents — the model
A domain is a category of revenue work — a motion you can run on your customer base. North Metric ships three domains: Protect, Recover, Expand.

An agent is a specific scanner that detects opportunities within a domain. Each agent reads Stripe data nightly, scores accounts, and produces dollar-priced findings with named customers. North Metric ships ten agents — three in Protect, three in Recover, four in Expand.

Domains are concepts. Agents are the units of work. You don't run a domain; you run the agents inside it. The daily briefing aggregates findings from every agent and sorts them by dollar impact.

## The three domains at a glance
- Protect — stop losing customers you could save. 3 agents.
- Recover — recapture revenue that already slipped. 3 agents.
- Expand — find growth hiding in your customer base. 4 agents.

## Protect domain (3 agents) — stop losing customers you could save
The average SaaS at $25K MRR loses $2,400/mo to preventable churn. By the time it shows up in a dashboard, the customer is already gone. Protect agents monitor every account daily and flag risk before cancellation.

• Churn Radar (live): identifies at-risk customers and structural churn patterns. Scores accounts on health, payment history, and engagement signals over the last 90 days; compares cohort churn rate against the P50 benchmark for your MRR tier.
• Whale Alert (planned): flags concentration risk in your largest accounts when warning signals appear.
• Cohort Decay (planned): tracks retention decay patterns by acquisition cohort age.

## Recover domain (3 agents) — recapture revenue that already slipped
About 7% of subscription charges fail every month due to expired cards, insufficient funds, or bank declines. The retry window is roughly 72 hours.

• Payment Recovery (live): finds failed payments and expiring cards. Scores failures by failure code (insufficient funds, expired card, do_not_honor) to estimate recovery probability; buckets card expiries into ≤30-day and ≤60-day windows.
• Cash Collector (planned): automates outreach sequences for overdue invoices.
• Win-Back Engine (planned): scores recently-churned customers by re-engagement probability × former MRR.

## Expand domain (4 agents) — find growth hiding in your customer base
26% of customers on lower plans use features consistently above their tier. They're upgrade-ready but haven't been asked.

• Upgrade Finder (live): scores customers most likely to upgrade. Composite score out of 100: spending vs plan (40 pts), tenure (20 pts), plan gap (20 pts), payment health and price sensitivity (20 pts combined). Threshold is 60.
• Growth Pulse (live): diagnostic on your growth engine. Quick Ratio (new + expansion MRR ÷ churn + downgrade MRR), growth percentile, expansion trend, and the dominant bottleneck.
• Trial Closer (live): surfaces expiring free trials, the MRR at stake, and recommended conversion plays. Buckets trials into ≤7-day (urgent) and 8–30-day (watch).
• Pricing Power (planned): detects pricing room through ASP vs ARPA trends, coupon dependency, and feature adoption.

## How agents reach the daily briefing
Every finding from every agent — at-risk customers, recoverable failed charges, upgrade-ready accounts, expiring trials — is priced in dollars, attached to a customer name, and surfaced in The North. The briefing sorts by urgency × dollar impact.

## Tier availability
Pro includes one agent per domain: Churn Radar, Payment Recovery, Upgrade Finder. Scale includes all ten. See /pricing for full tier details. For the platform context, see /platform.