Catch the usage drop before the churn email

An early-warning system that turns a falling usage curve into a human check-in while the account can still be saved — instead of a churn post-mortem after the cancellation lands.

When to run it

Churn reviews keep finding the same story: the account's usage fell off a cliff weeks before the cancellation, and nobody was looking. By the time the cancellation email arrives, the customer has already re-solved the problem somewhere else — the save conversation you're having is the one that should have happened at the drop.

Why it works

At the moment usage drops, the customer still has the workflow, the data, and the muscle memory in your product — switching costs are still on your side. Two weeks later, none of that is true. An early-warning play converts churn from a verdict you receive into a signal you act on, and the outreach that works at this stage is a genuine check-in ("did something change on your side?"), not a discount.

The play

1. Define the drop upstream — a scheduled job or an analytics cohort that compares this week's core-action count to the trailing average — and turn it into a single "usage dropped" signal carrying the size of the drop.
2. Route big drops to a human: an internal alert with the account, the number, and the owner. Small drops get the automated check-in.
3. Send the check-in as plain text from a person, one question, no CTA button. It should read like an email a founder would actually write.
4. Give the account two weeks to recover before it counts as saved or lost.

Ship it with Hogsend

The detection lives in your analytics; the journey handles the response. Property conditions on the trigger split the human path from the automated one.

import { days, defineJourney, hours, sendEmail } from "@hogsend/engine";

export const usageDropWarning = defineJourney({
  meta: {
    id: "usage-drop-early-warning",
    trigger: {
      event: "usage.dropped",
      where: (b) => b.prop("dropPct").gte(50),
    },
    // A slow slide re-fires the detector; don't re-enter for a month.
    entryLimit: "once_per_period",
    entryPeriod: days(30),
    suppress: hours(0),
  },
  run: async (user, ctx) => {
    await sendEmail({
      to: user.email,
      userId: user.id,
      template: "retention/founder-checkin",
    });
    const recovered = await ctx.waitForEvent({
      event: "usage.recovered",
      timeout: days(14),
      label: "await-recovery",
    });
    if (recovered.timedOut) {
      // Escalate to a human while there's still an account to save.
      await ctx.trigger({ event: "account.at_risk", userId: user.id });
    }
  },
});

Routing the account.at_risk escalation into Slack or Discord is the lifecycle alerts recipe.

How you'll know

Recovery rate: accounts whose usage returns to baseline within 14 days of the drop signal. Over a quarter, compare churn among flagged-and-contacted accounts against what your churn reviews used to find — the play pays for itself with a small number of saves, because these are existing-revenue accounts.