agentsclimarketplace

Churn signal identification

Skill Dragoon0x/product-skills/skills/retention-activation/churn-signal-identification

The product thinking layer for AI agents. 580 skills. 48 commands. 8 domains. - Product strategy, discovery, execution, analytics, growth, experience, conversion, communication and more.

Install
npx -y skills add Dragoon0x/product-skills --skill churn-signal-identification

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Identify behavioral, usage, and contextual signals that predict a customer is about to leave — before they cancel. Build an early warning system, not a postmortem. Use when churn is too high, building health scores, or designing proactive retention interventions.

SKILL.md

3.2 KB, as published. Nobody here has run it

Churn Signal Identification

Spot the warning signs before customers leave, not after.

How to use

  • /churn-signal-identification Apply churn signal constraints to this conversation.
  • /churn-signal-identification <product> Build a churn early warning system for the described product.

Constraints

The Churn Timeline

Churn has phases. Interventions MUST happen in the early invisible phase, not the decision phase.

  • Invisible phase (weeks/months before): subtle behavior changes, slight usage dips
  • Consideration phase (days/weeks before): evaluating alternatives, asking about exports
  • Decision phase (days before): customer has decided. Usually too late.
  • Action phase: cancel, don't renew, stop paying

Signal Categories

MUST monitor all six categories:

  1. Usage decay: login frequency drops, core action frequency drops, session duration decreases, feature breadth narrows
  2. Engagement decay: stopped opening emails, unsubscribed from updates, support tickets drop to zero (gave up, not satisfied)
  3. Team decay: active users on account decreasing, champion goes inactive, no new users added
  4. Integration decay: disconnected integrations, reduced API calls, stopped syncing
  5. Financial signals: downgraded plan, reduced seats, disputed invoice, asked about cancellation terms
  6. Competitive signals: asked about data export, mentioned competitor in support, viewed comparison pages

Signal Weighting

  • MUST assign high/medium/low weight to each signal based on historical correlation with churn
  • Champion going inactive = critical. Always investigate.
  • Support ticket about export = high. Often means they're evaluating alternatives.
  • Email open rate drop = low. Only meaningful as part of a pattern.
  • NEVER treat a single low-weight signal as actionable. Look for clusters.

Health Scoring

  • Green: active usage, growing engagement. No action needed.
  • Yellow: some decay signals. Trigger a check-in.
  • Orange: multiple decay signals, clear downward trend. Proactive intervention.
  • Red: major signals firing, usage near zero. Escalate to retention or senior contact.

Intervention Matching

  • Usage decay → re-engagement (personalized outreach, highlight unused features)
  • Champion inactive → relationship (reach out directly, connect with replacement)
  • Downgrade → value demonstration (ROI review, plan optimization)
  • Export/cancellation inquiry → retention (understand the concern, don't hard-sell)
  • Competitive signals → positioning (share differentiation, address specific gaps)
  • MUST match intervention to signal. Generic "we miss you" emails make things worse.

Anti-Patterns

  • Only measuring churn at cancellation — that's measuring failure, not preventing it
  • Treating all churn the same (free user vs. enterprise are different problems)
  • Bombarding disengaged users with emails and calls — this accelerates churn
  • Ignoring involuntary churn (failed payments) which is fixable with dunning

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.