Reforge — Retention Is the Output

Summary

Retention is the most important growth metric, but most teams treat it wrong -- they try to "improve retention" directly instead of recognizing it as an output of three inputs: activation, engagement, and resurrection.

The mental model: retention answers one question -- "How many users remained active within a defined time period after signing up?" But you can't optimize it directly. You optimize the inputs.

Three common mistakes when defining your retention metric:

1. Wrong frequency -- Borrowing your retention window from competitors or blog posts instead of understanding your product's natural usage cadence. Over-applying (measuring daily for a monthly product) or under-applying (measuring monthly for a daily product) sends you in the wrong direction.
2. Wrong core action -- Choosing the wrong behavior as your "active" signal leads the team to optimize for the wrong things. The core action should reflect genuine value delivery, not just any product interaction.
3. Wrong audience -- Not all users are equal. Some segments are your target use case; others aren't. Optimizing retention across ALL users dilutes the signal from the users who actually matter.

Relevance to projects:

• [[01-projects/squarely-puzzles/index]] -- What's the natural usage frequency for a puzzle game? Daily? Weekly? Defining this wrong will either make retention look artificially terrible (daily for a casual weekly player) or mask real problems (monthly for what should be a daily habit). The core action matters too -- is it "opened the app" or "completed a puzzle"?
• [[01-projects/data-marketplace/index]] -- Retention here is tricky because the frequency depends on the use case. A data consumer who buys one dataset and never returns isn't necessarily churned -- they might come back when they need different data. Need to define the right cadence and core action carefully.
• [[01-projects/newsletter/index]] -- Newsletter retention is relatively straightforward (opened an email within X weeks) but the "wrong audience" trap is real. Tracking retention across all subscribers including disengaged ones masks the health of the core audience.

Connects to [[06-reference/2026-04-03-activations-three-moments]] (activation as one of the three inputs to retention), [[06-reference/2026-04-03-reforge-defining-strategy]] (retention metric should be grounded in strategic understanding), and [[06-reference/2026-04-03-saas-metrics-that-matter]] (retention as the foundation of LTV calculations).

Open Questions

• For [[01-projects/squarely-puzzles/index]], what's the right retention frequency? Need to look at actual usage data before committing to a metric.
• The "resurrection" input is interesting -- what does a resurrection loop look like for each project? Email re-engagement? Push notifications? New content alerts?
• How do you separate "wrong audience" from "product isn't good enough"? If non-target users churn, is that signal or noise?