Retention

Earn return visits by making lasting value easy to resume—so people come back because the product still fits their life, not because they were nudged.

Prerequisite: Customer Discovery — know who this is for and what state they arrive in. Related concepts: Feeling North Star, Tools, Techniques, and Practices.

What it is

Retention is whether people who already found value keep coming back to get more of it. It sits downstream of Activation—you cannot retain someone who never experienced the product working—and it is measured in cohorts and curves, not single snapshots: what share of a signup cohort is still active on day 1, 7, and 30, and does that curve flatten into a durable core or slide toward zero. A flattening curve means the product has found a group of people for whom it is genuinely useful; a curve that never flattens means either the wrong people are arriving, the value is one-time, or something about return visits is broken.

Retention is not the same job as engagement or habit formation, even though they share tools. Engagement is about how deeply someone uses the product within a session; Habit Formation is about whether return becomes automatic and context-cued. Retention sits between them: it asks whether the product still fits the person’s life well enough that they choose to resume it, deliberately, when the need recurs.

Why it works

People return to things that keep working for them—not to things that remind them the most often. The mechanism is resumption, not reminder: a returning user needs to find their progress, context, and prior work exactly as they left it, with the friction of re-entry as close to zero as the value allows. This is why Effort Moat, Value Replay, and Personalisation do more for retention than most notification strategies: they honour what the person already invested, rather than asking them to reconstruct it.

Retention curves are also the cleanest way to detect a product-market mismatch, because a curve does not lie the way a survey can. If even highly activated users churn steadily, the issue is rarely a UI polish problem—it is more often that the product’s natural usage interval was misjudged (forcing a daily frame on a weekly tool inflates apparent churn) or that the value simply does not recur for this segment. Treating retention as a design problem to be nudged, rather than a signal to be diagnosed, is how teams end up bolting streaks and notifications onto a product nobody actually needs again.

Core Tools, Techniques, and Practices (TTPs)

• Gamified Progress: Make meaningful progress visible, achievable, and fully in the user’s control.
• System Widget: Put one timely, glanceable piece of user-chosen value where people already look.
• Value Replay: Bring a user’s meaningful progress back into view, only where and when they have chosen to receive it.
• Effort Moat: Honour what people have built, so their progress remains useful and recognisable.
• Personalisation: Make the product fit the user’s stated needs—without making assumptions they cannot inspect or change.
• Variable Reward: Add optional novelty around reliable value, without making users chase uncertainty.
• Commitment: Let users declare their intent, then design the product around it.

Supporting Tools, Techniques, and Practices (TTPs)

• Success Moments: Mark meaningful progress with feedback that feels earned, then let users decide what comes next.
• Micro Interactions: Small responses that make an action feel acknowledged, understandable, and complete.
• Intent Mirroring: Notice a meaningful signal, then offer a small, optional next step that fits the user’s likely goal.
• Pattern Alignment: Use familiar mental models where they help people act with confidence.
• Deep-link: Drop users into the exact moment, feature, or context that matters.
• Graceful Exit: Make leaving as respectful as joining, so the last impression earns the return visit.

Insights & Metrics

Measure whether people return—and whether that return matches how often the product should naturally be used.

1. Day 1 / 7 / 30 cohort retention
   • Formula: (Users from a new-user cohort who return on day X / Cohort size) × 100
     (Prefer a clear starting event and return event; use “on or after” when you care about cumulative return.)
   • What it measures: How well a cohort sticks after first use
   • Why it matters: Early retention is the base of sustainable growth; compare cohorts, not vanity totals
2. Stickiness ratio (Daily Active Users / Monthly Active Users)
   • Formula: Daily Active Users (DAU) / Monthly Active Users (MAU)
   • What it measures: How concentrated activity is within the month
   • Why it matters: Higher ratios suggest frequent return—but judge against your product’s natural cadence (daily tools ≠ weekly tools)
3. Churn rate (period)
   • Formula: (Users who became inactive in the period / Active users at period start) × 100
   • What it measures: Loss of previously active users
   • Why it matters: Retention’s mirror metric—shows when value, fit, or trust is eroding

Behind the Data

Why people return, when they stop, and what still earns another visit.

Retention curve analysis

• Does your curve flatten (a core of lasting users) or keep sliding toward zero?
• Which acquisition or behaviour cohorts keep the strongest retention after the early drop?
• Are you measuring return on the product’s real usage interval—or forcing a daily frame that misreads health?

Feature usage and return value

• Which features predict long-term return versus one-session spikes?
• Does the order of feature adoption change retention strength?
• Where does resumed value feel hard to find—so people open the product and leave?

Return prompts without fatigue

• What legitimate reasons bring occasional users back (progress, people, deadlines, outcomes)?
• How do notifications and Value Replay affect return without training people to mute you?
• When users churn, what did they lose that Effort Moat or Personalisation could have preserved?

Further reading

• Interpret your retention analysis (Amplitude Docs) — How cohort retention methods and curves are defined in practice.
• Usage interval in retention analysis (Amplitude Docs) — Match retention goals to how often users naturally return.
• Fogg Behavior Model — Motivation, ability, and prompt must converge for a return behaviour to happen.
• Choose Customer Loyalty over Short-Term Profit (Nielsen Norman Group) — Why trust-preserving experience beats retention tricks that tax loyalty.
• Loyalty on the Web (Nielsen Norman Group) — Classic guidance on earning repeat visits without becoming an annoyance.

Agent skill

• Primary command / sequence: /productfeeling sequence — map → peaks-ends → states → trust → review
• Core TTPs to load: Gamified Progress, System Widget, Value Replay, Effort Moat, Personalisation, Variable Reward, Commitment
• Supporting TTPs: Success Moments, Micro Interactions, Intent Mirroring, Pattern Alignment, Deep-link, Graceful Exit
• When the agent should use this strategy: “users churn after week one”, “D7 falling”, “need return without spam”, “resume value”, “win-back flow”
• Companion handoff: Impeccable — return surfaces, widgets, progress UI, and re-entry states; DocSlime — when retention definitions, cohort metrics, or return policies belong in docs/strategy/; RedTeam — before streak mechanics, notification cadence changes, or variable-reward programmes ship (/redteam premortem on the proposed return loop)
• Feeling north star this strategy serves: trusted familiarity — progress still waiting for me
• Anti-goals: streak punishment, addictive variable-reward loops, unsolicited resurfacing, retention at the cost of agency
• Reference path: skill/reference/sequence.md