Purpose
Turn raw customer usage data and qualitative signals into a clear set of behavioral patterns — who does what, where they drop off, what separates retained users from churned ones — and translate those patterns into ranked product interventions a team can act on this quarter.
When to Use
Invoke this skill when the user asks to:
- Diagnose why users are churning or going inactive
- Identify the behaviors that predict activation, retention, or expansion
- Segment a user base by behavior rather than demographics
- Understand where users drop off in onboarding or a key workflow
- Turn analytics exports or session data into product decisions
- Make sense of mixed quantitative and qualitative signals (usage logs + survey responses + support tickets)
Inputs Required
Before starting, collect:
- Product type: Marketplace, B2B SaaS, B2C app, etc. — and the core value action (the thing users do that delivers value)
- Data available: What's been provided — event/usage logs, funnel metrics, retention cohorts, survey responses, support tickets, interview notes, or a description if no raw data
- Key metrics: Current activation rate, retention curve, churn rate, or whatever the team tracks (with numbers if available)
- The question driving this: What decision the analysis should inform (reduce churn, improve onboarding, find expansion triggers, etc.)
- User segments of interest: New vs. power users, free vs. paid, buyer vs. seller, or none specified
- Known hypotheses: Anything the team already suspects, so the analysis can confirm or disprove it
- Time window: The period the data covers, if relevant
If raw data isn't provided, work from the qualitative description and flag which patterns are inferred vs. observed.
Instructions
Step 1: Establish the Behavioral Baseline
Before finding patterns, define what "normal" and "successful" look like.
- Identify the core value action — the single event most tied to retention (e.g., for a marketplace: completing a transaction; for SaaS: inviting a teammate)
- Establish the activation moment — the point where a user has experienced enough value to likely stick (the "aha" event or threshold)
- Define the retained user profile — the behaviors that retained users share that churned users don't
- State assumptions explicitly when data is thin
Step 2: Behavioral Segmentation
Group users by what they do, not who they are. Build 4-6 behavioral segments:
- Power users: High-frequency, high-depth — using multiple features or completing the core action repeatedly
- Steady users: Consistent but shallow — getting value but not expanding usage
- Dabblers: Signed up, took a few actions, never activated
- Dormant/at-risk: Previously active, usage declining or stopped
- Single-session: Came once, never returned
For each segment: estimated % of base, defining behaviors, and the trajectory (growing, stable, shrinking).
Step 3: Funnel & Drop-Off Mapping
Trace the critical path from signup to value and beyond.
- Lay out the key stages (signup → setup → first value action → habit → expansion)
- Identify the single largest drop-off point and quantify it if data allows
- Distinguish leaks (users who try and fail) from no-shows (users who never attempt the step)
- Note any time-to-value signal — how long it takes successful users to reach activation
Step 4: Surface 8-12 Behavioral Patterns
This is the core deliverable. Each pattern must be structured as:
- Pattern name: A short, memorable label
- What's happening: The observed behavior, with a number or directional signal
- Why it likely matters: The mechanism or motivation behind it
- Evidence type: Observed (in data) / Inferred (from qualitative signal) / Hypothesis (needs validation)
- So what: The implication for the product or business
Cover patterns across at least three of these lenses:
- Activation patterns: What separates users who reach the aha moment from those who don't
- Retention/churn patterns: Leading indicators of churn that appear before the user leaves
- Engagement-depth patterns: Feature adoption sequences, habit formation, frequency cliffs
- Friction patterns: Where users repeatedly stall, abandon, or contact support
- Segment-specific patterns: Behaviors unique to a buyer/seller, free/paid, or persona split
Step 5: Rank Interventions by Expected Impact
Translate the top patterns into 6-8 product interventions, ranked. For each:
- The intervention: What to build, change, or test
- The pattern it addresses: Which finding it acts on
- Expected impact: High / Medium / Low — tied to how many users the pattern affects
- Effort: High / Medium / Low — rough build/operational cost
- How to validate: The metric that proves it worked
Rank by impact-to-effort ratio, not just raw impact.
Step 6: Define the Measurement Plan
Specify how the team will know patterns are shifting:
- The 2-3 leading indicators to instrument or watch
- The review cadence for re-running this analysis
- The single north-star behavioral metric the team should optimize against
Output Format
- Structure: Six sections following the steps; behavioral patterns and interventions presented as structured cards/blocks; segmentation and funnel as tables where data supports it
- Length: 1,100–1,500 words
- Format: Structured document — scannable blocks, not dense prose; tables for segments and funnel
- Sections: Behavioral baseline → Behavioral segmentation → Funnel & drop-off mapping → 8-12 patterns → Ranked interventions → Measurement plan
- Tone: Senior product analyst — analytical, hypothesis-aware, willing to label uncertainty, decisive on recommendations
Quality Checklist
Before delivering output, verify:
- [ ] Between 8 and 12 patterns are surfaced (count them)
- [ ] Every pattern is labeled Observed / Inferred / Hypothesis — no pattern is presented as fact when it's a guess
- [ ] The single largest drop-off point in the funnel is explicitly named
- [ ] Behavioral segments are defined by behavior, not demographics, with an estimated % of base each
- [ ] Interventions are ranked by impact-to-effort — not listed arbitrarily — and each has a validation metric
- [ ] A single north-star behavioral metric is named in the measurement plan
Notes
- When data is qualitative-only, the value is in structuring hypotheses for testing — be explicit that these are directional and propose how to instrument validation
- The most actionable patterns are usually leading churn indicators, not lagging ones — a usage decline three weeks before cancellation is more useful than the cancellation itself
- Resist over-segmenting: 4-6 behavioral segments are usable by a team; 15 are not
- If a known hypothesis is contradicted by the data, say so directly — disproving a wrong assumption is often the highest-value output
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills