Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Perform cohort analysis on user engagement data — retention curves, feature adoption trends, and segment-level insights. Use when analyzing user retention by cohort, studying feature adoption over time, investigating churn patterns, or identifying engagement trends.
.claude/skills/phuryn-cohort-analysis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 595% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 63% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 37% | 0% |
Analyze user engagement and retention patterns by cohort to identify trends in user behavior, feature adoption, and long-term engagement. Combine quantitative insights with qualitative research recommendations.
Example 1: Upload CSV Data
Upload cohort_engagement.csv with columns: cohort_month, weeks_active,
user_id, feature_x_usage, engagement_score
Request: "Analyze retention patterns and identify why Q4 2025 cohorts
underperform compared to Q3"Example 2: Describe Data Format
"I have monthly user cohorts from Jan-Dec 2025. Each row shows:
cohort date, user ID, purchase frequency, and support tickets.
Analyze which cohorts show best long-term retention."Example 3: Feature Adoption Analysis
Upload feature_usage.xlsx with cohort adoption data.
Request: "Compare adoption curves for our new feature across cohorts.
Which cohorts adopted fastest? Any patterns?"You'll receive:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,119 | 20,474 | +235% | 1 | 1 | 0% | 219 | 5,212 | +2280% | 0 | 0 | — |
case-02 | pass→pass | 24,147 | 23,355 | -3% | 1 | 1 | 0% | 3,991 | 5,846 | +46% | 0 | 0 | — |
case-03 | pass→pass | 24,748 | 17,670 | -29% | 1 | 1 | 0% | 4,171 | 4,680 | +12% | 0 | 0 | — |
case-04 | pass→pass | 16,580 | 16,827 | +1% | 1 | 1 | 0% | 2,863 | 4,135 | +44% | 0 | 0 | — |
case-05 | fail→pass | 20,739 | 29,256 | +41% | 1 | 1 | 0% | 2,882 | 3,917 | +36% | 0 | 0 | — |
case-06 | fail→fail | 15,343 | 16,509 | +8% | 1 | 1 | 0% | 2,521 | 3,836 | +52% | 0 | 0 | — |
case-07 | fail→pass | 4,249 | 21,551 | +407% | 1 | 1 | 0% | 738 | 5,129 | +595% | 0 | 0 | — |
case-08 | fail→pass | 13,721 | 11,975 | -13% | 1 | 1 | 0% | 2,021 | 3,303 | +63% | 0 | 0 | — |
case-09 | pass→pass | 12,964 | 18,339 | +41% | 1 | 1 | 0% | 2,296 | 4,551 | +98% | 0 | 0 | — |
case-10 | fail→pass | 13,659 | 17,133 | +25% | 1 | 1 | 0% | 2,501 | 3,655 | +46% | 0 | 0 | — |
case-11 | fail→pass | 10,927 | 9,212 | -16% | 1 | 1 | 0% | 1,934 | 2,658 | +37% | 0 | 0 | — |
case-12 | pass→pass | 12,714 | 22,301 | +75% | 1 | 1 | 0% | 2,113 | 3,998 | +89% | 0 | 0 | — |
case-13 | pass→pass | 18,348 | 18,197 | -1% | 1 | 1 | 0% | 3,033 | 4,787 | +58% | 0 | 0 | — |
case-14 | pass→pass | 12,058 | 14,516 | +20% | 1 | 1 | 0% | 2,070 | 3,419 | +65% | 0 | 0 | — |
case-15 | pass→pass | 12,918 | 10,746 | -17% | 1 | 1 | 0% | 2,437 | 3,076 | +26% | 0 | 0 | — |
case-16 | pass→pass | 14,096 | 14,867 | +5% | 1 | 1 | 0% | 2,292 | 3,573 | +56% | 0 | 0 | — |
case-17 | pass→pass | 15,744 | 19,068 | +21% | 1 | 1 | 0% | 2,621 | 4,275 | +63% | 0 | 0 | — |
case-18 | pass→pass | 12,171 | 10,468 | -14% | 1 | 1 | 0% | 1,782 | 2,916 | +64% | 0 | 0 | — |
case-19 | pass→pass | 12,062 | 11,447 | -5% | 1 | 1 | 0% | 1,947 | 2,887 | +48% | 0 | 0 | — |
case-20 | fail→pass | 12,839 | 17,643 | +37% | 1 | 1 | 0% | 1,826 | 4,185 | +129% | 0 | 0 | — |
case-21 | fail→fail | 7,530 | 5,035 | -33% | 1 | 1 | 0% | 1,247 | 1,949 | +56% | 0 | 0 | — |
case-22 | pass→pass | 14,237 | 18,118 | +27% | 1 | 1 | 0% | 2,272 | 4,314 | +90% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 21 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.