ai-analyst-lab/ai-analyst-plugin/drop-off-format
Claude Code and agent skill by ai-analyst-lab/ai-analyst-plugin
- community
Checking that the SKILL.md file is still in the repository.
Every step-to-step drop-off must show three numbers: raw users entering the step, raw users exiting, and the drop-off percentage. Trigger whenever output contains funnel steps, conversion rates between stages, or step-to-step comparisons, in tables, narrative, chart annotations, or Slack posts. A drop-off shown as only a raw count or only a percentage is incomplete.
Install
npx vskill@latest install ai-analyst-lab/ai-analyst-plugin/drop-off-formatSKILL.md
Skill: Drop-Off Format
Purpose
A percentage without a count hides scale. A count without a percentage hides severity. "We lost 12,000 users" sounds bad until you learn that's 3% of the base. "34% drop-off" sounds bad until you learn that's 850 users out of 2,500. This skill enforces one rule: every drop-off shows three numbers.
When to Use
Before presenting ANY step-to-step drop-off — in a funnel table, in narrative text, in a chart annotation, in a Slack message, or on a slide. This runs on every output that contains funnel or conversion-step data.
The Rule
**Every drop-off must show: entering count, exiting count, and drop-offpercentage.**
All three, every time. The format:
In a table, this becomes three columns:
| Step | Users | Drop-off | |
|---|---|---|---|
| checkout_started | 43,490 | -- | |
| payment_attempted | 35,543 | 18.3% (7,947 lost) |
Either format is acceptable. What is not acceptable is showing only one or two of the three numbers.
Instructions
Step 1: Find every drop-off in your draft output
Scan the response you are about to send. Every place where two adjacent funnel steps are compared — in a table row, a sentence, a chart label — is in scope.
Step 2: Check for all three numbers
For each drop-off, confirm you have:
- Entering count — unique users at the upstream step
- Exiting count — unique users at the downstream step
- Drop-off percentage —
(entering - exiting) / entering x 100
If any of the three is missing, add it before sending.
Step 3: Verify the arithmetic
The percentage must equal (entering - exiting) / entering x 100, rounded to one decimal place. Do the division — do not estimate. A wrong percentage is worse than a missing one.
Step 4: Apply to the "biggest leak" callout too
When the readout names the biggest drop-off, it must follow the same rule. "The biggest leak is checkout_started -> payment_attempted" is incomplete. "The biggest leak is checkout_started -> payment_attempted: 43,490 -> 35,543, an 18.3% drop-off" is complete.
Examples
Bad -> Good
| Bad | Good | |
|---|---|---|
| "34% drop-off at checkout" | "checkout_started (43,490) -> payment_attempted (35,543): 18.3% drop-off" | |
| "We lost 7,947 users at payment" | "payment_attempted lost 7,947 of 43,490 users (18.3%) from checkout_started" | |
| "The biggest leak is 47%" | "The biggest leak is product_view -> add_to_cart: 49,899 -> 48,083, a 3.6% drop-off" |
Anti-Patterns
- Never show only a percentage. The reader cannot tell if 34% means 340
users or 340,000.
- Never show only a raw count. The reader cannot tell if 7,947 lost is
catastrophic or routine without knowing the base.
- Never round the percentage to a whole number. Use one decimal place.
The difference between 18% and 18.3% matters when comparing steps.
- Never skip the rule for the "biggest leak" callout. That line is the
most-read sentence in the readout — it needs all three numbers more than any other.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingai-analyst-plus/skills/drop-off-format/SKILL.md
Popularity
- Repository stars
- 32
- Repository forks
- 13
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ai-analyst-lab/ai-analyst-plugin/drop-off-format)