GPT Tasks/Analyze Funnel Drop-Off and Plan Fixes
Task library

Research · GPT Task

Analyze Funnel Drop-Off and Plan Fixes

Turn supplied funnel-stage definitions and metrics into a stage-by-stage drop-off analysis with data-quality checks, friction hypotheses, alternative explanations, prioritized experiments, and measurement criteria.

Context → structured prompt → explicit AI handoff

Runs in this browser. No upload by this page. External AI handoff is explicit. Review before sharing.

Prompt workbenchConfigure → review → use
01

Configure task

Setup
Describe the funnel, each stage in order, what counts as entering/completing a stage, and the business or user outcome the funnel should support.
Provide the stage-level counts, rates, segment/variant data, time period, definitions, and any qualitative evidence available.
Prompt style
Additional options
Add instrumentation changes, traffic/source changes, product changes, segment differences, known incidents, missing data, or testing constraints that affect interpretation.

What you’ll need

  • Funnel stages plus the stage-level counts, conversion rates, or other evidence you want analyzed
  • Funnel stages and goal
  • Stage metrics and evidence

What you’ll get

  • Funnel and metric integrity check
  • Stage-to-stage observation table
  • Largest drop-off/friction signals
  • Segment or variant differences where supplied
  • Hypotheses and alternative explanations
02

Prompt preview

Editable

Review or edit the generated prompt before you copy it or explicitly open an AI chat.

Export

ChatGPT opens with the current prompt prepared. Gemini copies the current prompt, then opens Gemini so you can paste it. Keep secrets and sensitive information out of URL-carried prompts.

Task referenceExamples, expected output, boundaries, and routing.

Examples

1. Signup onboarding funnel

Context: A product team supplies visit, signup, email-verification, profile-completion, and activation counts for the same period plus one instrumentation change.

Goal: Find the strongest observable drop-off and plan 30-day tests.

Useful output: Stage table, integrity warnings, observed losses, hypotheses, alternate explanations, tests, and decision criteria.

2. Lead funnel by source

Context: A B2B team supplies landing, form, qualified-lead, meeting, and opportunity counts split by two supplied channels.

Goal: Compare channel-stage friction without inventing lead intent.

Useful output: Channel comparison, stage observations, data gaps, hypotheses, test plan, and measurement rules.

3. Checkout conversion funnel

Context: An e-commerce team supplies cart, checkout, payment, and order counts plus a known checkout change.

Goal: Create an immediate diagnostic and fix/check plan.

Useful output: Metric integrity check, drop-off observations, alternative explanations, instrumentation questions, prioritized checks/tests, and next actions.

Expected output

  • Funnel and metric integrity check
  • Stage-to-stage observation table
  • Largest drop-off/friction signals
  • Segment or variant differences where supplied
  • Hypotheses and alternative explanations
  • Data-quality and instrumentation questions
  • Prioritized fixes or experiments
  • Measurement and decision criteria
  • Assumptions, limits, and next actions

Truth boundary

Use only supplied funnel definitions and metrics. Do not claim causation, tracking access or user motives without evidence. Label data gaps, alternative explanations and experiments explicitly.

Overlap guard: Owns multi-stage funnel path analysis and stage-to-stage friction from supplied metrics. Use Content Performance for content-level results, Email Campaign Review for one email campaign/segment/variant analysis, and Email Funnel for designing the messaging sequence before or apart from performance diagnosis.

Privacy and handoff

Runs in this browser. No upload by this page. Copy/download locally or explicitly open ChatGPT or Gemini. Source files are attached in the chosen AI service, not uploaded by Mycelgrid. Review the generated prompt and AI output before using or sharing it.