---
name: improve-a-conversion-funnel
category: web
description: Diagnose an ethical signup, purchase, application, or activation funnel and test one evidence-backed reduction in user friction. Use when people enter a web journey but fail to complete it, conversion rates fall, or a team wants growth without dark patterns or unsupported redesign.
---

# improve-a-conversion-funnel

Improve the user's ability to reach a wanted outcome. Start with journey evidence, locate one material
barrier, and test a reversible repair while protecting quality and consent.

## Inputs

- Name the audience, desired user outcome, funnel steps, business outcome, and decision owner.
- Gather event definitions, counts, recordings or feedback where permitted, support cases, and recent changes.
- Define guardrails for errors, refunds, complaints, accessibility, privacy, and downstream quality.

## Procedure

1. Verify the funnel events and eligible population by reconciling a sample from source records.
2. Calculate step conversion, absolute drop-off, time to advance, error rate, and meaningful segments.
3. Walk the journey on representative devices and states, including new, returning, error, and assisted users.
4. Combine quantitative drop-off with direct evidence from support, usability sessions, or user feedback.
5. List plausible barriers: unclear value, missing trust, unnecessary fields, surprise cost, technical error, slow load, inaccessible control, or wrong audience.
6. Rank barriers by impact, confidence, and effort. Select one with evidence and an owned repair.
7. Define the change, primary outcome, guardrails, minimum meaningful effect, audience, duration, and stop rule before launch.
8. Implement the smallest reversible change and verify events plus critical journey behavior.
9. Run the comparison using an appropriate controlled test or cautious before/after analysis with limitations stated.
10. Inspect guardrails, segments, downstream quality, and support effects before expanding.
11. Record the decision and update the funnel model with what was learned.

## Boundaries

Do not hide price, force consent, manufacture urgency, obstruct cancellation, or optimize accidental clicks.
Never treat more completion as success when refunds, harm, low-quality leads, or unwanted enrollment rise.
Protect small and sensitive segments.

## Done

- A reconciled funnel report identifies the highest-impact step and evidence for the selected barrier
- The change has a pre-written outcome, guardrails, audience, duration, and stop or rollback rule
- Instrumentation and critical journey checks pass before exposure
- A decision report shows outcome, uncertainty, guardrails, segments, downstream quality, and next action

Then use analyze-an-experiment for controlled results or accessibility-pass when interaction barriers dominate.