Conversion variants with sample sizes, lift and Wilson intervals — working example

Compare conversion variants with sample sizes, percentage-point lift and honest uncertainty intervals. A conversion comparison, sample-aware confidence intervals, explicit baseline and readable metric table. PopNow can add optional media after a quote; the core workflow runs locally.

Uses PopNow for
Images
Needs
Node.js

The skill is free. Optional AI images, video and sound are quoted for your approval before generation. API-key calls use your separate API balance in USD; generation through ChatGPT or Claude connected via OAuth/MCP uses your website credits.

v1.0.0 · Updated Oct 24, 2026 · 40 KB · Use it for personal and client work; please don't resell or share the package.

What you get

  1. A conversion comparison, sample-aware confidence intervals, explicit baseline and readable metric table.

  2. A validated local data model and self-contained HTML builder

  3. Representative data, editable source and meaningful regression checks

  4. Optional PopNow media with approval before generation

Good for

  • Compare control 10000 visitors / 620 conversions with new checkout 9800 / 701; show uncertainty without declaring a winner.
  • Teams that need an editable local artifact with traceable inputs

Not for

  • Not a sequential-testing or causal-inference tool; repeated users, peeking and nonrandom assignment need separate treatment.
  • A hosted backend or synchronized multi-user service

How to use it

You say

  • Use the PopNow skill "Experiment Comparison Chart" (experiment-comparison-chart) for this: Compare control 10000 visitors / 620 conversions with new checkout 9800 / 701; show uncertainty without declaring a winner.

Then try

  • Apply our actual data and explain any rejected fields.

  • Adapt the layout to our brand and quote optional PopNow media first.

What to have ready

  • CSV or JSON variants with variant,visitors,conversions. One independent user observation per experiment arm.
  • Your final audience, labels and publication destination
  • Owned assets and verified claims for any public version

Similar skills

View all →

Ratings

No ratings yet. Installed it? Be the first to rate it.

Rate this skill

Questions

What input does this workflow need?

CSV or JSON variants with variant,visitors,conversions. One independent user observation per experiment arm.

What does it calculate or validate?

Validate arm counts, compute observed conversion and Wilson 95% intervals, then show absolute percentage-point and relative lift against the first arm. Do not equate intervals with a significance test.

What is outside its scope?

Not a sequential-testing or causal-inference tool; repeated users, peeking and nonrandom assignment need separate treatment.

Can I use it without an API key?

Yes. The local template, validation and exports work without a key. PopNow media is optional and requires a quoted, approved API USD budget.