Design Your Best Sticky Call To Action Experiment

When done right, past lifts have ranged:

High Power Estimate +0.3% 5% MDE · not stat-sig Highest Estimate +9.1% 10% MDE · stat-sig

Neg vs Pos Rate 26 tests 27% (7 down) 69% (18 up)

-75.1 -11.3 -5.7 -3.9 -3.9 -2.1 -0.2 0.0 +0.3 +0.4 +2.0 +2.3 +2.5 +4.0 +4.9 +7.9 +8.2 +8.8 +9.1 +9.6 +10.5 +12.6 +18.4 +19.1 +43.1 +70.6

Make sure it's adequately powered — we'll help with the MDE calculation.

We'll surface the best combination of variables out of 86,400 possible ones, supported by 26 experiments.

This experiment qualifies for a free review by Kohavi, Vermeer & Linowski through the Replication Project. Ron Kohavi

Scope

Where the test will run. Select one or more page types.

Homepage

The site's main landing page.

Product Page

Pages displaying a single product or item.

Checkout

Pages in the purchase or signup flow.

Shopping Cart

The cart or basket review page.

Listing Pages

Category, search, or collection pages.

Targeting notes optional

What to include or exclude; specific triggering conditions, a list of URLs, or a regex.

Metrics

Which metrics will you use to evaluate the experiment? Mark each as Primary, Secondary, or Guardrail.

Metric	Primary	Secondary	Guardrail	Not negative	Not positive	Non-Inferiority w/ Margin
Add to carts						%
Sales						%
Revenue						%
Signups						%
Leads						%
Clicks						%
Bounce rate						%
Time on page						%
Scroll depth						%
Page views						%
Engagement						%
Page load time						%
Error rate						%
Cart abandonment						%
Refund rate						%
Support tickets						%
Latency						%
						%

Ship Criteria

How will you judge whether the variant won?

Superiority

The variant must perform strictly better than the control.

Non-Inferiority

The variant must perform no worse than the control by a defined margin.

Stop Criteria

MDE, Test Duration and Power.

Sample Size

Stop once a target number of visitors has been reached.

Duration

Stop once a fixed period of time has elapsed.

Early stopping (optional) — peek-safe rules to call it before the planned stop.

Early Ship

Stop early to ship once the variant shows a decisive win.

Early Abort

Stop early to cut losses if the variant is clearly losing or futile.

Even weeks. Round duration to complete weeks to balance weekday effects.

Minimum Detectable Effect (MDE)

Relative % lift you need to reliably detect.

Test Duration (days)

How many days you plan to run the experiment.

Segments To Track

Which audience segments should this experiment include?

Page Length

Benefits most from a persistent sticky CTA

Number of Choices

More choices may increase value of a sticky anchor

Device

User Type

Page Type

Traffic Source

Scroll Depth

Cart Status

Price Range

Evidence Basis

What body of evidence to base the decision on.

Your own experiment

Decide based on the results of this experiment alone.

Meta Data

Combine this experiment with results from other similar tests.

Experiment Design

Choose the structure of your experiment.

Simple

Single A/B test with one variant.

Exploratory

Open-ended test to discover directional signals.

Exploratory + Confirmatory

Exploratory test followed by a confirmatory run.

Leap Compound

Multiple simultaneous changes in one variant.

Intensity

Same change tested at varying levels of strength.

Cascade

A sequence of treatments where each builds on the previous one.

AI Starter Prompt

Copy this prompt and paste it into your AI tool to generate the experiment code.

Import to Experimentation Tools

Export this configuration and open it in your preferred platform.

Clicking a tool copies a JSON config to your clipboard and opens the tool's site in a new tab. “Share link” copies a URL that reopens this page with your configuration.

Get Feedback

Share this experiment design to get expert input before you run it.

All feedback options are paid services. Discord access is included with a GoodUI subscription.

GoodUI

            Attempt to read property "title" on false

            site/templates/template.php:23