A/B Test Idea Generator
Produce a focused test roadmap — ad copy, landing page, and offer — that actually moves performance
What This Prompt Does
Too many SEM tests fail because they test the wrong thing (button colors, filler copy) or never define a win condition. This prompt generates 5 ad copy tests, 3 landing page tests, and 3 offer/CTA tests — each with a clear purpose and a measurable success criterion, so you finish each test knowing whether it won.
When to Use It
- •Plateaued performance that needs structured experiments to unstick
- •Building a Q1/Q2/Q3 testing roadmap for an account
- •A new team member wants to start experimenting and needs direction
- •Pitching a test program to a skeptical stakeholder
The Prompt Template
Act as a paid search testing strategist. Your goal is to generate simple, useful A/B test ideas for this campaign so performance can improve through structured testing. Context: - Business type: [BUSINESS TYPE] - Campaign goal: [GOAL] - Current ad copy: [PASTE COPY] - Landing page summary: [PASTE SUMMARY] - Current performance issues: [PASTE ISSUES] Task: Suggest: 1. 5 ad copy tests 2. 3 landing page tests 3. 3 offer or CTA tests For each test, explain: - what to test - why it matters - what result would count as success Process: 1. Review the current messaging and performance issues. 2. Identify the most likely bottlenecks. 3. Suggest focused tests with a clear purpose. 4. Keep the tests manageable. Constraints: - Use plain English. - Do not suggest too many tests at once. - Focus on high-value, realistic tests. - Keep explanations short. - Favor recommendations that reduce wasted spend and improve conversion quality, not just click volume. Output format: Return a table with these columns: - Test - Type - Why - Success Measure
How the Prompt Is Structured
Three Types of Tests
Ad copy, landing page, and offer/CTA. Each test type lives in a different part of the funnel, and separating them ensures the plan covers all three rather than stacking 11 ad copy tests.
"Success Measure" Is Mandatory
Without naming a win condition, tests drift into "interesting" and never conclude. The output format requires a measurable success criterion for every test — no exceptions.
"Do Not Suggest Too Many Tests at Once"
Test programs die from overload. The 5-3-3 cap is generous enough to fill a quarter but constrained enough to stay manageable.
Tied to Current Issues
Asking for "Current performance issues" in the context means the tests target real bottlenecks, not theoretical ones. Tests without a diagnosis are noise.
Example Output
| Test | Type | Why | Success Measure |
|---|---|---|---|
| Add pricing language to one ad variant | Ad copy | May filter out poor-fit clicks and improve lead quality. | Lower CPA or higher conversion rate. |
| Replace "Call Now" with "Get Same-Day Help" | Ad copy | Lead with benefit instead of command. | Higher CTR, equal or better conversion rate. |
| Put phone number in hero instead of footer | Landing page | Mobile users don't scroll to find the CTA. | Higher tracked call volume on mobile traffic. |
| Trim lead form from 9 fields to 3 | Landing page | Fewer fields reduce abandonment on urgent-service traffic. | Higher form completion rate. |
| Add "Free estimate on major jobs" as offer | Offer | Removes commitment pressure for bigger-ticket intent. | Higher conversion rate in Remodeling campaign specifically. |
Tips for Better Results
One Test at a Time
Running three concurrent landing page tests makes it impossible to tell which change moved the number. Serialize.
Run Until Significance, Not a Calendar Date
Calling a test at an arbitrary date invites false positives. Use a simple significance calculator and let the math tell you when to stop.
Kill Tests That Won't Move the Number
If a test needs 8 weeks to hit significance on a tiny effect size, it's not worth running. Drop it and test something bigger.
Keep a Test Log
Document every test with hypothesis, result, and what you learned. Even losing tests produce insight if you write them down.
Test the Things That Actually Move the Number
We build testing roadmaps that turn paid search optimization from guesswork into a repeatable process.