Google Ads Experiments: Running Valid A/B Tests (2026 Guide)

"I think we should switch to Target CPA bidding." "I feel like Broad Match might work better now."

Opinions are expensive. Data is free. Google Ads Experiments (formerly Drafts & Experiments) is the only way to validate a major change without risking the performance of your entire campaign.

In this "Mega-Authority" guide, we cover:

1. The Methodology: Science applied to marketing.
2. The Setup: Creating a Cookie-Split test.
3. What to Test: Bidding, Match Types, Creative.
4. Interpreting Results: Statistical Significance (P-Value).

Part 1: The Financial Impact of Testing

Imagine you switch your main campaign from Manual CPC to Maximize Conversions. It tanks. CPA doubles. You lose $5,000 in a week. You panic and switch back.

Now imagine you tested it on 50% of traffic. It tanks. You lost $2,500. You stop the test. Or, it succeeds. It lowers CPA by 30%. You roll it out confidently.

Experiments are your "Sandbox." They protect your downside while unlocking upside.

Part 2: Theory - Cookie Split vs Search Split

Google uses a Cookie-Based Split.

• User A falls into the "Control" bucket. They always see the original campaign.
• User B falls into the "Experiment" bucket. They always see the test campaign.

This ensures data integrity. It prevents a user from seeing Ad A in the morning and Ad B in the evening, which would muddy the conversion attribution.

Part 3: Framework - The Testing Hierarchy

Don't test random things. Test high-impact levers.

Part 4: Execution - Setting Up a Test

Let's test Manual CPC vs Target CPA.

1. Campaigns → Experiments → All Experiments.
2. Click + -> Custom Experiment.
3. Base Campaign: Select your current "Search - Generic".
4. Suffix: - Experiment - tCPA.
5. Configuration:
   • Change Bidding Strategy to Target CPA.
   • Set Target (use your historical 30-day average).
6. Split: 50% (Recommended).
7. Schedule: Start Date (Tomorrow). End Date (None - manually end it).

Part 5: The "Don't Touch" Rule

Once an experiment is live, DO NOT TOUCH IT. Do not change the budget. Do not add keywords. Do not change ads.

If you change variables mid-test, you invalidate the results. The "Learning Phase" takes 7 days. The "Data Collection" phase takes 14-30 days. Patience is the skill here.

Part 6: Interpreting Results

After 30 days, check the dashboard. Google will show a "Confidence Interval" (Star icon or Blue/Grey text).

• Metric: Conv. / Cost (ROAS) or Cost / Conv. (CPA).
• Result: "Experiment outperformed Base by +15% (95% Confidence)." -> APPLY.
• Result: "No significant difference." -> END. (The change isn't worth it).
• Result: "Experiment underperformed." -> END. (Good thing you didn't switch!).

How to Apply: Click "Apply Experiment." You can choose to:

1. Update Original: Converts the base campaign to the new settings. (Preserves history).
2. Convert to New: Pauses base, creates new campaign. (Resets history - Avoid).

Part 7: Summary & Checklist

If you aren't testing, you aren't growing.

Your Action Plan:

1. Identify a campaign that has plateaued.
2. Hypothesize a change (e.g., "Broad Match with Smart Bidding will get more volume").
3. Launch a 50/50 experiment today.
4. Wait 4 weeks.

Be the scientist.

The "Big Three" Growth Tests

Most accounts only need to run three types of experiments — ever.

1. "Brain Transplant" — Test Manual CPC → tCPA. Your biggest potential efficiency gain. Set experiment to 50/50 and run for minimum 30 days or 50 conversions per arm.
2. "Net Cast" — Test Phrase Match → Broad Match on your top-performing keywords. Broad Match in 2024 is dramatically better than 2019. Don't assume it still fails.
3. "Landing Page Duel" — In experiment settings, use Find and Replace in the Final URL field to swap Page A for Page B across every ad. One change, account-wide, clean data.

Naming Convention That Keeps You Sane

Every experiment must follow: EXP - [Date] - [Hypothesis]

• EXP - Jan 2026 - Test tCPA vs Manual
• EXP - Feb 2026 - Broad Match on Brand Terms

Without this convention, 6 months later you won't know what "Experiment 3" was testing.

The "Testing Too Much" Failure Mode

The single most common experiment mistake: changing multiple variables at once.

You change the bid strategy AND the headlines AND the landing page simultaneously. CPAs improve 30%. Was it the bid strategy? The headlines? The landing page? You will never know.

One experiment = one variable. If you want to test three things, run three consecutive experiments.

"Apply" vs "Convert" — End-Game Decision

When an experiment wins, you have two options:

• Apply to Base Campaign — Merges the winning variant into the original. All history preserved. Recommended for bid strategy and match type tests.
• Convert to New Campaign — Creates a brand-new campaign from the winner. The original stays unchanged. Use this when you want to maintain the original as a control.

Default recommendation: Apply to Base Campaign unless you have a specific reason to preserve the original.