How many email variants should I use per campaign for A/B testing?

The recommended number of email variants per campaign step depends on your campaign size and testing goals.

Recommended Variants by Step

Initial email: 2-3 variants Follow-up emails: 1-2 variants each

Why These Numbers Work

More than 3 variants per step dilutes your data. You need enough sends per variant to draw meaningful conclusions about performance.

AI-generated campaigns can draft more variants than this. Treat those as a starting set to review, not a rule that every variant should stay active. Before launching, open Sequences and disable or delete any version you would not be happy sending.

For small campaigns, stick to 1-2 variants maximum per step so you do not split already-thin data too many ways. Treat early results as directional until the campaign has real learning volume.

For a quick learning test, use at least 1,000-1,500 leads. If you already know the niche is worth testing, 2,000-2,500 leads gives variants a better chance to reach meaningful volume. Larger campaigns can support 3 variants in the initial email.

What to Test in Your Variants

Focus your A/B tests on elements that impact performance:

• Subject lines have the biggest impact on open rates
• Opening lines have the biggest impact on reply rates
• Call-to-action variations (book a call vs. reply to discuss)

Keep the overall message similar between variants. Change one element at a time for clean testing results.

How to Create Variants in Campaigns

1. Navigate to your campaign and click the Sequences tab
2. Click Add Variant next to any step
3. The system generates a new variant you can edit
4. Use the Preview button to see how each variant looks with personalization

You can also use the Variations tool to generate fresh versions of existing scripts before adding them to your campaign.

Monitoring Variant Performance

Check variant performance in the Analytics tab:

1. Scroll past the metrics grid and the Campaign Performance Trend card — the per-step variant table sits at the bottom
2. Each step row breaks down performance per variant (labelled A, B, C)
3. Look for meaningful differences in reply rates once each variant has enough sends
4. Use the toggle in the row to disable an underperforming variant (disabling the only active variant in a step removes the whole step — you'll see a Disable the last variant? confirmation)

Related

• Campaigns
• Email Accounts
• Email Methods