Flo

How can AI make A/B testing email designs in Klaviyo practical?

By the Flo team · Updated

Short answer

Create two complete versions of the email as separate templates, then use Klaviyo's A/B test on the campaign (or an A/B test in the flow email) to split your audience between them. Change one big idea at a time, such as the hero, the offer framing or the layout, and judge the winner on clicks or orders rather than opens.

Why subject line tests are not enough

Subject line tests are easy, so they are what most teams run. But opens are an unreliable signal since Apple Mail Privacy Protection, and the subject line is only the doorway. What the email looks like, and what it leads with, often decides whether anyone clicks.

How design tests work in Klaviyo

  1. Build version A and version B as two complete emails, each saved as a template.
  2. In the campaign, create an A/B test and assign a template to each variation.
  3. Choose the test size and how the winner is picked, then let Klaviyo send the winner to the rest, or split the whole audience.
  4. In flows, add an A/B test to the flow email and give each variation its own content.

What to test

Test big ideas, not button shades:

  • Hero: product shot versus lifestyle image.
  • Offer framing: percentage off versus a free gift.
  • Layout: one hero product versus a grid of best sellers.
  • Length: short and punchy versus a longer story.
  • Proof: reviews up top versus further down.

Change one idea per test, or you will not know what made the difference.

Picking the winner

Prefer clicks or placed orders over opens. Make sure each variation gets enough recipients to mean something; small lists may need the whole audience split rather than a test and winner send.

Learning across clients

Agencies have an advantage here: you see results across many brands. Keep a simple log of each test, what changed and what won, by brand and by type of email. Patterns that hold across several clients, such as product shots beating lifestyle images for replenishment emails, become your starting point for new clients. Patterns that only hold for one brand become that brand's rules.

The production problem

The reason teams avoid design tests is that every variant is a whole extra email to design, review and get approved. That is where an AI designer helps most. In Flo, you ask for an A/B version and Flo makes a second, complete email from the same brief with one idea changed, and both go through review together. Both can be exported to Klaviyo as templates for your test. See also AI-designed Black Friday series.

Questions agencies also ask

Can Klaviyo A/B test different email templates?
Yes. Campaign A/B tests can compare different content, so each variation can use a different template. Flow emails can be A/B tested as well.
How long should an email A/B test run?
Long enough for most clicks and orders to come in, often several hours for campaigns. For small lists, split the whole audience and compare results afterwards.
Should agencies charge extra for A/B testing?
Many include a testing plan in the retainer. If variants are cheap to produce, regular design tests are a strong reason clients stay.