I have audited email programmes for dozens of DTC brands. The pattern is consistent. Welcome series built in 2024, never tested. Abandoned cart flow with a 10% discount baked in since launch, never challenged. Subject lines written by a copywriter on a Tuesday afternoon, treated as permanent. The flows are technically on. They are not optimised.
Your Klaviyo flows generate 41% of total email revenue from just 5.3% of sends. That ratio means small improvements compound at a rate that campaigns cannot match. A 15% lift in your abandoned cart open rate does not just improve one campaign. It works every single day, on every single abandoned cart, for as long as the flow runs. That is why systematic A/B testing is the highest-leverage activity most DTC brands are not doing.
Why Most DTC Brands Never Test Their Flows
There are three reasons brands do not test. The first is volume anxiety: they assume their list is too small to reach significance. The second is configuration confusion: Klaviyo has two separate A/B testing mechanisms and most operators do not know which to use for which goal. The third is priority blindness: testing feels slower than launching the next flow, so it keeps getting deprioritised.
All three are solvable. You do not need a huge list. You need patience with your highest-volume flows. The configuration is straightforward once you know the distinction between a flow email test and a flow branch test. And on the priority question, consider that brands running ongoing A/B tests see 47% higher email revenue per recipient and 34% better customer lifetime value than brands that do not. That is not marginal. That is a compounding advantage.
A flow that was good on launch day is not necessarily good today. Audience expectations change. Your product range changes. Seasonality shifts behaviour. The only way to know if your flows are still performing at their ceiling is to keep testing them.
Understanding Klaviyo's Two A/B Test Types
Before you build your first test, you need to know which tool to use. Klaviyo has two distinct A/B testing mechanisms and using the wrong one for the wrong goal wastes weeks of data.
Flow Email A/B Test
Best for
Testing copy, subject lines, creative, CTA wording, and offer within a single email
How it works
Splits recipients 50/50 between two versions of the same email at the same position in the flow. Klaviyo tracks opens, clicks, placed orders, and revenue per recipient.
Flow Branch A/B Test
Best for
Testing structural decisions: email count, sequence length, discount presence, and offer type across the entire flow path
How it works
Splits recipients between two entirely different paths through the flow. One path might send 3 emails, the other 5. One path might include a discount, the other might not.
Rule of thumb: if you are testing words and design, use a flow email test. If you are testing structure and strategy, use a flow branch test.
The 6 Tests to Run, In Order
Run these in the sequence below. Each test builds on the last. Start with the highest-volume flows (abandoned cart, welcome series) to get data fastest.
Subject Line Test: Personalisation vs Curiosity vs Urgency
What to test
Run three variants if volume allows; two if not. Test: (A) first-name personalisation ("[Name], you left something behind"), (B) curiosity gap ("We noticed something"), (C) urgency without discount ("Your cart expires tonight"). If limited to two, pit personalisation against urgency first.
Benchmark
Personalisation lifts open rates in 60% of tests. A real split test across 5,000 recipients comparing personalised vs generic subject lines showed 24.5% vs 27.8% open rate, a 13.5% lift at 98% confidence. Optimised subject lines add an average 23% to flow open rates.
Setup
In Klaviyo: open your flow, click the email you want to test, select "Create A/B test", enter your subject line variants, set split to 50/50. Let it run until each variant has at least 500 recipients. Klaviyo will flag the winner based on your chosen metric.
CTA Button Copy: Urgency vs Benefit vs Generic
What to test
Test three CTA variants: (A) urgency ("Complete your order before it sells out"), (B) benefit-led ("Get yours delivered by Thursday"), (C) generic ("Shop now"). Most brands are running variant C. That is leaving conversion rate on the table every day.
Benchmark
Urgency-based CTAs have shown up to 89% higher conversion rates versus generic "Shop Now" copy in tested DTC email campaigns. Even modest urgency framing ("Claim your order") consistently outperforms passive alternatives. This is consistently the highest-lift single change in a flow email test.
Setup
Build this as a flow email A/B test on the primary CTA button. Keep everything else identical. Measure on placed orders and revenue per recipient, not just click rate. A higher click rate that does not convert is not a win.
Discount Presence Test in Abandoned Cart
What to test
Most brands bake a discount into email 3 of their abandoned cart flow from day one and never question it. Test: (A) current sequence with the discount offer, (B) same sequence where email 3 is replaced with a strong social proof email and no discount. You are testing whether the discount is actually required to close the sale, or if your brand is strong enough to convert without it.
Benchmark
Many brands discover that 40 to 60% of their recovered carts would have converted without the discount. If your abandoned cart recovery rate does not significantly drop on the no-discount branch, you are giving away margin on customers who were already going to buy. Removing an unnecessary 10% discount at 200 recovered orders per month at a 60 GBP AOV saves 1,200 GBP in margin per month.
Setup
Use a flow branch A/B test. Build a duplicate of your abandoned cart flow without the discount in email 3. Split incoming traffic 50/50 between paths. Run for a minimum of 4 weeks before drawing any conclusions. Measure revenue per recipient across the full flow, not just the conversion rate on email 3.
Send Time Delay Between Flow Emails
What to test
The delay between emails in a flow is rarely tested, despite having a measurable impact on revenue. Test your abandoned cart email 1 delay: (A) 1 hour after abandonment, (B) 4 hours after abandonment. For post-purchase flows, test email 2 at 3 days versus 7 days after the first purchase. For welcome series, test email 2 at 24 hours versus 48 hours after signup.
Benchmark
Send time testing consistently shows that most brands send their second and third flow emails too late. Thursday evening sends show 56% higher engagement than Tuesday morning sends in direct campaign comparisons. In flows, tightening the delay between emails 1 and 2 in high-intent flows like abandoned cart typically improves revenue per recipient by 8 to 18%.
Setup
For delay testing in Klaviyo flows, you cannot use the standard A/B test button. Instead, use a flow branch test with identical email content but different time delay settings on each branch. This is more setup but gives you clean, attributable data.
Sequence Length Test: 3 Emails vs 5 Emails
What to test
Test whether adding emails to your flows actually adds revenue or just adds unsubscribes. Run your existing flow against an extended version with 2 additional emails. This is most valuable in your welcome series (where most brands are under-emailing) and your post-purchase flow (where most brands stop too early to capture the 60-day repurchase window).
Benchmark
Top-performing ecommerce brands run 12 to 16 active automated flows with longer sequence lengths. Welcome series benchmarks show top performers send 5 to 7 emails with a 7-day window, while average performers send 2 to 3. The gap in conversion rate between a 3-email and 5-email welcome series is typically 4 to 8 percentage points, representing significant first-order revenue.
Setup
Create a flow branch test in your welcome series or post-purchase flow. Branch A keeps your existing sequence. Branch B adds 2 emails: one value-led content email and one final social-proof-plus-soft-CTA email. Run for 6 weeks minimum. Measure on revenue per recipient across the full branch, not just email-level metrics.
Offer Type Test: Discount vs Bundle vs Gift
What to test
If you are already offering an incentive in your flows (welcome series, abandoned cart, win-back), test the type of offer, not just whether to offer one. Run: (A) percentage discount (10% off), (B) free gift with purchase above a threshold, (C) a bundle deal at a fixed price. The goal is to find the offer that drives the highest revenue per recipient while protecting margin.
Benchmark
Gift-with-purchase offers outperform equivalent percentage discounts in most A/B tests at the brand level. GWP offers also avoid training customers to expect price cuts, which erodes long-term margin. In practice, a free shipping upgrade (moving threshold from 40 GBP to 50 GBP) frequently outperforms a 10% discount in welcome flows because it is perceived as higher value while costing less.
Setup
Build three branches in your welcome series or win-back flow, each with the same email structure but a different offer in the final incentive email. Split traffic three ways (34/33/33). Run for a minimum of 6 weeks. Measure on revenue per recipient and contribution margin per converted customer, not just conversion rate.
How to Build a Testing Cadence That Compounds
Running one test is not a testing programme. Running tests in sequence, applying winners, and moving to the next variable is what builds compounding lift. Here is the operating cadence.
Start one test at a time
Running multiple tests simultaneously across the same flow muddies the data. You cannot isolate what caused the lift. Start with your highest-volume flow (usually abandoned cart) and run tests sequentially: subject line first, then CTA, then offer. Only once you have a winner on variable one should you move to variable two.
Set a minimum run window before reviewing
Do not review test results before 2 weeks have passed, regardless of early patterns. Early data is noisy. Set a calendar reminder for your minimum review date and resist checking before then. For lower-volume flows, extend the window to 4 to 6 weeks.
Declare winners on revenue per recipient, not open rate
Open rate and click rate are directional signals. Revenue per recipient is the metric that connects to your P&L. A variant with a 5% higher open rate but 12% lower revenue per recipient is not a winner. Always set your Klaviyo test winner metric to placed orders or revenue.
Document every test, result, and hypothesis
Keep a simple spreadsheet: flow name, test variable, variant A, variant B, run dates, sample sizes, winning variant, and lift percentage. This becomes your knowledge base. It tells you which principles apply across your audience and saves you from re-testing the same variable in a different flow.
Roll winners into campaigns
Every winning subject line formula, CTA structure, or offer framing discovered in a flow test should be applied to your campaign strategy. Flows generate the data. Campaigns amplify the learnings. A subject line formula that lifts open rates 15% in your abandoned cart flow will almost certainly do the same in your next sale campaign.
What This Looks Like at Scale
I worked with a wellness brand doing 80k GBP per month. Their Klaviyo account had been running for two years without a single A/B test. Flows were live, technically functioning, generating around 18% of total revenue. Not bad. But not anywhere near where it should have been for the size of their list.
We ran tests 1 through 3 over twelve weeks. Subject line personalisation in the abandoned cart series lifted open rates by 19%. A switch from generic to urgency CTA on the second cart email lifted placed orders by 31%. The discount test revealed that 52% of their recovered carts would have converted without the 10% incentive. Removing it from half the traffic increased contribution margin per recovered cart by 8.6%.
Net result after 90 days: email as a share of revenue moved from 18% to 27%. Not from adding a single new flow or growing the list. From testing what was already there.
Your list is already big enough to run these tests. Your flows are already generating enough data. The only thing missing is the decision to start testing and the discipline to run it systematically.
Common Mistakes to Avoid
Testing too many variables at once
One variable per test. If you change subject line and CTA at the same time and the test wins, you cannot tell which change caused the lift.
Measuring on opens instead of revenue
Set your winning metric to placed orders or revenue per recipient before the test goes live. Changing it after the test starts invalidates the result.
Declaring a winner too early
Set a minimum sample size (500 recipients per variant minimum) and a minimum time window (2 weeks) before reviewing. Early results are almost always misleading.
Not applying flow test winners to campaigns
Your flow tests are your most valuable creative research. Apply every winning formula to your campaign subject lines and CTAs immediately.
Running tests on inactive flows
Only test your highest-volume flows first. Testing a low-volume flow like a sunset or VIP flow means waiting months for statistical significance. Build your testing muscle on abandoned cart and welcome first.
Email Audit
Find Out What Your Klaviyo Flows Are Leaving on the Table
I will audit your Klaviyo account, identify the flows and variables worth testing first, and give you a prioritised testing plan with projected lift. No agency pitch. No retainer required to get started.
Book Your Email AuditFrequently asked questions
Can you A/B test Klaviyo flows?
Yes. Klaviyo lets you A/B test individual emails within a flow using the flow email A/B test feature. You can test subject lines, preheader text, email body content, CTAs, and send time delays. For testing entire branches (such as a 3-email sequence versus a 5-email sequence), use Klaviyo's flow branch A/B test feature, which splits recipients between two different paths.
How long should you run a Klaviyo A/B test before declaring a winner?
Run until you have at least 1,000 recipients per variant and 95% statistical confidence. For flow email tests, this depends on flow volume. High-volume flows like abandoned cart may reach significance within 2 weeks. Lower-volume flows like win-back can take 4 to 8 weeks. Never declare a winner based on fewer than 500 recipients per variant, regardless of how dramatic the early results look.
What should DTC brands A/B test first in Klaviyo?
Start with subject line tests in your highest-volume flows, usually your welcome series and abandoned cart flow. These generate the most data fastest and subject lines directly impact open rates, which gate all downstream revenue. Once you have subject line winners, move to CTA copy and offer type tests.
How much revenue can A/B testing add to Klaviyo email flows?
Analysis of 1,847 Klaviyo A/B tests across 89 DTC brands found systematic testing generated $3.2M in incremental email revenue, with brands running ongoing tests seeing 47% higher email revenue per recipient. Subject line personalisation commonly lifts open rates by 13 to 23%, and urgency CTAs lift conversion rates by up to 89% versus generic copy.
What is the difference between A/B testing a Klaviyo flow email versus A/B testing a flow branch?
A flow email A/B test splits recipients between two versions of a single email within the same flow path. It tests elements like subject line, content, or CTA. A flow branch A/B test splits recipients between two entirely different paths, letting you test different sequence structures, email counts, or different series of offers. Use email tests for copy and creative decisions. Use branch tests for structural decisions like sequence length and offer strategy.
Should DTC brands run A/B tests on campaigns or flows first?
Start with flows. Flows run continuously and accumulate data without requiring you to send extra campaigns. They also generate 41% of total email revenue from just 5.3% of sends, meaning small improvements compound over time. Once your flow tests are generating consistent learnings, apply the winners to your campaign subject lines and CTA copy.
About the author
Caner Veli built Liquiproof from zero to 3,000+ global retailers in under 6 years. He now helps DTC and CPG brands fix broken growth engines and scale 2x-15x in 90 days.