See why your emails moved revenue. Connect Klaviyo and combine the campaign A/B tests you've already run. Each new test builds on the last.
Read-only access. 60 days free, no card. See pricing
Across 64 campaign tests, discount framing moves revenue per recipient more than anything else in the email.
Percent-off beats dollar-off for first orders
First-order revenue per recipient, 95% CI
+3% to +8%
This week
The 'Recent buyers: care guide' test strengthened the subject-line lesson.
Urgent subject lines beat curious ones for recent buyers
Revenue per recipient, 95% CI
+2% to +6%
Revenue per recipient, 95% CI
Percent-off
+1% to +4%
Dollar-off
−4% to −1%
Urgency
+1% to +4%
Scarcity
−1% to +3%
Social proof
−2% to +2%
Curiosity
−3% to +1%
Short
−1% to +3%
Long
−3% to +1%
Lifestyle
−5% to +5%
Product-only
−5% to +5%
Revenue per recipient, compared with your average
2 more features: button verb, body copy length
Across 64 campaign tests, discount framing moves revenue per recipient more than anything else in the email.
Percent-off beats dollar-off for first orders
First-order revenue per recipient, 95% CI
+3% to +8%
This week
The 'Recent buyers: care guide' test strengthened the subject-line lesson.
Urgent subject lines beat curious ones for recent buyers
Revenue per recipient, 95% CI
+2% to +6%
Revenue per recipient, 95% CI
Percent-off
+1% to +4%
Dollar-off
−4% to −1%
Urgency
+1% to +4%
Scarcity
−1% to +3%
Social proof
−2% to +2%
Curiosity
−3% to +1%
Revenue per recipient, compared with your average
4 more features: preview text length, hero image style, button verb, body copy length
The overview combines every past test: the lead lesson first, then each feature of the email with its range.
Results sit inside each campaign, so comparing tests means opening each one and copying numbers into a sheet. The sheet records which variant performed better, not why. Ask what the past six months of tests taught you about hero images, and you face an afternoon of moving data around before you can even start to combine them.
You can reference results in the Experiments tab, but we recommend maintaining a centralized spreadsheet.
Klaviyo, on its blog
Klaviyo's advice is right. We keep that record for you, and turn it into what your tests have taught.
Join the waitlistRead-only. Your first findings come from the tests you've already run.
We read what each variant says and shows: the urgency in a subject line, how an offer is framed, the style of a hero image. Each feature is a hypothesis about your customers, and every test checks it. Below are five hypotheses checked in one test. Your next test can build on them to narrow in on what moves revenue most.
Revenue per recipient, 95% CI
Subject line, urgency: +2% to +6%
Preview text, one product only: −8% to +6%
Hero image, lifestyle: −7% to +2%
Button, "Shop" verb: −1% to +5%
Body copy, short: −9% to +8%
Where the dots pile up, the values are more likely. A range that crosses zero is not a failed test.
Each quarter's findings rest on the last. The next test aims at what is still uncertain instead of settling a question twice. The result is a floor under every send that keeps rising.
Tests don't get faster. Late in a program they get harder to call, because the easy wins are already found. What compounds is what you know, not how quickly you know it.
Revenue per recipient, 95% CI
Q4 2026
Urgency for VIPs: −9% to +11%
Q3 2026
Lifestyle hero images for new arrivals: −8% to +9%
Q2 2025
Percent-off beats dollar-off for first orders: first-order revenue per recipient +3% to +8%
Q1 2025
Short preview text beats long for sale announcements: +1% to +4%
Each quarter's findings rest on the last. Nothing here is re-tested from scratch.
Q1 2025, then Q2 2026 to Q4 2026. The 4 quarters between are not shown.
The floor rises each quarter. Individual tests get harder to call, not faster.
Whichever metric a test was judged on, we read the others as well. In this example, the test picked the variant with more clicks. The revenue didn't follow. We show you the measures side by side, so you can choose revenue when it matters.
Judged in Klaviyo on: clicks
−1% to +2%
+4% to +9%
−6% to +5%
−10% to +10%
Judged in Klaviyo on: clicks
−1% to +2%
+4% to +9%
−6% to +5%
−10% to +10%
More clicks likely did not bring more revenue.
One document your team reads and your AI writes with. You write the email; we never do.
What your tests have taught is one document, in plain sentences, each with its range. You forward it to your CMO, hand it to your agency, or paste it into your own AI tool beside a brief. That way, your next email starts from what you already know, whoever writes it. The document comes as Markdown, a PDF, or a link.
In the CMO review, you are the one who can say what your customers respond to.
Klaviyo still splits and sends your test. What changes is what you learn from it.
| A spreadsheet | Klaviyo's A/B test on its own | Klaviyo's A/B test, read by Deciding Data | |
|---|---|---|---|
| Splits and sends the test | No | Yes[1] | No. Klaviyo does, and we leave it there |
| Every test in one place | If someone keeps it up | Each test in its own campaign[1] | Yes, kept for you |
| What about the email moved the result | Only if someone works it out by hand | Which variant won its metric[1] | Each feature, with its range |
| What earlier tests carry into the next | Whatever the keeper remembers | Each test starts fresh[1] | Every finding, every quarter |
| How strong the evidence is | Rarely recorded | A win confidence percentage[2] | A 95% range for every result |
[1] Klaviyo's guide to A/B testing email campaigns: help.klaviyo.com/hc/en-us/articles/115005228148
[2] Klaviyo's deliverability reference, on the win confidence percentage and the winning metric: help.klaviyo.com/hc/en-us/articles/25620771311643
Connect Klaviyo. You sign in through Klaviyo and approve read-only access to campaigns, metrics, and templates.
We read the tests you've already run. We read their variants, what each one said, and how each one did.
Read what you've learned. Your first document is built from your own history.
Your team keeps writing and sending in Klaviyo.
Campaigns, read only. Which tests you ran and what each variant said.
Metrics, read only. How each variant did.
Templates, read only. What each email looked like.
Nothing else. We never write to your Klaviyo account.
Join the waitlistFree for 60 days. No card.
With read-only access, Deciding Data informs your team and your tools.
Agencies already promise their brands compounding results. The document makes those results visible for each client, quarter by quarter, in the client's own numbers. Each client connects its own Klaviyo account, and one agreement covers every client account, so onboarding a new brand is one connection, not a new toolchain.
Talk to us about agency pricingA lifecycle analyst who could do this is a $100K hire. The tiers below differ by the size of your list, and the price is the only thing that changes between them.
60 days free on every tier.
No card to start, and we never charge automatically.
Nothing to cancel. If you don't choose a plan, the account turns read-only on day 60.
Customers who join before January 1, 2027 keep their founding price through 2028.
No. We ask for three read-only permissions, for campaigns, metrics, and templates, and no others.
No. You keep building and sending tests in Klaviyo exactly as you do now.
No. Your team, your agency, or your own AI tools write them, and the document helps them write from what your tests have shown.
Email campaign A/B tests. Flows and SMS are on the roadmap, and we will say so here when they arrive.
The product works best from about 30,000 active profiles. Below about 10,000, tests teach slowly, and it is better to know that before a trial than to find out in one.
We use Bayesian models that carry what each test learned into the next one, and every result is a range. There are no p-values and no "significant" stamps.
Nothing is charged automatically, and there is nothing to cancel. If you don't choose a plan, the account turns read-only, and the data is kept for 90 days and then deleted.