Test anything your visitors see
Every experiment splits traffic between the experience you have and the one you suspect is better — on the pages where money is made.
Page templates
Duplicate your product, collection or home template, redesign the copy in the theme editor, and split traffic between the two. No code, no risk to the original.
Copywriting
Test titles, descriptions and the metafield copy your theme renders. When a version wins, Varyant writes it back to the product for you.
Content blocks
Swap a section of the page — banner, offer, guarantee, social proof — and find out which one actually moves orders, not opinions.
Three steps to a proven winner
Create & split
Pick what to test, set the traffic split with a slider, choose the goal — checkout, add to cart, or order value. Target by device, traffic source, or new vs returning visitors.
Visitors see one store
Each visitor is assigned once and keeps their version for the whole test. The decision runs before first paint — nobody ever sees the page flicker or swap.
Apply the winner
When the math clears the guardrails, Varyant says so in plain words. One click ships the winning experience to 100% of visitors — and cleans up after itself.
Numbers you can act on
Most tools hand you a p-value and wish you luck. Varyant runs Bayesian statistics and answers the only question that matters: which version makes more money, and how sure are we?
P(best) — probability to win
The chance each variant is truly the best, updated live. No stats degree required.
Risk — what shipping early could cost
Expected loss in conversion points if you call it now and you're wrong. Ship when it's near zero.
Guardrails against false winners
Minimum samples, minimum runtime, and a traffic-split integrity check — the peeking traps are handled for you.
The ranges still overlap — keep the test running. Drag the slider to add visitors.
Your store stays fast. That's the point.
A/B apps are notorious for heavy scripts that swap content after the page loads — visitors see the flicker, and Google sees the slowdown. Varyant decides inline, before the first paint.
All variants render server-side in your theme; a tiny inline script reveals the assigned one before anything paints. No external calls in the critical path, no layout shift, no cumulative-layout penalty.
The right test, for the right visitors
Run a mobile-only layout test. Try a bolder pitch on paid traffic. Show returning customers something new. Everyone else simply sees your store.
Built different from typical A/B apps
| Varyant | Typical A/B apps | |
|---|---|---|
| Storefront flicker | ✓None — decided before first paint | ✕Content often swaps after load |
| Decision math | ✓Bayesian P(best), risk & guardrails | ✕p-values that invite calling it early |
| Revenue truth | ✓Real purchase revenue on every test | ✕Clicks standing in for money |
| Weight on your store | ✓≈2 KB inline decision, zero blocking scripts | ✕Heavy tags loaded in <head> |
| Your data | ✓Full raw export — daily, orders, segments | ✕Locked dashboards, sampled numbers |
Fair questions
Will it slow down my store?
No. The variant decision is a ~2 KB inline script that runs before the page paints — there are no external calls in the critical path and no content swapping after load. Tracking happens in the background via Shopify's own web pixel sandbox.
How does Varyant decide the winner?
A Bayesian engine computes each variant's probability of being best, the expected loss ("risk") of shipping it, and a 95% credible interval — with guardrails on sample size, runtime and traffic-split integrity so a lucky streak can't be crowned a winner.
Do I need a developer?
No. You create tests, edit variants in Shopify's own theme editor, and apply winners — all from the admin. Nothing to code, and nothing left behind in your theme when a test is deleted.
What exactly can I test?
Page templates (product, collection, home, pages), copywriting — titles, descriptions and metafield copy — and individual content blocks. Tests can target device, traffic source, specific referrer domains, country, and new vs returning visitors.
Does it work with my theme?
Varyant works with any Online Store 2.0 theme using standard app blocks and app embeds — no theme code edits. Template tests duplicate your existing template, so the original is never touched.
When can I get it?
Varyant is in private early access while we onboard the first stores personally. Join the waitlist and you'll hear from a human — not a drip campaign.
Stop guessing what sells. Prove it.
Join the waitlist and be one of the first stores running proven, flicker-free experiments with Varyant.