What Luria does
Luria runs split tests on your own pages, measures them against a held-out control, and moves traffic to the winner.
Luria is a split tester for your own pages. It writes small changes to your store, shows them to some of your visitors, keeps your original page in front of the rest, and reports which one made more money.
What you'll need
An installed site (the Shopify app embed, or the script tag anywhere else) and traffic. Nothing else. You do not write variants, pick selectors, or configure a stats engine.
What it does
Step 1: It reads your pages and builds the tests
The onboarding scan reads your homepage and top product page, and your seven answers set the goal (Sales, Booked calls, Leads / emails, App downloads). Luria turns that into tests. You never build one yourself.
Step 2: It patches text and style, nothing else
A variant is a small text or style patch: a headline, a call to action, an offer, social proof, imagery or body copy. Layout is off by default. Checkout legal text, privacy and terms pages, and prices without approval are always off and not negotiable.
Luria never edits your theme code. Patches are layered on at page load, capped at 250ms of anti-flicker, and any error means the page renders exactly as if Luria were not there.
Step 3: You approve first, unless you turn that off
On the Pages tab, "How Luria applies changes" has two settings. Approve first is the default: every change waits for your go-ahead. Autonomous lets Luria apply and promote winners on its own, and everything stays reversible either way.
There is no approve button on a specific pending change today. The status pill counts "{n} awaiting your approval" and Luria handles approvals with you over email at hello@luriart.com.
Step 4: It measures against a held-out control
Part of your traffic always keeps the original page. Both arms run at the same time on the same traffic, so the comparison is not a before-and-after guess. A visitor is assigned by a sticky hash and keeps the same arm on every visit.
Step 5: It moves traffic while the test is still running
As results come in, more traffic goes to the arm that is winning. The control never drops below 10% of traffic. A variant is promoted only above 95% probability of beating control with at least 200 sessions per arm, and is killed below 5%. Details: How a test is measured.
Step 6: It reports in the dashboard
Six tabs at luriart.com/app: Overview (conversion rate, lift versus your 4-week baseline, conversions and tracked revenue, active tests, a 12-week chart), My Site (every change with a status and a roll-back button), Customers (converted sessions), Split Tests (per-test bars, sessions, P(beat)), Pages (the rules above) and Plan & Billing.
What Luria is not
- Not ad tracking and not ad attribution. Luria measures conversions on your own site against its own control arm. It does not tell you which ad or campaign made a sale.
- No ad platform connections. There is no Meta, Google, TikTok, Klaviyo or Google Analytics integration.
- Not a session recorder. There are no heatmaps and no session replays.
- Not a page builder. Luria patches text and style on pages you already have.
There is also no per-source targeting, no team access (one user per site), no goal selector in the dashboard (the goal is set in onboarding), and no API keys screen.
Common misunderstandings
- "It changed my theme." It did not. Remove the app embed or the script tag and your site is byte for byte what it was.
- "I can't stop it." You can. Pausing stops serving within about 60 seconds, and switching the Shopify embed off does the same. Ask us at hello@luriart.com; it is not a dashboard toggle today.
- "The numbers should match Shopify." They will not. Luria counts attributed orders net of full refunds, tied to sessions that saw a test. See Numbers don't match Shopify.
- "Variants stopped." After 5 free attributed conversions with no plan active, variants pause and measurement keeps running. See Plans and metering.