What to expect

An honest account of how Luria results behave over the first months, including the parts that look like problems but aren't.

This page exists so nothing surprises you. Some of what's below will look like Luria isn't working when it's doing exactly what it should. Read it once now and you'll read your dashboard correctly later.

Everything here assumes roughly 3,000 or more sessions per month on the pages being tested. Lower traffic stretches every timeline.

Expect a visible first win inside the first week

On most stores with enough traffic, the first measurable improvement shows up during the first tests stage, day 3 to 10. It's usually a headline or offer framing change, and it's usually modest.

"Most" is doing real work in that sentence. If your traffic is thin, your conversions are few, or your page is already very well optimized, the first win takes longer. That's not a sign something's wrong; it's a sign Luria is waiting for enough evidence to be sure.

Expect the numbers to move around early

In the first couple of weeks, the lift number in your dashboard will jump. Up 12% one day, up 4% the next, maybe negative for an afternoon. Small samples are noisy, and early on every sample is small.

Luria doesn't call a winner until it can defend it. Until then, treat the early number as a preview, not a verdict. Lift and significance explains how to tell the difference in the dashboard.

What not to do: pause Luria because a Tuesday looked bad, or edit the page because a Thursday looked great. Both reset the evidence.

Expect at least one variant to lose

Some variants will convert worse than your original. You'll see them in the dashboard. This is the system working, not failing. Luria can only find what beats your page by also finding what doesn't, and losers are information.

Losing variants get retired automatically. Traffic shifts away from them as soon as Luria is confident they're worse, and you keep the winner. No visitor is stuck on a losing page longer than it takes to prove it's losing.

If you want to approve variants before they go live, that's an option. See Approvals.

Expect low-traffic stores to take longer

Confidence comes from conversions, not calendar days. A store doing 50 orders a month reaches a defensible result later than a store doing 500, because each stage of training needs a certain number of conversions to clear.

If that's you, the levers that help most are in Getting better results faster: concentrate traffic on fewer pages, track every conversion, and don't pause.

Expect lift to compound, not spike

Luria is not trying to give you a good week. It's trying to raise your floor and keep it raised.

On stores with enough traffic and a clean, consistent goal, the pattern we aim for is a permanent lift in the range of 5 to 15% that holds after the first 30 to 60 days, then continues to inch up as Luria personalizes, adjusts for season, and tests offers. Those conditions matter: enough traffic, a clean goal, no mid-training resets. We don't promise that range to any individual store, because no one can know your number in advance. What we can promise is that we won't show you a lift we can't defend.

Expect the lift to leave if Luria does

The variants that produce your lift are served by Luria. If you pause Luria, switch off the app embed, or uninstall, your site goes back to your original page within about 60 seconds, and the lift goes with it.

This isn't a lock-in tactic; it's how the product works. The winning copy is yours to read and reuse (see Your Brand Brain), and you can manually rebuild a winning page into your theme any time. But the ongoing testing, per-channel splits, and personalization only run while Luria is running.

What you should never expect

A guaranteed percentage. Anyone who promises one is guessing.
Results on day 1. Luria needs a baseline first.
A straight line. Early numbers wobble, then settle.
Zero losers. Losers are how winners get found.
Luria to fix a broken funnel. If checkout is slow or the offer is wrong, Luria will find the best framing for a page that still has a problem underneath it. Fix the problem; Luria will optimize around the fix.

If something really does look wrong

No variants after day 5 at normal traffic, lift that stays flat at zero for weeks, numbers that don't match Shopify at all: those are worth a look. Start with No variants serving and Numbers don't match Shopify, or contact support.