What is CRO?

CRO stands for conversion rate optimization. It is the work of getting a larger share of the visitors you already have to buy — rather than paying for more visitors and converting the same share of them.

Your conversion rate is one division: orders ÷ visitors. A store with 25,000 visits a month, a 2% conversion rate and a $50 average order makes $25,000. Move that rate to 2.2% — a 10% relative improvement, not a dramatic one — and the same traffic makes $27,500. Nothing about the ad budget changed.

Why merchants reach for it

Paid traffic is rented. You stop paying and the visitors stop arriving, and the price of each one has been rising for years. A conversion improvement is different in kind: it applies to every visitor you already have — including the expensive ones — and it keeps applying after the work is finished. That is the whole argument for CRO, and it is a good one.

The argument has an honest limit, which most people selling CRO skip over. Most changes do not win. Run ten experiments and expect a couple of clear winners, a couple of clear losers, and a majority that move nothing you can distinguish from noise. CRO pays because you keep the winners and revert the rest cheaply — not because someone can pick the winner in advance.

How CRO is actually done

1. 1

   Find where the money leaks

   Analytics tells you which step loses people. Session recordings, reviews and support tickets tell you why. Neither one is enough alone: a number without a reason produces guesses, and a reason without a number produces work on a page nobody visits.

2. 2

   Write a hypothesis, not an opinion

   “Make the page nicer” cannot be tested. “Shipping cost is only revealed at checkout, so adding a free-shipping threshold near the buy button will raise add-to-cart” can be — it names the problem, the change, and the metric that would move.

3. 3

   Test it against the current page

   Half your visitors see the original, half see the change, at the same time, in the same conditions. Comparing this week against last week is not an experiment: it compares your change against the weather, the weekend and whatever your ads did.

4. 4

   Call it honestly, then keep the lesson

   Read a confidence interval rather than a live scoreboard, and do not stop the test the moment it looks good. Then write down what you learned — including from the losers, which are often more informative about your customers than the wins.

The part that decides whether you can test at all

An A/B test needs enough visitors to tell a real difference from a coin flip, and the number is larger than almost anyone expects. Detecting a 10% relative lift on a 2% conversion rate takes roughly 80,000 visitors in each arm at 95% confidence and 80% power — about 160,000 visitors in total. The store in the example above would need six months.

This is not a reason to give up; it is a reason to test bigger changes. A small tweak to a button colour is undetectable at any realistic sample size. A different offer, a restructured product page, a genuinely different value proposition — those produce effects large enough to see. Below roughly 10,000 monthly visitors, testing is usually the wrong tool entirely, and the right one is to apply well-evidenced fixes without proving each one.

You can put your own numbers through this with the CRO ROI calculator: it shows what a lift is worth on your store, and how long proving one would take.

What is worth testing first on a Shopify store

• The product page, above the fold

  What a shopper can see without scrolling decides most of it: the headline, the price framing, what the first image communicates, and whether the buy button is reachable on a phone. This is where the largest share of storefront experiments run, because it is where the largest share of the decision happens.

• The reasons not to buy

  Shipping cost discovered late, a returns policy nobody can find, no review anywhere near the price. These are not design problems — they are unanswered objections, and answering one where it is being asked is the most reliable win in ecommerce CRO.

• The mobile experience specifically

  Most Shopify traffic is on a phone and converts at a fraction of desktop. A test that improves desktop and quietly costs you mobile is a net loss you will report as a win unless you look at the split.

• The path, not just the page

  Collection → product → cart → checkout. A store that converts badly usually leaks at one specific step, and finding which one is cheaper than redesigning all four.

What CRO cannot fix

Optimization multiplies what you have. It does not create demand for a product nobody wants, rescue traffic from the wrong audience, or fix a price the market has rejected. If your ads bring the wrong people, a better product page converts the wrong people slightly better. Check that first — it is a cheaper answer than a testing programme.

There is also a hard line on a Shopify store between what a testing tool can change from the outside and what is genuinely development work. Any tool that blurs it is describing something it cannot do.

Where Looplift fits

Looplift is an AI CRO and A/B testing product for Shopify stores. It does the first two steps for you — it audits your live storefront, names specific conversion problems, and turns each into a ranked proposal with a hypothesis and a variant already built against your real page content — then runs the approved ones as experiments and calls the results with confidence intervals. You approve every launch; nothing goes live on its own, and nothing is written into your theme.

If you want the mechanics rather than the category, what is Looplift covers them, and Shopify A/B testing covers what it will and will not touch on a store.