What is conversion rate optimization

What conversion rate optimization means

Conversion rate optimization is the practice of raising the share of visitors who complete the action a site exists to produce, using controlled tests, not opinion. For a store, that action is an order.

Untested is not optimised.

The word optimization implies a comparison. A change never measured against the version it replaced has not been optimised, it has just been made. The comparison matters because most changes lose: published success rates put even the strongest programs between 8 and 33 percent.

How the number is calculated

Divide conversions by visits in the same period, then multiply by one hundred. A store with 50 orders from 1,000 sessions converts at 5 percent. The formula is the easy part. The word visits is where the trouble starts, because there are two published answers to it.

Two metrics, one name.

Google Analytics defines two separate metrics in its Data API schema, and both are called a rate of conversion. sessionKeyEventRate is the percentage of sessions in which any key event was triggered. userKeyEventRate is the percentage of users who triggered any key event.

The same denominator question follows a paid click. What you pay in a Google Ads auction is set per click, so whether that click paid for itself depends on what the order was worth, not on the rate alone.

This is why two dashboards disagree about the same week, and why a benchmark can flatter you without anyone lying. IRP Commerce states its own basis explicitly: transactions divided by sessions.

The two rates count different things on both sides of the line, so yours are not interchangeable. Work out both on your own data before you hold either against a benchmark, and check which basis that benchmark used.

Why the rate alone can mislead you

A comparable denominator tells you whether the rate moved. It does not tell you what the movement cost, and the cheapest way to move it is also the most expensive. A discount lowers what each order is worth, and it is usually run in the expectation that more people will buy. Judged on the rate it reads as a win. Judged on revenue it can be a loss.

Sector averages are not targets.

The discount case is arithmetic, not bad luck. A lower price raises the rate and lowers average order value at the same time, and revenue per visitor is the product of the two.

A benchmark can state its scale and still leave the question open. Littledata's ecommerce benchmark page puts the Shopify median at 1.4 percent across 2,800 sites, and gives the year. That page does not say whether the rate is per session or per user. Read any benchmark for that line before you compare yourself to it, because two people quoting the same number can mean different things.

The second way it misleads is comparison. In IRP Commerce's July 2026 data the highest sector converted at 5.23 percent and the lowest at 0.55 percent, a spread of 9.5 times inside one dataset in one month. An all-markets average of 2.26 percent sits between them and describes neither. Unbounce's landing-page benchmark shows the same shape with different numbers: a 7.1 percent median for food and beverage against 1.3 for fashion. Traffic source, business model and calculation method all differ between you and the average, so the comparison is structurally broken whatever the numbers. A benchmark describes the arithmetic in its own dataset, not a store.

Speed is the exception with a published number: the Core Web Vitals thresholds are written down, which is more than can be said for the rate itself.

What people say stops them is not subtle. It is also not a measurement of what happened. Read the next two figures as a shortlist to test, not a finding. In Bizrate Insights' 2026 survey of 1,010 US shoppers, 42 percent said unexpected fees make them abandon a cart often or very often. A 10 percent return fee put 72 percent of them off ordering from you at all.

Why CRO is not the same as SEO

Those are all things you can do to the store. The most common confusion is about which discipline owns them. SEO mostly changes how many people arrive and CRO mostly changes what happens once they do, and the two overlap wherever the landing page is the thing being ranked. They fail differently, so a single combined number hides which half is broken.

They fail in different places.

The other frequent mix up is conversion rate with click through rate, which Google Ads defines as clicks divided by impressions: it counts people who clicked an ad. High clicks with low conversions tells you the ad worked and something after it did not; which thing is what the next step has to find.

Where to read the full method

How the work is run, from sizing a test to knowing when a result can be trusted, is covered in our full guide to conversion rate optimization for ecommerce. The mechanics of a single experiment are in how an A/B test is run.

One rule from it belongs here: a test recorded only when it wins teaches nothing.

Before you compare your rate to anyone else’s, find out which of the two your dashboard is showing you. Most reports do not say, and the two answer different questions.

Once you know which of the two you are reading, the page changes that move it are our landing page optimization service.

An analytics report, two ratesWhy the same behaviour reports two different rates

Two rows that look interchangeable and answer different questions. Most reports show one of them and do not say which.

What counts as a conversion, according to the tool measuring it

1. 1Google puts the choice on you. Anything meaningful to the business can be a key event, and nothing on the list is a conversion by nature.
2. 2So two stores can report the same rate for completely different behaviour.