CRO: conversion rate optimisation

What is CRO (Conversion Rate Optimization)?

CRO stands for Conversion Rate Optimization, in other words optimising the conversion rate. The term covers the methods that get a larger share of a website’s or an app’s visitors to take action. Traffic, for its part, does not necessarily increase.

A conversion is not necessarily a sale. On a B2B website, it is often a qualified contact request. On a media site, it is a newsletter subscription; on an app, the creation of an active account. CRO therefore starts with a simple question: which action moves the business forward?

How do you calculate a conversion rate?

The conversion rate is simple to calculate. Divide the number of conversions by the number of visits, then multiply by 100. Some tools divide conversions by visitors rather than visits: always state which convention you use.

Fictional example: 30 quote requests for 1,500 sessions give a rate of 2%.

Type of conversion	Examples	Role in the analysis
Macro conversion	Purchase, quote request, appointment booking	Measures the value created for the business
Micro conversion	Add to cart, click on the phone number, guide download	Shows where the journey gets stuck before the final goal
Quality conversion	Lead turned customer, order not cancelled	Avoids optimising forms filled in by useless contacts

In Google Analytics 4, a conversion corresponds to a key event. You create or find the event that measures the important interaction. You then mark it as a key event (Google Analytics Help). Setting up tracking is covered in our web analytics guide.

Is there such a thing as a “good” conversion rate?

The industry averages that get passed around are rarely sourced. They mix unrelated websites, traffic and definitions of conversion.

The only useful benchmark is your own history, segmented by traffic source, device and page type. A visitor who arrives from a branded query does not convert like a visitor who comes from a display ad.

Why does CRO make acquisition pay off?

CRO works on the traffic you already have. Every visit costs something: a paid click in SEA, months of work in SEO, a budget in social ads. Improving the conversion rate lowers the cost of each customer acquired, across all channels at once.

Take a fictional example. A website gets 10,000 visits a month and converts at 1.5%, or 150 leads. Moving to 2% gives 200 leads, without a single extra euro spent on acquisition. If those visits come from Google Ads, the cost per lead drops by a quarter.

This is why a Google Ads account that has plateaued is often unblocked on the landing page rather than in the bids.

CRO also improves customer knowledge. Tests and session recordings reveal what reassures, what blocks and what is not understood. These lessons then feed the ads, the content and the digital strategy.

Which method should a CRO programme follow?

A CRO programme runs in a loop. Each cycle starts from a measured observation and ends with a documented decision.

1. Measure: define the conversions, check the key events, segment by source and device.
2. Diagnose: find the pages with high traffic and low conversion, and the funnel steps that lose the most visitors.
3. Understand: click maps, session recordings, user tests, questions to customer service.
4. Formulate hypotheses. For example: “If we show delivery costs on the product page, checkout abandonment will drop.” The reason: surprise costs drive people away.
5. Prioritise: expected impact, level of confidence, effort to deliver. The ICE or PIE frameworks help you decide.
6. Test: an A/B test if traffic allows. Otherwise, a rollout followed by a careful before/after comparison.
7. Document: hypothesis, result, decision. A losing test also provides useful information.

Why observing a few users is often enough

Qualitative research does not require hundreds of participants. According to Jakob Nielsen, five users reveal most of the usability problems of an interface. It is therefore better to run several small sessions between two iterations than a single large test (Nielsen Norman Group, March 18, 2000).

In practice, ask five people close to your target audience to complete a real task: find a product, request a quote. They think aloud during the exercise. The obstacles that recur from one participant to the next become your first hypotheses.

Which levers improve the conversion rate?

Landing pages: keep the promise of the ad

A landing page converts when it repeats exactly what prompted the click: same offer, same wording, same intent.

• A headline that answers the query.
• A clear benefit above the fold.
• Proof: reviews, client references, guarantees.
• A single main call to action.

Every unnecessary exit link is a door to abandonment.

UX and user journey: reduce the effort

Visitors must understand within a few seconds where they are, what is on offer and what to do next. That requires clear navigation, effective site search, content that answers objections and a controlled loading time. Page speed and stability are measured with the Core Web Vitals, which matter for SEO as much as for conversion.

Forms: every field has a cost

Ask only for what is needed to process the request. Provide visible labels, explicit error messages and input suited to mobile. The W3C’s WCAG 2.2 accessibility guidelines (Recommendation of December 12, 2024) cover these points: labels, error identification, input assistance. An accessible form is also simpler for everyone.

Conversion funnel: remove the obstacles to payment

The same Baymard survey ranks the reasons for abandonment. Extra costs considered too high come first. They are followed by slow delivery, lack of trust, mandatory account creation and an overly long checkout.

The CRO answers are well known: total cost visible early, guest checkout, short steps. Add suitable payment methods and reassurance near the button.

How do you run a successful A/B test?

An A/B test compares two versions of a page with two randomly selected groups of visitors. It measures which one converts better, and is only as good as its rigour:

• a hypothesis written before launch, and a primary metric chosen in advance;
• enough volume: calculate the required sample size before launching, otherwise the result is down to chance;
• complete cycles: run the test over whole weeks to smooth out day-of-week effects;
• no early stop because the variant “looks like” it is winning after three days;
• a single reading: the result is judged on the planned metric, not on the one that suits you.

On a low-traffic website, an A/B test can take months. It is then better to fix the obvious problems revealed by user research directly. Then track the before/after trend with caution.

Does A/B testing hurt SEO?

No, if you follow the Google Search Central guidelines (updated on December 10, 2025):

• do not show Googlebot different content from what visitors see (cloaking);
• add a canonical tag pointing to the original URL on the variants;
• use temporary 302 redirects, not 301;
• stop the test as soon as it has given its result.

The difference between the two redirects is explained on our page about the 301 redirect.

On the tools side, Google Optimize and Optimize 360 stopped working on September 30, 2023. Google points to third-party solutions integrated with GA4 and names AB Tasty, Optimizely and VWO (Google help on Optimize).

CRO and GDPR: what do you need to know?

CRO tools set trackers and collect browsing data. In France, the CNIL (the French data protection authority) allows a consent exemption for certain audience measurement tools, under strict conditions. Measurement must be carried out solely on the publisher’s behalf and produce anonymous statistics. It rules out any cross-referencing with other processing and any tracking on other websites (CNIL, audience measurement, July 4, 2025).

A/B testing, click map and session recording tools are not covered by this exemption page. Check their configuration and how consent is collected with your DPO before any rollout.

Another practical consequence: visitors who refuse trackers do not enter your measurements. The sample of your tests is reduced as a result.

Which tools should you use to get started with CRO?

Need	Type of tool	Examples
Measure conversions	Web analytics	Google Analytics 4, Matomo
See behaviour	Click maps, session recordings	Microsoft Clarity (free), Hotjar
Test variants	A/B testing platform	AB Tasty, Optimizely, VWO
Listen to users	User tests, surveys	Moderated sessions, end-of-journey questionnaire

Prices and features are compared in our selection of marketing and SEO tools. A spreadsheet and five user tests are enough to launch a first cycle. The tool comes after the method.

CRO, SEO and SEA: how do they fit together?

SEO and SEA bring in visitors; CRO converts them. The three feed each other. The queries that convert best in SEA show which SEO pages to prioritise. Conversely, pages that convert well deserve more paid traffic.

The only point of friction: a change designed for conversion can remove content that is useful for rankings. Before deleting a block of text or merging pages, check their organic traffic. Our SEO guide details what counts for ranking.