Conversion Rate Optimisation (CRO): A Step-by-Step Guide

Short answer: Conversion rate optimisation (CRO) is the process of increasing the share of website visitors who take the action you want, such as buying, booking or enquiring. It runs as a loop: research where and why people drop off, write a testable hypothesis, run a controlled test, then keep what wins and learn from what doesn’t.

The typical situation: traffic is steady, ad spend is rising, and the number of enquiries hasn’t moved in months. Someone suggests a new button colour or a homepage redesign, but nobody can say where visitors actually give up, or why.

This guide is for business owners, marketing managers and in-house teams who already get traffic and want more of it to become leads or sales. If you still need convincing that conversion rate deserves attention before traffic, start with why traffic isn’t the problem, your conversion rate is.

This post covers how to do CRO: the four-step process, a page checklist, sample size and significance in plain English, tools now that Google Optimize is gone, and the mistakes that waste testing budgets.

What is conversion rate optimisation (CRO)?

CRO is a structured way of finding and removing the things that stop visitors from converting. It relies on evidence (analytics, recordings, customer feedback and controlled tests) rather than opinions about what looks good.

A “conversion” is whatever action matters to your business: a purchase, a form submission, a demo booking, a WhatsApp click, a call. Your conversion rate is the number of conversions divided by the number of visitors (or sessions) in the same period. If 4,000 people visit a landing page and 120 submit the form, the conversion rate is 3%.

CRO is not a list of tricks such as “make the button orange”. A change that works on one site can do nothing on another, because the audience, offer and traffic source differ, so the process matters more than any single tactic.

What are the steps in the CRO process?

The CRO process has four repeating steps: research, hypothesis, test and learn. Each round feeds the next, so the site improves continuously rather than in one big redesign.

1. Research: find where and why people drop off. Start with numbers, then add context. In Google Analytics 4 (GA4), mark the actions that matter as key events, then build a funnel from landing page to conversion and look for the step with the biggest drop. Watch session recordings and heatmaps for those pages, then add qualitative input: on-page surveys (“What stopped you from booking today?”), sales and support team notes, and the questions customers ask before they buy.
2. Hypothesis: turn a problem into a testable idea. Write each idea as: “Because we saw [evidence], we believe changing [element] for [audience] will improve [metric].” For example: “Because recordings show mobile visitors abandoning the quote form at the phone-number field, we believe making that field optional will increase form completions on mobile.” Prioritise hypotheses by expected impact, confidence in the evidence and effort to build.
3. Test: compare the change against the current version. Run an A/B test where traffic is randomly split between the original (control) and the new version (variant). Decide your primary metric, the sample size you need and the test duration before you start (see the testing section below). If you don’t have enough traffic to test, make the change and monitor the before-and-after numbers carefully, treating the result as a signal rather than proof.
4. Learn: record the result and decide what’s next. Ship winners and record every result, including losses and “no difference” outcomes, because a losing test still tells you something about your customers. Keep a simple test log (hypothesis, dates, traffic, result, decision) so future tests build on past ones instead of repeating them.

Which page elements should you optimise first?

Check the elements closest to the conversion first: the headline and value proposition, the call to action (CTA), forms and checkout, trust signals and page speed. Small problems here cause large drop-offs. Run this checklist against your highest-traffic landing pages and main conversion path.

For ecommerce, checkout deserves special attention. Baymard Institute’s cart abandonment research puts the documented average abandonment rate at 70.22% across 50 studies. Among US shoppers who abandoned during checkout, the most common reasons were extra costs being too high (40%), slow delivery (20%), not trusting the site with card details (19%), being asked to create an account (18%) and a checkout that was too long or complicated (17%). Most of these come down to design and policy decisions you control.

How do you write headlines and CTAs that convert?

Write headlines that name the outcome the visitor wants, and CTAs that say exactly what happens when they click. Clarity beats cleverness: if visitors can’t tell what you offer within a few seconds, no button colour will help.

• Lead with the outcome. “Get qualified leads from your website in 30 days” tells a visitor more than “Digital solutions for growth”. Only promise results you can back up.
• Name the pain point. A headline that reflects the visitor’s problem (“Tired of quotes that take a week?”) shows you understand their situation.
• Use specific action verbs in CTAs. “Download the checklist”, “Book a 20-minute call” and “See pricing” set clearer expectations than “Submit” or “Learn more”.
• State the benefit or remove the risk. Add a short line under the CTA, such as “No credit card needed” or “We reply within one working day”, if it’s true.
• Use urgency honestly. Real deadlines and genuine stock limits can help. Fake countdown timers damage trust once people notice them.

How many form fields should you have?

Ask only for the information you need to take the next step, and nothing more. For many lead forms that means name, email or phone, and one qualifying question. You can collect the rest later in the sales conversation.

Each extra field adds effort, especially on mobile. Rather than following a fixed rule, use form analytics or recordings to find the field where people abandon, then test removing it or making it optional. Check the basics too: labels that stay visible while typing, error messages that say exactly what is wrong, and a form people can find without hunting.

If your customers prefer phone or WhatsApp, offer that as an alternative to the form. Our product Sandesh handles WhatsApp customer conversations, but the principle applies whatever tool you use.

How does A/B testing work, and how much traffic do you need?

An A/B test randomly shows half your visitors the current page and half a changed version, then compares conversion rates. You need enough visitors and conversions for the difference to be reliable, usually more than people expect.

What sample size does an A/B test need?

If you flip a coin ten times and get seven heads, you wouldn’t conclude the coin is biased. A/B tests are the same: small samples produce big random swings. Sample size is the number of visitors each version needs before the result means anything.

Three things drive it: your current conversion rate, the smallest improvement you care about detecting (the “minimum detectable effect”) and how sure you want to be. The smaller the improvement you want to detect, the more traffic you need. Here are approximate figures for a page converting at 3%, using a standard two-sided test at 95% significance and 80% power (our calculation; you can check your own numbers with a free tool such as Evan Miller’s sample size calculator):

The practical lesson: low-traffic sites should test bold changes (a new offer, a rewritten page, a much shorter form) rather than small tweaks like button colour, because only big effects show up with limited traffic.

What does statistical significance mean?

Statistical significance tells you how unlikely your result would be if the change actually made no difference. A result “significant at 95%” means that, if there were truly no difference, you would see a gap this large only about 5% of the time by chance. It does not mean the variant is guaranteed to be better, or that the lift will be exactly what the test showed.

The most common error is checking results every day and stopping as soon as the tool says “significant”. Evan Miller explains in How Not to Run an A/B Test that repeatedly checking and stopping early inflates false positives. Set the sample size in advance and let the test run until it gets there, or use a tool built for sequential analysis. Run tests for at least one full week, ideally two, so both weekday and weekend behaviour are included.

Which CRO tools should you use now that Google Optimize is gone?

You need three types of tool: analytics to find problems, behaviour tools to understand them, and a testing tool to validate fixes. Google Optimize, which many small businesses used for free A/B testing, was sunset on 30 September 2023, so you’ll need an alternative for testing.

Microsoft’s documentation describes Clarity as free, with session recordings, heatmaps and funnels, which makes it a sensible starting point for smaller sites. Google’s Optimize sunset notice says it is working on GA4 integrations with AB Tasty, Optimizely and VWO, so those are a natural shortlist if you want test results to flow into your analytics.

Respect privacy rules when you add these tools. Mask form inputs in recordings, update your cookie or consent setup and privacy policy, and don’t record pages where people enter payment or health details.

What improves conversion rates beyond the page itself?

Speed, trust and follow-up all affect conversions alongside layout. A well-designed page still underperforms if it loads slowly, looks untrustworthy or leaves warm leads waiting.

• Speed. In Deloitte’s Google-commissioned Milliseconds Make Millions study of 37 brand sites, a 0.1-second improvement in mobile site speed was associated with 8.4% higher conversion rates for retail sites and 10.1% for travel sites. That is a correlation across brands, not a forecast for your site, but it shows why speed belongs in a CRO programme.
• Trust and social proof. Reviews, testimonials and real photos of your team or products reduce uncertainty. If you need more reviews, see our guide on how to get more Google reviews.
• Remarketing. Ads aimed at past visitors, cart abandoners and existing customers bring people back to finish what they started.
• Personalisation. Showing content that matches a visitor’s location, device or previous behaviour can make pages more relevant. Start simple, for example by matching the landing page headline to the ad they clicked.
• Response time. For lead generation, how quickly you reply affects whether an enquiry becomes a customer. An automated acknowledgement and a fast human follow-up are part of the conversion path.

If you keep patching an old site rather than fixing it, check our list of signs your website is outdated to decide whether a refresh or a rebuild makes more sense.

What are the most common CRO mistakes?

The most common mistakes are testing without research, stopping tests too early and copying other sites’ “best practices” without checking whether they apply to your audience.

1. Testing random ideas. A test based on a colleague’s opinion is a guess. A test based on a problem you saw in the data has a reason to work, and teaches you something either way.
2. Stopping at the first “significant” result. Decide sample size and duration in advance and stick to them.
3. Testing tiny changes on low traffic. A button colour test on a site with 2,000 visitors a month is very unlikely to reach a reliable answer.
4. Measuring the wrong metric. More clicks on a CTA mean little if form completions or sales don’t rise. Pick the metric closest to revenue that you have enough data for.
5. Ignoring mobile. Compare conversion rate by device. If phones lag well behind desktop, the fix is often a simpler mobile form or a sticky CTA.
6. Changing several things without knowing which mattered. That’s fine for a full page redesign, but then treat it as one test of the whole page, not proof that each change worked.
7. Not writing results down. Without a test log, teams repeat failed tests and forget why winning changes were made.
8. Fake urgency and dark patterns. Hidden fees, pre-ticked boxes and false scarcity may lift short-term numbers but invite refunds, complaints and distrust.

Frequently asked questions

What is a good conversion rate?

There is no single good conversion rate, because it depends on your industry, price point, traffic source and what counts as a conversion. A newsletter signup and a high-value B2B purchase will never convert at similar rates. The most useful benchmark is your own: measure your current rate by page, device and channel, then aim to improve it steadily. Compare against industry figures only when you know they measure the same action.

How long should an A/B test run?

Run a test until it reaches the sample size you calculated before starting, and for at least one full week so that weekday and weekend behaviour are both included. Two weeks is safer for most sites. Avoid stopping early because the result looks significant after a few days, as repeated checking increases the chance of a false winner. If the required sample size would take months, test a bolder change instead.

Can you do CRO with low traffic?

Yes, but you’ll rely more on research and less on A/B testing. Use recordings, heatmaps, surveys and customer interviews to find clear problems, fix them and monitor the before-and-after numbers. When you do test, test big changes such as a new offer or a much shorter form, because small tweaks need far more traffic to detect. Watching a few real customers try to complete a task on your site can also reveal obvious issues quickly.

What replaced Google Optimize?

Google Optimize was sunset on 30 September 2023 and Google did not launch a direct free replacement. For A/B testing, Google points users to third-party tools and says it is working on Google Analytics 4 integrations with AB Tasty, Optimizely and VWO. For understanding behaviour, free tools such as Microsoft Clarity provide heatmaps and session recordings, which you can use alongside GA4.

Is CRO the same as UX design?

They overlap but aren’t the same. UX design focuses on making a product or website easy and pleasant to use. CRO focuses on a measurable business outcome, such as purchases or enquiries, and uses experiments to prove which changes improve it. Good UX usually supports conversion, and many CRO wins come from fixing usability problems, but CRO also covers offers, pricing presentation, copy and trust signals.

Does CRO affect SEO?

CRO and SEO are separate disciplines, but they support each other. Clearer pages, faster loading and a better mobile experience help visitors and are aligned with what Google says it rewards in page experience. Google’s guidance on website testing says not to show search engines a different version from users, to use rel=”canonical” on alternate test URLs, to use 302 (temporary) redirects for variants and to remove test elements once the test ends.

If you want help finding where your site loses visitors and fixing it, our UI/UX design team can audit your conversion path, prioritise what to test and redesign the pages that matter most.