Conversion Rate Optimisation: Turning More Visitors into Customers

Conversion rate optimisation (CRO) is the work of getting more of your existing visitors to do the thing you want, such as enquire, book or buy, by finding where they drop off and fixing those points one at a time. It follows a simple loop: measure accurately, find the biggest leak, fix it or test a fix, and check the result. Most small business sites get their best gains from fixing obvious problems found in analytics and session recordings, because they do not have enough traffic for reliable A/B tests.

This article is about the process. If your site gets visits and no enquiries and you want a quick list of usual causes, start with why a website gets visitors but no leads. Here I cover how to measure, where to look, what to change on each type of page, and what A/B testing can and cannot tell a smaller site. No uplift figures are promised, because they depend entirely on your site and traffic.

#The CRO process in one loop

CRO works as a repeating loop, not a one-off redesign. Each round targets one leak, makes one change, and measures it, so you learn what actually moves your numbers.

A hypothesis is a short sentence that names the problem, the change and the expected effect, for example: "Phone visitors leave the quote form at the budget field, so making it optional should increase completed forms." Writing it down stops changes being made on taste alone, and gives you something to check afterwards.

#Step 1: Make sure you are counting the right things

Before changing anything, confirm that your analytics counts conversions correctly. In Google Analytics 4, an action that matters to the business, such as a form submission, a call click or a purchase, has to be set up as an event and marked as a key event (opens in a new tab). Many small business sites have GA4 installed but no key events, so the reports show traffic and nothing else.

Check these:

• Each form sends a key event only when it is successfully submitted, not when the page loads.
• Phone, WhatsApp and email clicks are tracked as separate events.
• Purchases record the order value, and test orders are excluded.
• Your own visits and staff visits are filtered out where possible.

My articles on GA4 for business owners and GA4 vs Google Tag Manager explain the setup. Without this step, CRO is guesswork with charts.

#Step 2: Find the leaks

A leak is a step where a large share of visitors leave. You find them by combining three sources, each answering a different question.

Analytics funnels. GA4's funnel exploration (opens in a new tab) shows the steps users take to complete a task and where they drop off between them. Build one for your main journey, for example landing page, service page, contact page, form submitted. Split it by device; phone visitors often behave very differently from desktop ones.

Recordings and heatmaps. Microsoft Clarity (opens in a new tab) is free and offers session recordings, heatmaps and funnels. Watch twenty recordings of people who reached your key page and left. Patterns appear quickly: rage clicks on something that is not a link, scrolling past the price, a sticky banner covering the button, a form error nobody can see. Mask personal data in forms, and mention the tool in your privacy notice and cookie consent.

Form data. Track which field people were on when they abandoned a form. A field that asks for budget, company size or a full address on the first contact is a common culprit.

#Step 3: Fix by page type

Different pages lose people for different reasons. Start with the page type closest to the money and with the most traffic.

#Landing pages

A landing page is where paid or campaign traffic arrives, and it has one job.

• Match the headline to the advert or search term that brought the visitor. A mismatch is the most common reason people leave in seconds.
• Keep one primary action. Remove the main menu and secondary links if they pull people away.
• Put the action and a short proof point, such as a client logo, a review or a real number from your work, near the top on a phone screen.
• Keep the form short. Ask only what you need to reply.
• Check speed on a phone. Slow pages lose visitors before they see anything; my guide to Core Web Vitals explains the measures.

#Service pages

Service pages persuade visitors who are comparing you with others.

• Say who the service is for and what result it gives, in the first lines.
• Show proof: real projects, case studies, named reviews.
• Give a price range or "from" price where you can. Visitors who cannot judge cost often leave to find someone who shows it.
• Answer the objections you hear on sales calls, in a short FAQ.
• Repeat the call to action after each major section, not only at the bottom.
• For complex quotes, an estimator or quote form can turn browsers into leads; see lead generation calculators and quote forms.

#Product pages and checkout

On online stores, the product page and checkout carry most of the revenue.

• Product pages need clear photos, price including tax and delivery estimate, stock status and returns information near the add-to-cart button. My guide to WooCommerce product pages that sell goes further.
• At checkout, offer guest checkout, show all costs before the last step, keep fields to what is needed, and offer the payment methods your customers use, such as UPI in India.
• Test every payment method and error message on a real phone. A failed payment that shows no message looks like a broken shop.

Why checkout loses sales covers the main abandonment causes in detail.

#Step 4: A/B testing, honestly

An A/B test shows half your visitors version A and half version B, then compares conversion rates. It is the most reliable way to prove a change worked, but only when enough people go through both versions.

The required sample size depends on your current conversion rate and the smallest change you want to detect. Using the standard calculation at 95% confidence and 80% power (the usual settings), these are worked examples:

These numbers are arithmetic, not predictions: they say how many visitors you need to tell a real difference from chance, not what result you will get. A page with 3,000 visitors a month would need over nine months to test the 3% to 3.6% case, and many things change on a website in nine months.

Rules that keep tests honest:

• Decide the sample size and run time before starting, and do not stop early because one version looks ahead.
• Run for whole weeks, so weekday and weekend visitors are both included.
• Test one meaningful change, not five small ones at once.
• Count the key event that matters, such as enquiries or orders, not clicks.

Google Optimize, once the common free option, was shut down on 30 September 2023 (opens in a new tab), so testing now needs a third-party tool or a custom setup.

#When traffic is too low to test

Most small business sites are in this group. Do not run tests that cannot reach a result. Instead:

1. Fix clear problems directly: broken forms, missing prices, invisible buttons, slow pages.
2. Compare the same length of time before and after each change, and change one thing at a time.
3. Note anything else that changed, such as an ad campaign, a holiday or a price change.
4. Look at the quality of enquiries as well as the count. Fewer, better enquiries can be a win.

#What a CRO engagement includes

Whether you do it in-house or hire help, a proper CRO process should leave you with:

• Verified tracking of every key action, with a short note of what counts as a conversion.
• A funnel report by device and a written list of leaks, ranked by traffic and value.
• A hypothesis and expected effect for each change before it is made.
• A change log with dates, so results can be checked later.
• A monthly review of the numbers, not a one-time report.

My conversion rate optimisation service follows this process, and analytics and tracking setup covers GA4, Tag Manager and conversion tracking if the measurement side needs fixing first.