For many business websites, the quickest gain isn’t more visitors but more of the existing ones getting in touch. Conversion rate optimisation (CRO) is the discipline of finding out why interested people leave without acting, and fixing it.
It has a reputation for button-colour tests and countdown timers. On a service or B2B website the real work is plainer and more useful: a clearer headline, proof in the right place, a form that works on a phone, a page that loads before patience runs out. Conversion is one ingredient of what makes a good website, and it depends on nearly all the others.
The short answer
Conversion rate optimisation (conversion rate optimization in American spelling) is the practice of increasing the share of visitors who take a meaningful action, such as sending an enquiry, booking a call or requesting a quote.
- The formula: conversion rate = conversions ÷ visitors × 100. For example, 45 enquiries from 3,000 sessions is a 1.5% conversion rate.
- The process: measure, research, form a hypothesis, prioritise by impact and effort, change or test, then review. Repeat.
- The principle: remove friction (effort) and doubt (uncertainty) from the path to an enquiry, rather than pressuring visitors with tricks.
How to calculate your website conversion rate
The arithmetic is easy; the decisions behind the two numbers are where mistakes happen.
Decide what counts as a conversion
Count the actions that start a sales conversation: a form reaching its success state, a tap on your phone number, a click to a messaging app, a completed booking. Not visits to the Contact page, or Submit clicks that may have failed. Track softer signals, such as viewing pricing, separately.
In Google Analytics 4 you mark each of these as a key event, the name GA4 has used for conversions since March 2024. Our guide to tracking enquiries in GA4 walks through the set-up.
Choose sessions or users, and stay consistent
A session-based rate measures conversions per visit; a user-based rate measures them per person, and usually runs higher, because many people visit more than once before they get in touch. Neither is wrong. Switching between them month to month is.
Why conversion rate benchmarks mislead
Published “average conversion rates” are widely shared and rarely useful. The sites behind them count conversions and visitors differently, mix paid landing pages with blog-heavy sites, and sell everything from free downloads to six-figure contracts.
A site with a busy blog will have a low site-wide rate, and that can be perfectly healthy. The only comparison that tells you much is your site against itself.
How to set your own baseline
- Check tracking against reality. Compare last month’s key events with the enquiries your team actually received. If they don’t match, fix measurement first.
- Use a long enough window. Three months or more where you can, noting seasonal peaks, campaigns and outages.
- Segment it by device, channel, landing page and language. A healthy average can hide a weak mobile rate, which needs a different fix from a weak headline.
- Record counts and quality, not just the rate: how many enquiries arrived and how many were a genuine fit. Cutting unqualified traffic raises the rate without adding a customer.
- Write it down with dates and definitions, so later comparisons are like-for-like.
CRO is about friction and doubt, not tricks
Every visitor deciding whether to get in touch weighs two questions: is this worth the effort, and can I trust it? Nearly every conversion problem makes one of those answers worse.
| At a glance | Friction | Doubt |
|---|---|---|
| What the visitor thinks | “This is hard work” | “I’m not sure about this” |
| Typical causes | Slow pages, hidden contact options, long forms, layouts that break on phones | Vague headlines, no pricing guidance, thin proof, silence about what happens next |
| Typical fixes | Faster pages, shorter forms, a clear next step on every page | Specific copy, visible evidence, an explained process |
Fake countdown timers, invented scarcity and guilt-trip decline links may nudge a click, but in a considered purchase they add doubt at the worst moment, and in many countries fake deadlines and false scarcity can breach consumer protection law. Our guide to website design mistakes covers the quieter, more common culprits.
The conversion rate optimisation process: a six-step cycle
The CRO process is a loop, not a project: each pass sharpens the next.
1. Measure
Confirm tracking and the baseline, then tie the goal to the business: which enquiries matter most, and what is one worth? Those answers belong in your website strategy, and stop you optimising for the wrong action.
2. Research
Find where people drop out, then why.
- Where: GA4’s landing page report and path explorations, plus session recordings and form drop-off from a behaviour-analytics tool (with consent, and form fields masked).
- Why: ask sales what prospects ask before they buy, reread recent enquiries, add a one-question survey, and watch five people attempt a real task. A website UX audit combines these into one structured review.
3. Hypothesise
Turn each finding into a statement you can check:
Because we saw [evidence], we believe [change] for [audience] will [result]. We’ll know when [metric] moves over [period].
For example: because most sales calls open with “roughly what does this cost?”, we believe a pricing guide on the main service page will lift mobile quote requests over eight weeks without lowering lead quality.
4. Prioritise by impact and effort
Score each idea on impact (how many visitors it touches, how close to the decision), confidence in the evidence, and effort, as frameworks such as ICE (impact, confidence, ease) do.
| Idea | What to do |
|---|---|
| High impact, low effort | Do it now |
| High impact, high effort | Plan and scope it |
| Low impact, low effort | Batch with other work |
| Low impact, high effort | Drop it |
Broken things skip the queue. A form that fails on some phones is a bug, not a hypothesis.
5. Change or test
With enough traffic, run an A/B test. Most service sites don’t have that volume, so make the change, note the date in analytics, and change one significant thing per page at a time.
6. Review
After the agreed period, compare against a like-for-like baseline and check lead quality with sales. Log every result, failures included, then start the next loop. That rhythm is continuous improvement in practice.
The eight levers that decide conversions
Almost every conversion problem on a service or B2B site traces back to one of eight levers.
Message: can visitors tell what you do in seconds?
The first screen should state what you offer, who it’s for and the outcome, in customers’ words. Check it yourself: show the page to someone for five seconds, then ask what you do. See how to write landing page copy and the landing page structure that orders it.
Offer: is the next step worth taking?
“Contact us” asks for commitment and promises nothing. A free consultation, a fixed quote or a pricing guide gives people a reason to act. Offers, contact channels and lead routing are covered in website lead generation.
Proof: why should they believe you?
Put specific, checkable evidence beside the claims it supports: real project outcomes, client names you may use, photos of real work, clear company details. See how visitors judge website credibility.
Calls to action: is the next step obvious?
Use a specific verb and outcome, place it where decisions happen, and offer a lower-commitment alternative. Our call to action best practices cover wording and placement.
Forms: can people finish them?
Ask only what your first reply needs, test every form on a phone, and confirm each submission reaches a person. Web form design has the detail.
Speed: does the page load before patience runs out?
Google rates a Largest Contentful Paint of 2.5 seconds or less, Interaction to Next Paint of 200 milliseconds or less and Cumulative Layout Shift of 0.1 or less as good, at the 75th percentile of real page loads. Check it yourself: run key pages through PageSpeed Insights, mobile first, then work through the website speed optimisation guide.
Mobile: does it work with a thumb on mobile data?
Test key tasks on your own phone, away from office Wi-Fi. Mobile-first vs responsive design explains layouts that hold up.
Journey: does every page lead somewhere?
Visitors often land on service pages and articles rather than the homepage, and considered purchases take several visits and people. Map what they need at each stage with customer journey design, and match campaign traffic to the right landing page, homepage or service page.
Why the levers move together
A faster page with a vague headline still fails, and so does a sharp headline above a broken form. That is why conversion is never only a design, copy or development job. In our property developer redesign, several levers moved at once: load time cut from nine seconds to two, pages rebuilt around the buyer journey and for phones, and tracked enquiry forms on every project page. Buyer enquiries rose 40% in the three months after launch. The changes landed together, so no single one can take the credit.
Why low-traffic sites should research before they test
A/B tests need volume. From a 2% baseline, reliably detecting a rise to 2.4% (a 20% relative improvement) at the conventional 95% confidence and 80% power takes roughly 21,000 visitors per version. A site with 3,000 visitors a month would wait over a year even if every visitor entered the test, and stopping early because one version looks ahead is how false winners get declared.
For most service and B2B sites, it’s better to:
- Fix obvious problems outright; they don’t need proving.
- Lean on qualitative research, which shows why people struggle.
- Compare before and after, allowing for seasons, campaigns and traffic mix.
- Try new messages cheaply in ads or emails first.
A/B testing landing pages explains when testing is worth it, and why your landing page isn’t converting gives a diagnosis order for one struggling page.
A CRO checklist
Frequently asked questions
What is a good conversion rate for a website?
There isn’t a universal figure worth chasing, because sites count conversions and visitors differently. A good rate is one that rises against your own baseline while enquiry quality holds or improves.
Is CRO the same as A/B testing?
No. A/B testing is one way to validate a change, and it needs plenty of traffic. CRO is the whole cycle of measuring, researching, prioritising and improving; most of it needs no split test.
What is a CRO audit?
A structured review of the path to enquiry: tracking, key landing pages, messaging, proof, calls to action, forms, speed and mobile usability. It should end in a prioritised list of fixes, with the evidence behind each one.
Where to start
Check your tracking, record a baseline, pick your three most important pages and run the cycle once. One honest loop teaches you more than any benchmark.
For an outside view first, a free website audit checks whether a first-time visitor can tell what you do, why to trust you and how to get in touch, and flags the speed and mobile problems that add friction. If the findings point at the whole site’s structure, a website redesign fixes them at the source; if one campaign needs a focused page, a landing page is the better tool.