CRO is the work of getting more of your existing visitors to buy, enquire or book. Here is what it involves and how to run it properly.
Conversion rate optimisation, or CRO, is the work of getting more of your existing visitors to do what you want: buy, enquire, book, or call. You study how people behave on the site, change one specific thing, measure what happens, and keep the version that wins.
The maths is the point
Conversion rate is conversions divided by sessions. 40 enquiries from 2,000 sessions is 2%. Move that to 3% and the same 2,000 sessions produce 60 enquiries. That is a 50% lift in leads with no extra ad spend, no new content and no waiting for rankings. It also compounds, because every future visitor lands on the better page.
This is why CRO usually beats buying more traffic when budgets are tight. Traffic costs money every month. A fixed form or a rewritten offer is a one-time cost.
What the work looks like
Good CRO is unglamorous and evidence-led. It runs in a loop.
- Research. Analytics, session recordings, form drop-off reports, sales call notes and a handful of user tests. You are looking for the points where intent dies.
- Hypothesis. A written statement: we believe changing X will improve Y for Z visitors, because of this evidence. If you cannot write it down, you do not understand the problem yet.
- Test. An A/B test where traffic supports it, a before-and-after measurement where it does not.
- Roll out and record. Ship the winner, document the losers. Losing tests are cheap insurance against someone proposing the same idea again in six months.
Where sites usually leak
Four problems account for most of what we find. Forms that ask for information the business does not actually need at that stage. Pricing hidden when the buyer expects to see it. Mobile layouts where the main button sits below three screens of scrolling. And slow pages, because every extra second of load time on a commercial page costs conversions, and the effect is worse on mobile.
Trust is the quieter one. Reviews, real photography, named people, clear delivery and returns terms. These rarely win a dramatic test on their own, but their absence caps everything else you do.
How long a test needs to run
Longer than most people want. You need enough conversions per variant for the result to mean anything, usually a few hundred, and at least two full business weeks so you capture both quiet and busy days. Calling a test after three days because one version is 40% ahead is the most common way to ship a change that does nothing.
If your site gets 500 sessions a month, formal A/B testing will not work. Fix the obvious problems, measure quarter-on-quarter, and lean on qualitative research instead of statistics.
CRO and SEO pull in the same direction
Search brings people in. Conversion decides whether that traffic was worth paying for. We run the two together for most clients, which is why our CRO work starts with the pages that already rank rather than a site-wide redesign. A page sitting at position three for a commercial term is the cheapest place in the business to find extra revenue.
Common questions
What is a good conversion rate?
It depends entirely on the business. Ecommerce often sits between 1% and 3%, lead generation sites between 2% and 5%, and a booking page reached from branded search might hit 20%. Compare yourself to your own last quarter rather than an industry average.
Do I need a testing tool to start?
Analytics with properly configured goals, plus session recordings, will surface more than enough to work on for the first few months. Add a testing tool once you have the traffic to use it honestly.
Can CRO hurt my rankings?
Only if you strip out content or slow the page with heavy scripts. Serve the same content to Google and to visitors, keep tests running for weeks rather than months, and there is no ranking risk.
For plain English definitions of the terms used here, see conversion rate optimisation (CRO) in our SEO glossary.
