What is conversion rate optimisation (CRO)? A working definition
3 min read · updated
Conversion rate optimisation is the work of getting more of the visitors you already have to do the thing you built the site for: buy, sign up, book a call. It has three parts. Find where people who wanted to act gave up. Change the page so that point is easier, clearer or more trustworthy. Then prove the change helped by comparing it with visitors who did not see it, because without that comparison you are reading noise.
What counts as a conversion?
A conversion is the single action that pays for the visit. For a store it is an order. For a SaaS product it is a signup or a booked demo. For a clinic it is an appointment request. Pick one primary conversion per site and measure everything else as a step towards it, otherwise every team optimises its own number and the business number does not move.
Conversion rate is conversions divided by visitors over the same period. Decide early whether the denominator is sessions or people. A returning shopper who visits four times before buying is one converted person but a 25 percent session conversion rate, and mixing the two makes month on month comparisons meaningless.
Why is CRO usually cheaper than buying more traffic?
Paid traffic costs the same whether the page converts or not. If a store converts 1.5 percent of visitors and a fix lifts that to 1.8 percent, every rupee of ad spend now returns a fifth more orders. Buying the same increase through ads means a fifth more budget every month, for as long as you want the result.
The catch is that improvements compound only if they are real. A change that looked good for a week because of a sale, a newsletter or a lucky weekend gives nothing back next month. That is why the measurement half of CRO matters as much as the ideas half.
Where do visitors who wanted to buy give up?
Start where intent is highest and friction is visible. On most stores that is the product page and the cart, on most SaaS sites the pricing page and the signup form. Look for pages with plenty of visits and a sharp drop to the next step, then read those pages as a first time visitor on a phone with a slow connection.
The usual culprits are boring: a delivery date that is missing or vague, a total that grows at the last step, a return policy you have to hunt for, a form that asks for more than the action needs, and a page that takes long enough to load that people leave before it renders.
- Pages with many visits and a steep drop to the next step
- Questions people ask in chat or support before buying
- Steps where mobile converts far worse than desktop
- Visitors who return several times to the same page without acting
How do you know a change actually worked?
Compare against people who did not see it, at the same time. Before and after comparisons fail because everything else changes too: campaigns start and stop, festivals arrive, a competitor runs a sale. A control group that keeps seeing the old page absorbs all of that, so the difference between the two groups is the effect of your change and nothing else.
Then wait for enough evidence. Small samples swing wildly, and the first few days of any test tend to flatter whatever is new. Decide the sample you need before you start, and do not call a winner because one side is ahead on day three.
What does a CRO process look like week to week?
A sustainable rhythm is one or two well chosen changes in flight at a time, each with a written reason, a primary metric and a stopping rule. Ship the winners, keep a log of the losers, and look at the log every quarter, because the losers tell you what your customers do not care about.
Tools can carry more of this than they used to. Some, like IntellQ, run the loop continuously: they read where intent is high, propose a bounded change to an element you marked, hold back a control group, and keep the change only if it beats that group. Whether a tool or a person runs it, the discipline is the same.
Who should own CRO in a small company?
One person, with time protected for it, usually the founder or growth lead in a small brand. CRO touches design, copy, product and operations, so ownership spread across a team tends to mean nobody owns the result.
Questions
What is a good conversion rate?
There is no universal number. It depends on price, category, traffic source and device mix. Compare a page with its own history and with a control group, not with an industry average quoted without context.
Is CRO the same as A/B testing?
No. A/B testing is one way to measure whether a change worked. CRO also covers finding where visitors give up and deciding what to change, which no test can do for you.
How long before CRO shows results?
A single change needs enough visitors in both groups to read a difference, which is days on a busy store and weeks on a small one. The programme as a whole pays back when several real wins stack.
Do I need a developer for CRO?
For layout changes, usually yes. For copy, visibility and links on elements that already exist, tools that edit marked elements in the browser can do it without a code deploy.