Conversion rate optimisation basics: turn traffic into revenue

Most businesses obsess over getting more traffic and ignore what happens once it arrives. That is the wrong order. If your website turns visitors into enquiries at half the rate it could, then every dollar you spend on SEO, ads and social is working at half efficiency — and no amount of extra traffic fixes a page that does not convert. Conversion rate optimisation, or CRO, is the discipline that closes that gap. This is how we explain it to the founders and marketing leads we work with.

What CRO actually is

Your conversion rate is the share of visitors who take the action that matters — submit an enquiry, book a call, add to basket, complete a checkout. CRO is the practice of increasing that share without necessarily increasing traffic. It is website work, analytics work and copywriting work in equal measure, which is why it lives at the seam between our website design and development and everything upstream of it.

The reason it deserves attention out of all proportion to how much it usually gets is simple: it multiplies every other channel. Lift the rate at which visitors convert and you have not improved one campaign — you have improved the return on all of them at once, including the traffic you are already paying for and the organic visits you worked months to earn. That is a rare kind of leverage in marketing, and it compounds.

The method: measure, hypothesise, test, learn

The single biggest mistake in CRO is treating it as a matter of opinion. Somebody decides the button should be green, or the headline should be punchier, the change ships, and nobody ever knows whether it helped. That is redecorating, not optimising.

Real CRO runs a loop:

• Measure. Look at where people actually drop off, using analytics, session recordings and the funnel numbers. You cannot fix what you have not located.
• Hypothesise. Form a specific, falsifiable claim — "visitors abandon the enquiry form because it asks for too much too early" — not a vague feeling that the page could be better.
• Test. Change one thing, show the variant to a portion of traffic, and let the two versions compete on real behaviour.
• Learn. Keep the winner, discard the loser, and — this is the part people skip — write down why, so the next hypothesis is smarter than the last.

The point is not any single test. It is the habit. A team that runs this loop honestly for a year knows more about its own customers than any amount of best-practice blog reading could teach.

The highest-return fixes

Before anyone runs a clever multivariate experiment, there are a handful of fixes that reliably move the needle because they address why people leave in the first place. In rough order of how often we find them broken:

• Message match. The promise that got someone to click — the ad, the search result, the social post — must be the first thing they see when they land. When the page talks about something subtly different, people bounce before they have read a sentence.
• Clarity of the offer. If a visitor cannot tell what you do, who it is for and what to do next within a few seconds, no button colour will save you. Clarity beats cleverness every time.
• Page speed. A slow page loses people before they see the offer at all. This is where CRO and technical performance overlap; we treat sub-two-second mobile loading as a conversion issue, not just a developer's.
• Form friction. Every field you ask for costs you completions. Ask for what you genuinely need to have the next conversation, and no more.
• Trust signals. Real proof — genuine reviews, recognisable clients, clear guarantees, a visible address and phone number — lowers the perceived risk of acting. It has to be true; fabricated proof is worse than none.
• Mobile. A large share of traffic in Singapore is on a phone. If the experience is built for a desktop and merely tolerated on mobile, that is where most of your lost conversions are hiding.

How to test honestly

Testing is where good intentions go to die, because it is easy to run a test badly and easier still to read one wishfully. Two disciplines keep you honest.

First, respect sample size. A variant that is "winning" after forty visitors is telling you nothing — small numbers swing wildly by chance. A test needs enough traffic and enough conversions, over enough time, before its result means anything. Businesses with modest traffic have to accept that some tests will simply take longer than patience would like, and calling them early is the same as guessing.

Second, change one thing at a time where you can, so that when the result moves you know what moved it. Rewrite the headline, the hero image, the button and the form all at once and a lift tells you the page is better but not why — which means you have learned nothing you can reuse. Learning is the entire return on testing.

Where to start

You do not need a testing programme to begin. Start by finding the single biggest leak in your funnel — usually a step where a large share of people drop off — and fix the obvious things first: the message match, the clarity, the speed, the form. Measure honestly, keep what works, and build the habit from there. Optimisation is a loop you run, not a project you finish, and the businesses that treat it that way quietly out-earn the ones chasing the next traffic source.