On this page
Conversion rate optimisation is the cheapest growth there is. You measure it as the conversion rate: enquiries, orders or calls divided by visitors. A site with 2,000 visitors and 40 enquiries converts at 2 percent. Lift that to 3 percent and the same traffic brings half as many enquiries again, whether it came from Google, paid ads or a trade fair. Every click you paid for earns more, so your cost per acquisition falls with it.
The discipline took shape in the 2000s, when advertisers found that buying more traffic kept getting dearer while their pages wasted most of what they already had. Google released a free testing tool in 2007 and retired its successor, Optimize, in 2023. The job itself remains: understand why visitors leave, and remove the reason.
Same budget, different page
A heat pump installer spends €1,000 a month on Google Ads at €2 a click: 500 visitors. The figure shows what they bring at four conversion rates. At 1 percent a lead costs €200; at 5 percent it costs €40. If one lead in four signs, that is the difference between just over one and just over six customers a month, for the same money. No bid adjustment does that.
Five hundred visitors at four conversion rates
500 visitors for 1,000 euros
Worked example. Budget and traffic stay the same; only the page changes.
Where sites lose their visitors, in order of damage
- The promise does not matchSomeone clicks an ad for "bathroom renovation cost" and lands on a home page without a price. One page per question fixes more than any button colour.
- The page is slowGoogle found that more than half of mobile visits are abandoned when a page takes longer than three seconds to load.
- The form asks too muchEvery field you do not need for a first conversation costs enquiries. Name, phone and the question are often enough.
- There is no proofReal projects, real names and an address beat general promises.
- There is only one way inPeople who prefer to call or message must be able to. A second route for a second kind of buyer often beats a cleverer form.
Why A/B tests rarely prove anything on small sites
An A/B test shows two versions of a page to different visitors and measures which wins. That only works with a lot of traffic. Statistician Robert Lehr's rule of thumb: per variant you need about 16 x p x (1 - p) / d² visitors, where p is your current conversion rate and d the difference you want to detect.
With modest traffic, watch instead of test. Ask five customers to use your site while you say nothing; where they hesitate is the leak. Add one question to the confirmation email: "What nearly stopped you?" Change one thing and compare four weeks before with four weeks after. And make the page fast first; the website speed test shows where it stands.
Converting buyers in another country
Selling abroad adds leaks that no heatmap shows:
| Leak | What the buyer thinks | Fix |
|---|---|---|
| Prices in a foreign currency | "What will this cost me?" | Local currency, VAT shown the local way |
| No familiar payment method | "I cannot pay the way I always do" | iDEAL in the Netherlands, Bancontact in Belgium, cards and wallets elsewhere |
| A foreign phone number | "Who will answer, and when?" | A local number or clear hours in the buyer's time zone |
| A translated page | "This company is not really here" | Pages written for the market, not run through a tool |
Localised pages, local payment and a local contact route are conversion work too; see how we build multilingual websites.
For scale: across US industries, 8.18 percent of clicks on search ads converted in 2026, from 3.70 percent in real estate to 15.51 percent in car repair. Your own trend says more than any average. When the problem is too few visitors rather than too few enquiries, start with SEO instead, and when an online shop has traffic but no sales, read why is my online store not selling.
What owners ask before touching their site
How long does CRO take?
The big leaks, such as a slow page, no price or a long form, are fixed in a few weeks. Fine-grained testing is ongoing and only pays with high traffic.
Which tools do I need?
Measurement that counts enquiries, a heatmap or session recording tool once you have enough traffic, and above all five people using your site while you watch.
Should I add a chatbot?
Only if it answers real questions. A window that pops up after three seconds and knows nothing costs more enquiries than it brings.