A/B testing on your website
With A/B testing you let your visitors choose which version of a page brings in more quote requests. That way you decide on numbers instead of taste.
100+ businesses and growing
Marketingkenners gebruikt A/B testen om te bepalen welke versie van je pagina meer offerteaanvragen oplevert. Per test verandert één ding, zoals de kop, de knoptekst of de lengte van het formulier. Het verkeer gaat gelijk over beide versies, de test loopt in hele weken en we meten op aanvragen en telefoontjes, niet op klikken.
What is A/B testing?
A/B testing is a method where you show two versions of a page at the same time to different visitors. Half see version A, the other half version B. The version that brings in more enquiries wins and stays.
The great thing is that you no longer have to guess. Whether a shorter headline works, whether reviews higher up the page help, or whether a call button beats a form: your visitors show you with their behaviour.
A/B testing is one of the fixed parts of our conversion optimisation. We always measure what counts for you, so enquiries and phone calls. Not clicks.
Two versions, one winner
We change one thing at a time and measure what delivers more enquiries. Switch between the original, the variant and the heatmap.
The original: a generic button and a long form. Visitors drop off before they send anything.
What you test on a tradesperson’s page
We never test everything at once. Per test, one thing changes, so you know what makes the difference.
The headline and promise
What is at the top? A concrete result often works better than a generic slogan.
The button text
‘Request a quote’ or ‘Book a free inspection’. Small words, sometimes a big difference.
The length of the form
Fewer fields usually bring in more enquiries. Sometimes an extra question actually helps filter out the right jobs.
The position of reviews
Your Google reviews straight in view, or further down. We measure what builds more trust.
Project photos
Photos of your own work versus generic images. Real jobs almost always convince better.
Calling or a form
For big jobs a customer sometimes prefers to call. We test which form of contact brings in the most.
How many visitors do you need for an A/B test?
The more visitors and enquiries, the faster you get a reliable result. A result is reliable, or statistically significant, when the difference is big enough not to be down to chance.
Many tradesperson sites don’t have tens of thousands of visitors a month. In that case we test bigger differences, such as a completely new page structure, or we let a test run longer. If there is really too little traffic, we say so honestly.
In that case we work with heatmaps and session recordings. They show where people get stuck, even with little traffic. We then make targeted improvements and measure the effect on your enquiries.
Define the goal and measuring point
We agree on what counts, for example a submitted quote form or a click on the call button. With good tracking we measure that reliably.
Draw up a hypothesis
Based on data we write down what we expect. For example: if the reviews are higher up, more people request a quote.
Build the variant
We make version B, with one clear change. The rest of the page stays the same.
Split fairly and let it run
Traffic is split evenly across both versions. We test in whole weeks, so a busy Monday or quiet weekend doesn’t skew the result.
Analyse and implement
If B wins, we put it live. If B loses, you still learn something. You get a short report in plain language.
You recognise a good test by this
- One thing changes per test
- You measure on enquiries, not clicks
- The hypothesis is on paper beforehand
- The test runs in whole weeks
- You look at desktop and mobile separately
- You only stop when the result is reliable
Adjusting on gut feeling or testing on numbers
Adjusting on gut feeling
- The taste of the owner or designer decides
- You don’t know whether it works better or worse
- A wrong choice quietly costs you enquiries
- Every discussion starts over
A/B testing
- Your visitors decide with their behaviour
- You see exactly what a change delivers
- A bad variant disappears after the test
- Every test makes your site a little better
A/B testing within the Growth System
A test never stands alone. What we learn, we also use in your ads and other pages. If a certain promise works well on your service page, we carry it over into your Google Ads.
We discuss the results in the monthly meeting with your fixed point of contact. You see which tests ran, what they delivered and what we want to test next. That way your site grows month after month, within the complete Growth System.
Frequently asked questions
Don't see your question here? Call, WhatsApp or email us. You'll get an answer within one working day.
What is the difference between A/B testing and just changing something?
With a regular change you replace the old version and hope it works better. With an A/B test both versions run at the same time, so you know for sure which brings in more enquiries. Season or busy periods then play no part in the result.
How long does an A/B test take?
That depends on your traffic and the number of enquiries. A test runs for at least one or two full weeks, and with little traffic sometimes longer. We only stop when the result is reliable, not when it looks good after three days.
Can you A/B test on mobile and desktop?
Yes, and we always look at the result per device. People behave differently on their phone than at a laptop. What works on desktop can cost you enquiries on mobile.
Does A/B testing make sense if I have few visitors?
Sometimes not, and we say so honestly. With little traffic, a reliable result takes very long. Then we get more out of heatmaps, session recordings and targeted improvements.
Will I be involved in the tests?
Yes. We discuss in advance what we want to test and why, and you know your customers best. After each test you get a short report in plain language, and in the monthly meeting we look at the next step together.
Where do your visitors drop off?
A specialist looks with you for 30 minutes at where your site is leaking. Free, over video.
Marketingkennersonline · reply within one working day