You have visitors on your site, but the results are lacking. Quote requests are lagging behind, or the webshop sells less than the traffic suggests. That quickly leads you to conversion optimization. Before you start doing that, there is a question that is almost never asked, and it determines whether it will work for you.
What conversion optimization actually is
Conversion optimization means getting more out of the visitors you already have, instead of buying more visitors. You change something on your website, and you see if a larger portion of those same visitors does what you want.
That “what you want” is not always a sale. For a service provider, the conversion is usually an inquiry: a completed form or a phone call. That may seem like a detail, but it determines everything that follows, because you can only improve something once you know what you are measuring.
More traffic costs money every single month. A website that works better keeps working better.
The condition that is rarely mentioned
The standard way is the A/B test. You create two versions of a page, send half of your visitors to one and half to the other, and see which one does better.
There is a condition attached: you need enough visitors and enough conversions per version to be able to demonstrate a difference. If the scale of such a test is underestimated, the analysis cannot demonstrate a real improvement, even if you do everything else right. That is not our opinion; that is the conclusion of research into precisely this question: Zhou, Lu, and Shallah, All about sample-size calculations for A/B testing (2023).
Pay attention to how that works. The danger is not that the test invents something, but that it finds nothing. You actually improve your page, the test fails to prove it, and you throw the improvement away.
A fixed number of visitors does not exist. It depends on how small the difference is that you want to be able to see. You can prove a clearly better performing page this way. An improvement of a fraction of a percent requires significantly more traffic, and is therefore reserved for sites with a huge number of visitors.
Are you actually measuring the right thing?
Before you even get to testing, there is a step that goes wrong more often: is your conversion measurement correct. Is a submitted form actually being registered, including on a mobile phone. Is the correct action set as the primary one. And does it still work after the latest website update.
As long as that measurement is not reliable, every outcome of every test is a guess with a little graph around it. Yet this is the cheapest part of the whole story: an afternoon of checking, and everything that follows becomes usable.
What you do if you don't reach those numbers
Suppose your site has too little traffic to test. Then the work shifts from testing to watching.
An A/B test is meant to choose between two versions that are both reasonable. You don't need a test to fix something that is demonstrably broken. If your form partially falls off the screen on a phone, that is not a hypothesis, but a bug.
What yields the most during low traffic:
- See where people drop off, so you know which page deserves attention.
- Collect the questions you keep getting on the phone. Every question that comes up frequently should have been answered by your website.
- Browse through your own site on your visitors' devices, and try to do what you ask of them.
- Check if the page answers the question the user came in with.
When it is actually worth it right away
There is a situation where you don't have to hesitate. You structurally have a lot of traffic, there is one clear action that you measure and that measurement is correct, and your conversion is lagging behind what that traffic leads you to expect.
If you pay for that traffic, every improvement counts double. You are already paying for those visitors, regardless of what they do. A site that converts a larger portion of them thereby lowers your cost per lead.
Start with your measurement, not your pages. See if what you are counting is also what you want to know, and if it still works. With that basis, you immediately know how many conversions you have per month, and that number determines whether testing makes sense for you or is an expensive detour.
If it turns out you have enough traffic, we are happy to take a look with you to see where the profit lies.
