You want more visitors to your website to request a quote, call you or place an order. That is called conversion: the share of your visitors who take the action you want. A well-known way to improve it is the A/B test. You show half of your visitors version A and the other half version B. Then you look at which version works better.
There is just one problem. An A/B test needs a lot of visitors to give a reliable result. Most small and medium-sized business websites do not get that many. That does not mean you have to guess. In this article you will read how to keep improving based on facts, even with little traffic.
Why an A/B test does not work with low traffic
An A/B test compares two versions. The difference between those versions is often small. To know whether that difference is real and not down to chance, you need enough measurements. This is called statistical significance.
A rule of thumb: you often need at least a few hundred conversions per version. Exactly how many depends on the difference you want to detect. The smaller the difference, the more data you need.
If you stop a test too early, you may pick a winner that is not actually a winner. You think you have learned something, but mostly you have been lucky. That is more dangerous than not testing at all, because you are wrongly confident that you are right.
Do you receive a few dozen enquiries a month? Then a classic A/B test is usually not the right route. You need other sources.
Watch real visitors
Numbers tell you what is happening. They do not tell you why. That is why you start by looking at behaviour. Two tools help with this:
- Session recordings: short videos of a visit. You can see how someone scrolls, clicks, hesitates and leaves.
- Heatmaps: colour maps that show where visitors click and how far they scroll on a page.
You do not need large numbers for this. Ten to twenty recordings of one important page often reveal patterns already. That is frequently more than a month of figures will tell you.
Watch for signals like these:
- Visitors click on something that is not a link, such as an image or a piece of text.
- People scroll past your most important button without stopping.
- Visitors start a form and give up at a particular field.
- On mobile, people look for the menu or for your phone number.
Keep a record of what you see. If the same problem appears in several recordings, you have a strong indication. Do not forget privacy either. Make sure your tool does not store personal data such as completed form fields, and mention its use in your privacy statement and cookie notice.
Ask your customers
The best source of ideas is often not a tool. It is the people you have already convinced. They know why they chose you. And they know what almost held them back.
Ask a short question at a suitable moment, for example on the thank-you page after an enquiry or in the confirmation email. An example:
- What was the main reason for getting in touch?
- What almost stopped you?
- What information was missing on the website?
Keep it to one or two questions. The shorter it is, the more answers you will get. An open field works better than multiple-choice buttons, because people use their own words. You can use those words later in your copy.
Talk to your colleagues in sales or customer service too. They hear the same questions and objections every week. Is it the delivery time, the price breakdown or the question of what a first meeting looks like? Then the answer belongs on your website as well.
Change one thing and measure
Have you gathered a few indications? Then choose one important element to change. For example the headline at the top of the page, the text on the button or the number of fields in your form.
Why just one thing? Because otherwise you will not know what caused the effect. If you change the headline, image and form at the same time, you will not know afterwards which change worked.
Then compare the conversion rate over a few weeks with the period before. This is called a before-and-after comparison. It is not perfect. But it is far better than doing nothing.
Do this as carefully as possible:
- Write down beforehand what you expect and which figure you will track.
- Note the date of the change.
- Measure for at least four weeks, or longer if you have little traffic.
- Do not check the result every day. That leads to premature conclusions.
How to make a before-and-after comparison more reliable
A before-and-after comparison has weak points. The season changes. You launch a campaign. A competitor lowers its rate. All of these things affect your figures, regardless of your change.
You can reduce the risk:
- Compare like with like. Put the four weeks after the change next to the four weeks before, and if possible also look at the same period last year.
- Split by source. Look separately at visitors from Google Ads, SEO and email. This shows whether a shift in traffic is distorting your figure.
- Use an intermediate step. Do you have too few enquiries to see anything? Then also measure a micro-conversion. This is a smaller action on the way to the real conversion, such as a click on the phone button or starting the form.
- Combine with qualitative signals. Do you see in the recordings that fewer visitors give up? Then that is an extra indication that the change works.
You will rarely be certain. But you can get better and better at justifying why you make a choice. In practice, that is already a big step.
An example from practice
Suppose a garden landscaping company has a website with a quote form of eight fields. It receives around twenty enquiries a month. An A/B test is not realistic in that case.
The owner watches fifteen session recordings of the quote page. He sees that many visitors stop at the budget field. Some fill it in, delete it and leave. On the thank-you page he also asks new enquirers a question: what almost stopped you? Several people answer that they did not want to talk about money without knowing what a garden costs.
He makes one change. The budget field becomes optional, with a short explanation of why that question helps. He also adds a paragraph about what the cost of a garden depends on.
He notes the date and tracks the number of started and submitted forms for six weeks. He compares that with the six weeks before. He also checks whether the distribution of traffic across sources stayed the same. So he has no proof, but he does have a well-founded indication of whether the change worked. Then he moves on to the next element.
Frequently asked questions
What is the minimum number of visitors I need to test?
There is no fixed number. For an A/B test you often need a few hundred conversions per version. If you are far below that, choose recordings, surveys and before-and-after comparisons. These also work with small numbers.
How long should I wait before drawing conclusions?
Measure for at least four weeks, and preferably several weeks more if you have few enquiries. That way you capture fluctuations within the week and the month. Do not draw a conclusion based on a few days.
What if conversion drops after my change?
First check whether something else changed, such as your ads or the season. If the drop persists, you can restore the old version. You will still have learned something, namely that this approach does not work for your visitors.
Book a free 30-minute introductory meeting
Do you want to know where your website has the most to gain, even if you have few visitors? Book a free 30-minute introductory meeting with Adventio Marketing in Amersfoort. Together we will look at your website, your figures and your goals. You will get concrete ideas to start with, with no obligation.
