Conversion Rate

How to A/B Test on a Low Traffic Website

Post Thumbnail

A/B testing is vital during conversion rate optimisation. Evaluating a different approach to some elements of your website will help you increase sales. Optimised user experience will be a huge positive for your visibility in search engines. But can you carry out successful A/B tests when your traffic is low?

An A/B test requires significant input to reach trustworthy results. It is a luxury which start-ups and small companies are not entitled to.

If your website has a low volume of traffic, you are in a chicken-and-egg situation.

Websites that get little-to-zero visits have other problems, too.

Let’s focus on those two points.

Methods which can help you understand why people are not converting

High-traffic websites run A/B tests and decide on a variant that works better. In the case of a low traffic website, this isn’t possible. Why? Because you would either have to wait forever for the arrival of any meaningful data or invest a lot of money from the get-go. I assume none of these options is viable otherwise you wouldn’t be here.

However, hope is not lost. There are ways to learn why people are not converting and to gain traction.

User testing

It’s the easiest and the most accessible method. Ask someone you know to use your website while you observe and take notes. You need to decide what it is that you are testing first. Do you want them to spend a moment on your website and then tell you what they think your company does? Do you want them to try to find certain information without your help? Do you want them to buy a product? Deciding on the purpose of a test makes the exercise more fruitful.

When you feel confident about your website, hire user testers for more customer-centric results.


Many businesses don’t realise the potential of asking their website users about their experience. It’s a great opportunity to hear about struggles as they happen. Additionally, they are a great way to enhance your website copy and expand the range of products or services.

It’s our human nature that we’re much more likely to point out negatives rather than positives. It’s often visible in ratings for fairly “feedback-neutral” places, like airports or post offices which rarely ever get good feedback from the majority of people that had a pleasant experience.

For website testing, this actually works in your favour. You want people to point out the negatives. Sure, some will be very nitpicky, but it’s unlikely a disappointed visitor won’t leave feedback.

Session recordings

Watching sessions gives you an insight into how your website visitors use your website. You can track their cursor movement, scrolling patterns and at what point they have decided to exit the website. Plus, you will see your designs through a fresh pair of eyes.

On top of that, it’s anonymous and secure, as the cursor movement and interactions are recreated rather than truly “recorded”, and you can’t see any sensitive information the recorded user entered during their visit.

Analyse competitors’ websites

It is better to learn from someone’s mistakes. Additionally, they might have done something you haven’t thought of yet. If you don’t have any direct competitors, search for other businesses which target similar audiences.

Be careful of falling into the trap of replicating design or feature choices based on your intuition, though. Your business is unique and what works for others, even if they’re direct competitors, might not work for you. Note it down and carefully analyse why it’s there – or use it as an example when requesting website optimisation from a web agency.

The entire reason for conversion rate optimisation is to ensure your business website is a perfect match for your customers. You don’t want your website to become a blend of solutions that work for your competitors.

Talk to people within your industry

Find people who sold similar types of products or services. If they are willing to share, you will gain access to valuable knowledge and possibly learn about common concerns and objections that your customers might have before purchasing from you.

Add a phone number

Adding an incentive to get in touch might help. Even if you are not big on talking on the phone, it presents a chance to listen to what prospects have to say and questions they ask.

Many business websites feel like they’re hiding the phone number on purpose. Some explain that’s because they’re looking to increase the conversions on the site – but sending a contact form or purchasing directly from the site aren’t the only types of conversions.

Adding a prominent phone number into the mix can bring you a whole lot of customers who wouldn’t buy from you before.

How to measure what is working

Just because your website doesn’t get much traffic, it doesn’t mean you can’t A/B test. Many people hold an opinion that there is no reason to run any experiments on a website with low traffic. We disagree. When done correctly, there is no reason to not test your website in search of a better conversion rate.

Someone from our LinkedIn group “Business Website Owners in the UK” asked, “How much traffic is enough to get CRO right?”. The answer is lawyer-ish: it depends.

  1. Your current conversion rate. The lower conversion rate, the longer it takes to detect an improvement.
  2. Increase in conversion rate that you are trying to detect. A bigger bump doesn’t always mean waiting longer.
  3. How confident you want to be with the results. Sometimes settling for less brings more value.

The short answer is that it can be calculated. We’re using a simplified probability model where we can roughly estimate how confident we are of the result. Let’s take the following example:

Knowing this, we can calculate that the test would have to last about 4 weeks to give significant results with at least 95% of confidence. This means that the long-term results will land very close to a 2% conversion rate – they can be slightly worse, but also better.

Two of these assumptions won’t generally apply to small businesses and startups:

Unfortunately, both mean you’ll need an unreasonable amount of time for A/B tests to work. If you aim for 50% conversion increase instead of 100%, you’re looking at 12 weeks of experiments, so triple the time for an experiment. Cutting the traffic in half on top brings us to a whopping 24 weeks – half a year of testing. Although possible, it significantly devalues the process.

So how do you approach testing on a small website?

Here are several effective tactics that you can use instead.

Always do big changes

Test big and bold ideas. When your website doesn’t get much traffic, it is important to make significant changes to the website and understand what works. See if the changes tackle your users’ objections. Play around with your offer. Focus on bits which are likely to get attention.

Bigger changes usually cause a stronger reaction and these experiments thrive on that. You might even find out that a goal of doubling sales isn’t unrealistic – which might open up the opportunity to thoroughly A/B test your site.

Measure micro conversions

It’s difficult to measure the ultimate conversion goal too. Instead, focus on those small conversions that happen throughout the user journey.

What are the baby steps that a user needs to take before counting as a conversion? The less traffic your website gets, the more you need to optimise for micro conversions.

If you consider ecommerce, the ultimate goal would be to sell. But if your traffic is low, the sample of converters might not be enough to justify testing. Instead, you can focus on the click-through rate of your search engine.

Are the visitors even visiting the product page? Or perhaps they’re struggling at an earlier stage because of a bad search engine? These steps of the user journey will generally have lots of page views, meaning you can test them before you’re able to test the entire funnel.

Test major pages

Following up on the previous example, look for your top landing pages and optimise the heck out of them. Since they get the most portion of your traffic, they will provide you with meaningful results sooner.

It’s okay if they don’t convert at all. You won’t be able to perform a full A/B test but if you can drive a couple of conversions, this will be a sign you’re going in the right direction and long-term, it will build the foundations for a proper A/B test.

Combine pages into one test

If a few landing pages share the same element that needs testing, apply the change to all of them. Then, include them in the same test. Many companies like to create landing pages for every keyword they target or paid campaign they run. We prefer to use fewer landing pages and optimise them as much as possible.

Decide when you declare a winner

The standard threshold for a winning variation in an experiment is 95%. However, you can use a different figure. It directly relates to the question of how confident you want to be with the results. Don’t put on your perfectionist mask because it will get you nowhere. Even in 2020, most marketing decisions are based on intuition so ending a test at 90% or 85% confidence works too.

Temporarily increase the volume of traffic

If you are fine with spending a bit of profit on a paid campaign, use them to drive traffic to the tested pages. The spike in traffic might give you just what you need.


While it may seem to you that you are at a disadvantage, there are several methods that can work. We don’t know a better way to grow start-ups and small companies.

Now that you have a good overview and understanding, go ahead and play with them. Apply a mixture of tools and see what works for your website.

Good luck in increasing conversions.

Originally published Feb 05, 2020 3:28:07 PM, updated October 22 2020.

Google rating Navigation line
Menu Scroll button