Follow us: Twitter Behance Linked In Facebook

iD30 Blog

All the latest news, events and projects from iD30...

How, where and why to use A/B testing

A/B testing

A/B testing is a method of improving how effective a webpage’s design is at converting visitors into customers through side by side performance comparisons of two variations of the webpage.

A/b testing split testing illustration title image

How?

To carry out an A/B test you need to take one of your webpages, which is the “A” part, and tweak one aspect of it, which becomes the “B” part.

You could change the:

- Colour

- Content

- Layout

- Size

- Format

…of the:

- Call to action

- Title

- Body

- Image

- Video

Once you have your A and B you then show each one equally to visitors and collect data on which one is most effective at achieving the goal of the page, which is measured as a percentage and reffered to as conversion rate. The more successful variation will then become your new permanent page and hopefully more people will stay on the page long enough to hear what you have to say.

Rinse and repeat

You can A/B test more than once, if the first test didn’t show the scale of improvement you aimed for then you can continue to add a B variation to continuously improve the conversion rate of the page; it’s good practice to A/B test every now and then even if your page is performing well, because you don’t really know if the page could be performing even better until you try and see.

Where?

Most commonly A/B testing takes place on the landing page of a website, though it can also be beneficial anywhere else on the website where the page has a goal.

For example a blog may have a goal of signing people up for an email newsletter, so the layout of a the blog page could be A/B tested quite easily, but the content of the blog itself is not as easy to A/B test because of the sheer volume of information and number of variations for that content.

Something to note: when creating your “B” you should try to change only one aspect of the original “A” at a time. If you change too many you won’t be able to identify which change caused the improved or diminished results.

If you use pay per click (PPC) you can A/B test the content of the adverts you display. The success of adverts is usually measured by impressions and clicks, which as a ratio is called click through rate, this is what you must aim to increase through A/B testing.

To carry out an accurate A/B test on PPC adverts, you will have to keep who the advert appears to (keywords, location etcetera) the same on both and ensure that the rotation of the adverts is equal and not which is predicted to be best being shown more (I swear this is an option, it just has a catchier name).

Why?

There is always room for improvement in web design, no page is perfect, some come close but there will still be some people who do not like the page they see enough to want to spend their time inspecting it, especially when they can just go to a competitor who might have a page that they like more. And with people’s preferences changing all the time, the best way to adapt is by using A/B testing regularly.

If you have access to both analytics for your landing page and the PPC service your business is using, you can combine the data to increase the power of A/B testing.

Starting from when a user sees your advert; you can determine if “A” or “B” has had more clicks to your website, you can then use analytics on your landing page to test which of the two adverts has caused more conversions, which might not be the advert with the most clicks.

Using this method will save you money by matching expectations created by the advert to the information on the landing page.

For example if you have an advert which promises gold for free you are going to get a lot of clicks (Woo!), but if your landing page is trying to sell paper your visitors will click and leave (Not so woo), which means you pay the advertising company for nothing, so, in a way you kind of do give gold away for free, and it doesn’t take me to tell you that it is not a good business plan to have.

Obviously it is very rare that the advert and the landing page will be this unaligned, but the more aligned they are, the less money you waste on clicks that don’t convert.

Check out this infographic from Quicksprout for a guide to the process of A/B testing and ideas for what to tweak in your “B” test.

Why not Contact us if you need any help carrying out an A/B test.

STAY UP TO DATE...

Get the latest goodies and special treats from the iD30 camp