A/B Split Testing.

[stevegill]

Hi all,

Has anyone been able to set up A/B split testing with a website that was created with Xara?

I've never tried it myself but now I'd like to.

I'm guessing I would need to have two sites, and have a Javascript script controlling which site is displayed.

Thanks.

= Steve

[Acorn]

Hi all,
Has anyone been able to set up A/B split testing with a website that was created with Xara?
I've never tried it myself but now I'd like to.
I'm guessing I would need to have two sites, and have a Javascript script controlling which site is displayed.
Thanks.
= Steve

Steve are you really after A/B Split testing.
My take is you have a pool of customers that are randomly assigned Path/Site/A or Path/Site/B and you then measure suitable metrics to achieve an impartial understanding of the differences.

Or are you after the optician lens check: "Which is better? The first one or the second one?"
This latter is easy to achieve.
I would treat the two site as a multi-language design: https://www.talkgraphics.com/showthr...-Language-Site.

perhaps the best approach is through Collaboration with Xara Cloud+: https://www.xara.com/webinars/collaboration/.
You can publish a site in Cloud and the customer can comment live to your live changes or you can accept comments added directly to the design.

Acorn

[stevegill]

Steve are you really after A/B Split testing.
My take is you have a pool of customers that are randomly assigned Path/Site/A or Path/Site/B and you then measure suitable metrics to achieve an impartial understanding of the differences.

Or are you after the optician lens check: "Which is better? The first one or the second one?"
This latter is easy to achieve.
I would treat the two site as a multi-language design: https://www.talkgraphics.com/showthr...-Language-Site.

perhaps the best approach is through Collaboration with Xara Cloud+: https://www.xara.com/webinars/collaboration/.
You can publish a site in Cloud and the customer can comment live to your live changes or you can accept comments added directly to the design.

Acorn

Hi Acorn,

The way I understand A/B Testing is it's about increasing conversions. You develop a theory of what could increase conversions and then test one at a time. Say, for example, you may have a red button for your CTA (Call to Action) but think a green button may work better, so you test that theory. If green works better then that becomes the control. Then you try another colour and so on.

Then you might try different images, headlines, page copy, etc.

= Steve

[Acorn]

Hi Acorn,

The way I understand A/B Testing is it's about increasing conversions. You develop a theory of what could increase conversions and then test one at a time. Say, for example, you may have a red button for your CTA (Call to Action) but think a green button may work better, so you test that theory. If green works better then that becomes the control. Then you try another colour and so on.

Then you might try different images, headlines, page copy, etc.

= Steve

Only asked as it's a long haul.
What you appear to be after is a random selection of your visitors to be given a token that thereafter directs them to the same site, A or B.
This can be set in Local Storage.

Google would not like two near identical versions so you would have to code at the page level, forcing an automatic switch.
I could achieve this at component level if there's no major size change.

Acorn

[RKissane]

Steve,

A/B testing really came into being for Email marketing campaigns because it is much easier to configure and control.

The big thing to consider is whether the site has enough traffic to support even doing A/B testing. Google only likes to have the testing done for short period of time and that means you need to have enough traffic to support get any meaningful results.

Now the testing will be done on a landing page that you want to see if changes make a difference on user interaction. But you may find that it is more useful to spend your time working on how you get users to the landing pages. Then once you have a steady larger user base, you then could look at A/B testing to see if site changes will make a better conversion rate.

There are some programs out there that you enter their code snippet into your site and they will do the analytics and redirecting to your variant. Google has one and another one I have looked at in the past is Optimizely. I have never done any type of A/B testing because the sites I work with do not have the traffic to support this kind of testing.

Let us know how you make out on this.

Ray

[stevegill]

Thanks Acorn in Ray,

The site has thousands of visitors per month. I came across the Google A/B Testing and I'm going to give that a try.

= Steve