We would like to test a specific element across multiple pages ( but there may be a catch)
We would like to conduct a simple test in our Ecommerce site
The hypothisis is that if we change the wording from "Add to cart " to "BUY NOW" that this stronger call to action will increase revenue.
The issue we are having is applying this change site wide to all current buttons that say "add to cart"
we have read all the instruction on URL targeting but still cannot seem to get it to work.......
One possibility that may be an issue is how our URL structure works ( and this is the possibile catch)
our category pages are as follows
and our Product pages are as follows
BUT .....the add to cart buttons only appear on the Product pages and NOT the category pages.
In URL Targeting we have selected http://www.qispackaging.com.au/Product/ and the match validator say that it works ...
any help/ideas/suggestions/ would be greatly appreciated
Thanks Guys !!
Solved! Go to Solution.
I assume that you got it working because when I visit http://www.qispackaging.com.au/Product/paper-bags-coloured/coloured-bags-with-paper-twist-handle.asp... I am actually seeing the "Buy Now" CTA .
However, although it works, the targeting set-up is not ideal. The reason being that a substring match to http://www.qispackaging.com.au/Product/ includes pages that don't have the "Add to cart" button, thus inflating your visitor count (the experiment is running on pages it doesn't need to). This will skew your results.
Here's how I suggest setting up targeting:
Use a Regular Expression Match: .*qispackaging.com.au\/Product\/.*\/.*
This will ensure that the experiment runs on any page that follows the URL structure:[anything]ispackaging.com.au/Product/[anything]/[anything]. But, it will not run on the category pages due to the missing content detailing the specific product (the final /[anything] ). For additional help on Regular Expressions, I like to use http://rubular.com/
Product page example:
Category page example:
Let me know if you have any questions!
It looks good to me! I double checked your targeting and confirmed it's running on the correct pages. The button text also changes on the "read more" pages as well which seems correct.
However, there is at least one URL that doesn't follow the structure as outlined above: http://www.qispackaging.com.au/Product/Heat-Sealers.aspx. I recommend just adding this URL manually to the targeting as in the screenshot below. You can do the same if there are any others like this.
If you have dev resources or can add snippets of code to your page, you should check out the custom tag targeting. This will make it easy for you to isolate page types.
Since you're new to Optimizely and A/B testing, I strongly recommend the Academy as a place to start. There are some wonderful tutorials and hands-on activities.
Just a quick question , out of all the categories that we have , how did you find that particular one was not working ??
Regex and targetting can often be a bit of a minefield.
I always find it useful to pull out a report from Google Analytics with all traffic'd pages on the site and check my rules against that list just to be sure we are not targetting the wrong pages as this can often skew your results.
We're having a similar problem. Can you please explain how including pages that don't have the element we're testing will skew results? Can we still run the test and trust the results?
We don't have a lot of experience writing regular expressions. We're hoping to launch the test in next few hours.
@Gina123 - It will skew your results because your visitor count for the experiment will increase every time a unique visitor hits a page targeted by your experiment. So, if your visitors are landing on pages that do not have the element, then the visitor count increases, but the conversion rate will not increase since the visitor is not technically seeing any changes to the page. Does that make sense?
I'm sure that I can help point you in the right direction for regular expressions or alternate targeting structure. Can you post an example of the pages for which you are hoping to run an experiment?
Hi @Gina123 , as long as you recognize that unique visitors will be high, then you are okay. You are right in that the effect will be the equal for all variations. Let me know if you have additional questions
I would suggest that this is a much riskier way of testing and I certainly would not recommend taking this approach without knowing what traffic levels you get and how that traffic is distributed across the pages that will be affected by this test.
I do feel that custom targetting is one of the areas that is lacking in Optimizley and I often look at ways I can get around it on a test by test basis.
My main problem was that the element I'm testing only appears on some of the pages I'm URL targeting. It would be difficult to identify all of the regular expressions where the element does and does not appear.
Optimizely helped me focus the test using Conditional Activation. So now the test only tracks visits to pages where the element-in-question appears. Problem solved!