Trying to understand the user’s behavior and increasing the conversion rate is really a time-consuming and difficult task (a pain in the a**). Remember that we are in the business of “making money for the client or ourselves”; we are not on the business of “I was right no matter what”.
So, follow your data not your ego to avoid these top 5 mistakes on your CRO projects.
It’s just too easy to have a pre-concept of “why this page is not converting” you can just have a look on the page and say it’s because is awful, it’s because it’s too big or small or is that paragraph and that element or wherever.
The reality is that you don’t know what is wrong because you haven’t analyzed the data yet, in this scenario you are not doing any CRO, you are just guessing based on pre-concepts that may be or not be right.
The only results that you are going to have for this scenario is a bunch of losing A/B test because again you are just guessing.
So, my advice is really simple, don ́t guess, take your time to analyze the data and create a CRO strategy.
The rushing is the twin brother of the guessing, they normally come together, you rush to get results so you start to guess instead of analyzing.
You should avoid that and follow your plan, don’t ́t stop your A/B test until you have the required number of sessions and the possibility to be the best on your side.
Pausing your experiments early is going to generate a bunch of false-positives. As a result of this, the changes on the page are not going to improve the conversion rate
Original vs Variation 1
A really common mistake at the moment of setting up your A/B test is actually creating just a “Variant 1” vs the “original version”, this looks ok but is not.
When you set up a variant, the code and changes are going to be present on that variant, so the loading time of the page can be a bit slower so if you compare this to the original version your variant is always going to be slower and the page speed is really a big factor, you can lose around 7% to 12% of conversión for every 100 milliseconds of loading time.
The right way is creating 2 variants, you can call the first one “Control” and leave this without modifications and a second one called “Variant 1” you can apply your modifications on this variant; divide your traffic 50% each between these two variants and assign a 0% traffic to the “original version”. Using this method you are going to have equal conditions for your A/B test.
Breaking your page
Testing your A/B test variations is mandatory before going live,
and unfortunately, it happened to me once, believe me, coming back to the office on the Saturday of a long weekend is no fun at all.
It’s really simple, before you go live with your A/B test go to “Interactive Mode” this mode allows you to interact with the Variant page before going live, you should test that everything is working well in different devices, test the forms, the check-out, etc.
Another good way is going live with your experiment for an exclusive IP address, you can use your office IP, go live, test everything and when you are sure that the page is working well, then stop that experiment and finally put the test live.
Try to be always right
As I mentioned in the introduction, one of the golden rules of CRO is actually acknowledge that you are not going to win all the A/B tests and this is a mathematical fact.
Leave your ego out of this and follow the numbers, they normally show you the right way. Integrity and honesty are essential in this game because data is easy to manipulate to show something else.
I hope you enjoy this reading and let me know your thoughts about this top 5 mistakes on CRO.