Don’ts in Experimentation (A/B Testing, Multivariate, Personalization) while planning your testing Strategy aligned to digital strategy

“Search engines loves experimentation” 

A. I will love to start with don’t while you guys are experimenting on pages/ forms aligned to your  digital strategy idea we should avoid somethings as they will never give us any outcome & will only consume time & false expectation in the mind of the visitors coming to your web sites

1.       We all agree that it’s time to serve local content to local audience as it really help them know how locally you have become from being global that company looks.

2.       While Back-link strategy is liked by google from the trusted sites global page should try to avoid giving links to local country pages as it simply tells google to give them weight-age on a google.com console where you always want your global page to appear first than the local versions 

3.       Hence the first thing to avoid in testing give back links to local country pages from the global page

4.       So here is the real story:

Just look at the Variation below that will shown to audiences from china (Personalization Experiment) giving back-link from the global page to a local page can help move local page above the global page in GOOGLE.COM console which is really not we want to happen at all at any given circumstance

                                                                        

 Even after 9 days of running the exp & showing the variation to local audience we are still seeing only 4 visits of out 81 visits clicks on the CTA & engagement rate of the traffic https://www.xxxx.com/blogs/xxxx/cn-cn of the traffic clicking going from the global page to local page is down by -1.23% which clearly tells that on the global local content should not be mixed 

Click on the image to enlarge it

Personalization of these kinds should better be done using ECM likes sitecore, drupal, SharePoint from the server side itself any time  visitors  from a local country lands on a global page from the local page as it tells the search engines father & child relationship on localization journey 

Click on the image to enlarge it

B. Swapping page band by moving the band further up the fold without doing any changes on the page content 

Assumption made here is : Moving Client Success below “Featured Software & Services” section will help visitors gain visibility to Client Success section will uplift engagement

Position of the Client success has been changed without making any changes on the content of the section 

Click on the image to enlarge it

See what happens next?

                                                       

In next than 5 working days the results comes negative as engagement rate of the page further falls on the variation drastically. Hence make some changes before moving any section of up/down before testing them

Click on the image to enlarge it

C. Now what I am write here is like committing a crime which will make your digital strategy really not getting best outcome for sure, so I will say don’t try this at all while doing testing 

I do agree to fact that it is not easy to get organic traffic to your industry specific pages as the keywords targeted are not having high search volume at all hence you need to plan some merchandising plan for it to drive traffic to these pages ie, leveraging owned media or align little paid media to these pages to really start giving visibility to these pages

But if you are only using organic traffic for Industry page & you try personalization using a cookies pool which personalizes the experience depending upon the past visit even you run the experiment for an year you will not even see 1% statistical significance reached by your experiment

So what I mean to say is Personalization works only when it is supports by a good amount of traffic flow else it is waste of time & effort

Click on the image to enlarge it

 

So below is unique kind of personalization where on the basis of cookie pool if any visitors qualifies ( (Health care NOT FOR TARGETING) done using DMP will show up the variation page

So this experiment is getting traffic only from organic search & hence even after running the experiment for one year which is wow time.

Even we are seeing the improvements in idea implemented it has not even reached 1% stats significance hence we still can’t say that the idea works or not conclusively. Refer the results shown below 

Click on the image to enlarge it

Note: All the results shown can be seen inside the Testing tools like Google Optimize/Adobe Target/ Optimizely ( All are Industry leaders in web testing/ Personalization)