Rich-text Reply

Using test cookie at draft experiment possible

Mr-BS 05-22-16

Using test cookie at draft experiment possible


following the QA steps, 1. Previewmode, 2. using force parameters and 3. using test cookie.

Ist it right that I can test an Experiment in "draft mode" with the force parameter (when privacy settings are set) but for step 3 using test cookie the Experiment must be started?


And is it right that it´s only with the cookie possible to test an Experiment over multiple pages?

By testing with the query parameter I have to set this one after every pagerload new, right?


Level 2

JasonDahlin 05-23-16

Re: Using test cookie at draft experiment possible

Hi @sRBjoern,



1- Use the force parameter to make the experiment run on whatever page you are on, regardless of whether it meets URL or Audience Targeting criteria.  This is great for when you are building the experiment and for when you want to easily switch from one variation into the other.


2- Use a test cookie for when you want to set the experiment live but not actually receive any real traffic.  This is great for when you cannot or don't want to use the force parameter and when you want to check the URL targeting.

--Jason Dahlin
Analytics and Testing Guru Smiley Happy
DavidS 05-24-16

Re: Using test cookie at draft experiment possible

Hi sRBjoern,


Thanks for reaching out to Optimizely!

On top of what Jason said, if you choose to use a test cookie, don't forget to reset your results before setting your experiment live for the world to see so you reset your own test visitors and conversions. 


The test cookie also allows you to check that your goals are firing correctly. You can find out more about it via the following KB article.


Feel more than free to reply back with any additional questions at all.




Mr-BS 05-25-16

Re: Using test cookie at draft experiment possible

[ Edited ]

Hey David,

thanks for the hint to reset the results.

But after that, useres who are bucket in a variation are still in it.

So If I want start from zero it´s usefull to pause the experiment and duplicat it and start the duplicated, right?


On the other hand duplicating is unnecessary when I´m adding the IPs to the privacy seetings.


Hmmm, just think about whats best practice.
Will be nice to do it without duplicating the experiment but I think there is no chance to end-bucket an user?

Level 2