The experiment will be treated by the Google algorithm as a new campaign and it will also learn the experiment for a period of 7-14 days. Ending the experiment during this time may result in receiving unreliable data, based on which we would draw conclusions and make decisions as to the effectiveness of changes introduced in Google experiments. ‘ of the campaign budget allocated to the experiment and the percentage of auctions in which the experiment can participate. To be able to reliably test modifications, it is recommended that the budget and auction split be 50% for the original campaign and 50% for the experiment.
In the case of text campaigns
There is an additional section below – ‘Advanced options’ . It is worth looking into it and choosing whether the separation of the experimental part is to be based on searches or cookies. By default, Google will select the ‘Cookie based’ option for us. Separating the experimental portions based on cookies ensures that users will only see Kiribati Email List one version of your campaign, no matter how many times they search. Google suggests this version by default as it will keep other factors from affecting the results, the experiment. Will get more accurate data. If you choose to split based on cookies. You must ensure that your audience lists have at least 10,000 users per list.
If you have fewer users in your lists
The results may be less accurate. If your site doesn’t generate a lot of traffic, I suggest that you separate the experimental part based on searches. Selecting this option randomly assigns users to your experiment or original campaign each time CL Lists they perform a search. This way you can get statistically significant results faster than cookie-based splitting. For Display Network campaigns, cookie-only breakdown is used. 5. After completing all the information, click on the blue ‘Save’ button. The experiment continues – be patient! We can easily distinguish an active experiment from standard campaigns.
