On what basis we can split the traffic in drafts and experiment campaigns.

How to split the traffic in drafts and experiment campaigns

I guess you want to experiment with a new campaign in your Google ads account.

While setting up, you have come across an experiment split option.

Now, you are probably wondering which value to enter to maximize the results.

This might be one of the most asked questions about experimenting with new campaigns.

In this article, we are going to cover how to split the traffic in drafts and experiment campaigns and compare the search query splits vs cookie splits.

So, regardless of your business type, you should be able to find the best ratio of experiment and original campaigns for your business.

How experiment split works?

In a short answer, you are splitting the budget or traffic of the original campaign between the experiment and the original campaign.

And the value you are entering means how many percentages of the total budget is allocated to test a new campaign.

So, you aren’t spending any extra money on top of your budget.

You are just sharing it with your existing campaign.

As a result, your test campaign can also participate in the portion of all the auctions and can receive the same percentage of traffic proportionally.

NOTE: keep in mind that, experimenting only works for search and display networks.

Besides, your split percentage doesn’t mean you are going to receive the same amount of impressions in the google search results. That number can be more or less.

Percentages of the budget that can be allocated for experiments

When creating your test campaigns, under the Drafts and experiments, you have the Experiment split option for both Campaign experiment and Ad variation.  

50 percent split option chosen

Let’s see, how much this value should be to meet your expectations.

Here, we have listed some splitting options, and when you can implement them.

10 percent or less

  • To reduce the risk. When you are testing a completely new campaign without knowing what is going to happen. And you want to reduce the risk of that campaign to affect your total budget, you can use this method to split the traffic or budget.
  • Using many alterations. If you make a lot of changes, rather than one change at a time, results may be dramatically good or bad.
  • Need higher volume. To choose this option, you need to bring enough traffic volume for testing. As you can see, in this way experimented campaign receives 9 times less traffic than the original campaign. Therefore, you should use this with campaigns with more traffic volume. 
  • Takes much longer to test. When you use this percentage, you should be willing to wait for a longer period to get some accurate results. Because there will be very little data to analyze and make a decision compared to an average campaign.

10 to 50 percent

  • Fewer alterations. You can use this range when you have a little more confidence in your new campaign. Mostly because you made fewer changes that may not affect the results dramatically.
  • A longer time than the average. Even if it gives results faster than 10 percent or less split, it still requires a little more time than the average to generate enough data.

50/50 splits (50 to 50 percent)

This strategy is very famous and used by many advertisers. It has slightly different features than the other remaining splitting options.   

So, let’s see, what the 50/50 traffic split means in experiment campaign.

  • The total budget is shared equally. Half of the total budget of the original campaign is used for experimentation.
  • Easy to calculate. As both campaigns can receive an equal amount of traffic, you can compare their metrics directly by putting them on the same table. For example, cost per conversion (CPC), click-through rate (CTR), etc.
  • Brings results faster. As it receives more traffic, it can easily generate enough data within 3 to 4 weeks depending on your traffic volume. 
  • Results can be accurate. More traffic and faster testing means, for most of the campaigns your report can be accurate
  • Riskier. As you allocate more of your budget for testing, most of the time you want to avoid very risky ideas in which you don’t have confidence at all.

50 to 100 percent

  • For reliable experiments. This split value is mostly used when you have more confidence in your system. You might tweak very few things in certain areas. Then put it into A/B testing, if in case something goes wrong. 
  • Used for gradual increases. It can be a part of the plan for some advertisers. They can increase the share of the experiment campaign as it becomes more and more reliable. For example, they may start with a 10 percent share of the total budget, which is a 10/90 value. As they gain more conviction, they can test it with a 25/75 percent split value and over time, that number could grow to 70/30 or more.
  • Shorter time. If you split the traffic or budget with this method, you can make tests in a much shorter period than other strategies.   
  • Better with lower volume. You don’t have to have a high volume of traffic to test your Google ads. It can work greatly with slightly lower volume campaigns.

Which experiment split option to choose?

When creating your campaign experiments, you can see the Advanced options under the Experiment split.

If you click that, there are two types of split options to choose from.

Now, we are going to compare the Search-based vs Cookie-based experiment splits.


Search based split

This split option shows either your experiment or original campaign when someone sees your ad.


  • More traffic. If someone browses the internet more than one time, she can see two variations of your ads. That means one person can see both of your ads. So, you may have 100 unique users seeing your ad, but your ad can be seen 150 times by them. (PIC)From that, we can conclude that most probably, your ads going to be seen more than the number of users and in this way, you can accumulate more data to analyze in the same time frame than the cookie-based split.   
  • Faster result. Statements given above can bring another outcome. More data with the same amount of users bring statistically significant results faster than other methods.


  • Not accurate. It is not as accurate as of the Cookie-based split option. Because it is showing your ads randomly rather than making them personalized.
  • More expensive test. Generating more statistical numbers for the same number of users makes the cost per visitor experiment more expensive.

Cookie-based split

In short, it means users browsing the internet can only see one type of your ad depending on their interests.


  • More accurate.  The cookie-based experiment split option as the name suggests shows a more personalized version of your Google ads. That means it can be a little more accurate. Why I am saying little because it only shows one better ad from the two options.        
  • Chipper Test. It might be chipper compared to search based split option because it shows your ads not more than the number of users. So, your cost per visitor/ experiment per user is slightly cheaper here. For example, 100 people can see your ad 100 times, not more not less.


  • Takes more time.  Because it shows one version of your ad to one person, you have to wait a little bit more to get enough number of users for your ads. 
  • Less Traffic. The cookie-based split option works with the one person one type of advertisement approach. So, you generate less data with the same time interval than the search-based approach.       

Notes: Experiment split options works with only search network campaigns for now.

For the display network, there is a cookie-based approach used.


Now, you know how to split your traffic or budget by taking into account your resources, expectations, and time frame.

These strategies determine how fast you are going to achieve your goals with fewer expenses if you implement them wisely.

Which option you are planning to use?

Let me know in the comments section below…

For any help regarding your Google ads account management, you can reach out to us through the Contact us page.

Share it