Intra-campaign experiments

Intra-campaign experiments are used to test a specific feature within a single campaign. Unlike system-managed experiments where traffic is split between control and treatment campaigns, intra-campaign experiments split traffic within the campaign, based on whether the feature is enabled or not.

This workflow is supported for the following ExperimentType values:

  • ADOPT_AI_MAX
  • ADOPT_BROAD_MATCH_KEYWORDS

Setup

  1. Define the Experiment, providing an experiment type, a control ExperimentArm, and a treatment ExperimentArm. Each arm should reference the same campaign.
  2. Enable the test feature for the experiment by using a field mask. This is not necessary for ADOPT_BROAD_MATCH_KEYWORDS; instead, the broad match campaign setting will be enabled automatically upon experiment creation.
  3. Send a GoogleAdsService.Mutate request that includes mutate operations to create the experiment and experiment arms, and (if applicable) to enable the test feature.

Once set up, traffic is split within the campaign such that 50% of traffic is exposed to the enabled feature (the treatment group) and 50% is not (the control group).

Java

This example is not yet available in Java; you can take a look at the other languages.

C#

This example is not yet available in C#; you can take a look at the other languages.

PHP

This example is not yet available in PHP; you can take a look at the other languages.

Python

# Create the experiment resource name using a temporary ID.
experiment_resource_name = googleads_service.experiment_path(
    customer_id, "-1"
)

# Create the experiment.
experiment_operation = client.get_type("MutateOperation")
experiment = experiment_operation.experiment_operation.create
experiment.resource_name = experiment_resource_name
experiment.name = f"ADOPT_AI_MAX Experiment #{uuid4()}"
experiment.type_ = client.enums.ExperimentTypeEnum.ADOPT_AI_MAX
experiment.status = client.enums.ExperimentStatusEnum.SETUP

# Create the control arm. Both arms in an intra-campaign experiment
# reference the same base campaign.
control_arm_operation = client.get_type("MutateOperation")
control_arm = control_arm_operation.experiment_arm_operation.create
control_arm.experiment = experiment_resource_name
control_arm.name = "Control Arm"
control_arm.control = True
control_arm.traffic_split = 50
control_arm.campaigns.append(
    googleads_service.campaign_path(customer_id, campaign_id)
)

# Create the treatment arm.
treatment_arm_operation = client.get_type("MutateOperation")
treatment_arm = treatment_arm_operation.experiment_arm_operation.create
treatment_arm.experiment = experiment_resource_name
treatment_arm.name = "Treatment Arm"
treatment_arm.control = False
treatment_arm.traffic_split = 50
treatment_arm.campaigns.append(
    googleads_service.campaign_path(customer_id, campaign_id)
)

# Create a campaign operation with an update mask to enable AI Max and
# configure asset automation settings.
campaign_operation = client.get_type("MutateOperation")
campaign = campaign_operation.campaign_operation.update
campaign.resource_name = googleads_service.campaign_path(
    customer_id, campaign_id
)
campaign.ai_max_setting.enable_ai_max = True

for asset_automation_type_enum in [
    client.enums.AssetAutomationTypeEnum.TEXT_ASSET_AUTOMATION,
    client.enums.AssetAutomationTypeEnum.FINAL_URL_EXPANSION_TEXT_ASSET_AUTOMATION,
]:
    asset_automation_setting = client.get_type(
        "Campaign"
    ).AssetAutomationSetting()
    asset_automation_setting.asset_automation_type = (
        asset_automation_type_enum
    )
    asset_automation_setting.asset_automation_status = (
        client.enums.AssetAutomationStatusEnum.OPTED_IN
    )
    campaign.asset_automation_settings.append(asset_automation_setting)

client.copy_from(
    campaign_operation.campaign_operation.update_mask,
    protobuf_helpers.field_mask(None, campaign._pb),
)

# Send all mutate operations in a single Mutate request.
mutate_operations = [
    experiment_operation,
    control_arm_operation,
    treatment_arm_operation,
    campaign_operation,
]

response = googleads_service.mutate(
    customer_id=customer_id,
    mutate_operations=mutate_operations,
)

Ruby

This example is not yet available in Ruby; you can take a look at the other languages.

Perl

This example is not yet available in Perl; you can take a look at the other languages.

curl

Report on the experiment

Because control and treatment traffic are mixed within a single campaign, you must use direct experiment reporting to compare metrics between the control and treatment groups. Standard campaign-level reporting only shows aggregated metrics for the entire campaign and cannot distinguish between the two groups.

The following GAQL query can be used to retrieve click stats for an ADOPT_AI_MAX intra-campaign experiment.

SELECT
  experiment.resource_name,
  experiment.name,
  metrics.clicks,
  metrics.control_clicks,
  metrics.clicks_point_estimate,
  metrics.clicks_p_value
FROM experiment
WHERE experiment.type = 'ADOPT_AI_MAX'

Promote or end the experiment

After evaluating the results, you can either end or promote the experiment using ExperimentService.

  • End: If you are not satisfied with the results, use EndExperiment. The feature will be disabled, and the campaign will revert to serving all traffic without the experimental feature. This is a synchronous operation.
  • Promote: If you are satisfied with the results, use PromoteExperiment. This applies the experimental change as the new permanent state of the campaign. This is an asynchronous operation; see Asynchronous errors for details.

The graduate operation is not supported for intra-campaign experiments because there is no separate treatment campaign to graduate.

Previous
System-managed experiments
Next
Asset optimization experiments