Run an impact estimate (Beta)

Only available in Google Ad Manager 360.

This feature is in Beta

Features in Beta phase might not be available in your network. Watch the release notes for when this feature becomes generally available.

To understand how some features, such as different identifiers, may impact your revenue, you can run an impact estimate.

Impact estimates split allocated traffic into multiple groups known as variants. Each variant has different behavior applied to the traffic. For example, one variant may have a feature enabled and another variant may have the feature disabled. A variant with the feature disabled is also known as a baselinevariant. By comparing monetization on the traffic in the different variants, publishers can understand the impact of a given feature or set of features.

Set up an impact estimate

To get started, you select a type of impact estimate and choose your settings. You can then check the results and compare the baseline and variants.

Sign in to Google Ad Manager.
Click Optimization, and then Impact estimates.
Review the available types of impact estimates:
- PPID and first-party IDs: This option estimates the impact of publisher provided IDs (for programmatic) and first-party IDs on your traffic.
  Note: For the Beta, impact estimates are available for PPID and first-party IDs.
- Secure signals: This option helps you understand the revenue impact of your active secure signals. Learn how to run an impact estimate for secure signals .
On the type of impact estimate you want to run, click New impact estimate.

You can only run one impact estimate at a time. If "New impact estimate" isn't available, you may already have one running. You can review the results , and then end the estimate to start another.
Select your settings:
- Estimate options: Select PPID for programmatic, First-party IDs, or PPID and first-party IDs.
  Note: To run an impact estimate for PPID, you need to be sending PPID in ad requests to Ad Manager. If you're not sending PPID in ad requests, then an impact estimate for PPID would be irrelevant.
- Demand channel settings: Based on the type of impact estimate you selected, you can choose to override all demand channel settings and share IDs where not enabled, or choose to only turn off sharing the identifier type for your currently allowed bidders. To customize your demand channel settings before proceeding, click View demand channel settings.
- Estimate variants: To check the breakdown of baseline and variant settings, click Show variants breakdown.
  The baseline shows impact with the given feature off. The variants show alternative settings.
- Delivery settings: Select the time period and traffic allocation for the impact estimate.
- Estimate procedure: Review the explanation of the estimate procedure.
Click Run, and then Confirm.

After your setup, Google automatically enables the controls needed to conduct the estimate. This process doesn't impact or change your existing settings outside of your impact estimates.

Review the results

You can review your impact estimates, whether ended or running.

Note that impact estimates for PPID and first-party IDs don't include the first week of data in the results of the impact estimate.

This is because It can take a few days for buyers to ingest new identifiers and use them in their targeting and bidding models. Therefore, the first few days of data aren't an accurate reflection of the impact of the identifier.

Sign in to Google Ad Manager.
Click Optimization, and then Impact estimates.
For an overview of your impact estimates, review the table columns:
- Impact estimate : Shows the type of impact estimate you selected, such as "First-party IDs."
- Estimate period: Shows the time period you selected for the estimate to run.
- Status: Shows whether the estimate is scheduled to run, running, or complete.
Under "Impact estimate," click the name of an impact estimate.
- To view results by a different time frame, next to "View metrics by," click the dropdownand select an option, such as "Weekly."
- Under "Variant," review the results for revenue, eCPM, and impression lift.
  Tip: If there are multiple variants, click the variant you want to review. The breakdown table will update to show data for the selected variant.
- To break down the results, next to "Breakdown," click a dimension, such as "Browser category."
  Tip: Add a filter to view more specific data.
- To view a detailed chart for each row in your breakdown, click Expand.
(Optional) To end a running impact estimate, click End impact estimate, and then Confirm.

Beta note:To review the settings of your impact estimate, at the top, click Settings. For the Beta, the settings of impact estimates can't be changed. To start a new impact estimate, end a running impact estimate.

Statistical significance explained

Statistical significance tells you if an observed result is attributable to an experiment or just random chance. Impact estimates rely on a 99% confidence interval to assess this. This interval signifies that if we were to rerun the experiment many times, 99 out of 100 times the results would be expected to fall within this calculated range, suggesting the observed effect is highly reliable.

Statistically significant: The effect you see (such as an uplift in revenue) is likely due to the feature being experimented with, not just random fluctuation between the two test groups.
Not statistically significant: The observed effect could be random. This doesn't mean there's no impact, but the data obtained from the experiment is not sufficient to assume the impact is due to the feature, as opposed to random fluctuations between the two test groups. This can happen when the feature's impact is small relative to the general "noise" in ad serving data.

How lift is indicated:

Impact estimates compare a variant versus a baseline. If the baseline has a feature disabled and the variant has it enabled, lift % results in green indicate the feature is performing positively. Conversely, if the baseline has a feature enabled and the variant has it disabled, lift % results in red indicate the feature is performing positively.
If you see impact estimate lift results with either "insufficient data" or in grey with an asterisk (*) beside them, it means the lift value was statistically insignificant for breakdown.
If you see lift results in either green or red , it means there was a statistically significant result in the breakdown being examined.

Run an impact estimate for secure signals

Was this helpful?

How can we improve it?