Conversion Lift measures the incremental number of conversions, site visits, and any other action directly driven by your ad. Conversion Lift data can help you adjust and improve your ads to generate more sales, leads, and app installs.
Before you begin
Before you can set up Conversion Lift measurement, you need to create a campaign to serve on a supported Google Ads platform. You can use Conversion Lift based on users for Video, Discovery, and Demand Gen campaigns. If you’re interested in using Conversion Lift based on users for Display, Search, Shopping, or Performance Max campaigns, reach out to your Google account representative for more information. Learn more About Conversion Lift .
Conversion Lift feasibility
Conversion Lift Feasibility, also referred to as Study Power, is an essential tool that estimates the certainty of results from your lift study. It helps determine the likelihood of successfully measuring the true, causal impact of your campaigns.
Study power indicates your study's probability of achieving conclusive results, and it is shaped by several factors:
- Historical campaign data
- Estimated lift
- Your chosen study configuration, including:
- Selected conversion actions
- Daily campaign budget
- Study duration
- Traffic split (holdback percentage)
The estimated certainty of lift is shown as a percentage, ranging from 50% to 90%, in 5% increments. A higher percentage indicates a greater likelihood that your ads will generate lift. If your study power is below 90%, budget guidance will be provided to help you achieve 90% certainty of lift. Note that you should determine the optimal certainty of lift for your needs. For more details, read about certainty levels .
To use Conversion Lift, your Google Ads account must track at least one compatible conversion action , either directly or through a manager account. All conversion must be set up to fire unconditionally. Find the compatible conversion sources listed here:
- Google ads conversion tag (AWCT)
- Enhanced Conversions
- Enhanced Conversion for Leads
- GA for apps (Firebase)
- App conversion tracking via the app attribution partner framework (AAP)
You have the flexibility to adjust the conversion actions you want to track until the study officially starts. By default, the conversion actions selected are those identified as primary for the campaigns included in the study. Be aware that altering this selection could influence the Study Power. Should you choose to concentrate on a smaller number of conversions, you will still be able to access conversion lift reports for all conversion actions at both a total and goal level after the study concludes.
Keep the following in mind:
- Currently unsupported conversion actions: Store visit, store sales, offline conversion import without personally identifiable information (PII) data.
- The only way to measure in-app conversions for non-app campaigns is via Google Ads API or Google data manager. Please reach out to your account team for more information.
Study setup eligibility for campaigns
Campaigns can only be part of a single study at the time. If your campaign is not eligible at the set up time, it means that it’s included in a different study (this can be Brand Lift, Search Lift or Conversion Lift). Please remove that campaign from that study, before setting up the new experiment.
For advertisers looking for directional lift (lower than 90% confidence), any budgets above $5,000 USD and 1,000 conversions will allow access to these results. You will not be able to save your study if budget is below $5,000 USD.
Best practices to decide holdout size
When you set up the experiment, the tool will provide a recommendation that allows you to maximize your chances of detecting lift. If you would like to manually customize it, keep in consideration the following:
The optimal holdout size depends on how much data you want to collect, and how quickly you would like to do that.
- Higher holdouts (up to 50%) increase the sample, but also have a higher opportunity cost (not showing your ad to control users).
- Smaller holdouts will require you to have a longer experiment to collect enough samples to detect differences between baseline conversions and exposed groups.
Best practices for measuring Conversion Lift
Conversion categories
We’ve found that upper and mid-funnel conversion actions such as pageviews, submit lead form, add to cart, and more show higher lift, and thus have a higher likelihood of measuring statistically significant lift. We recommend measuring not only bottom-of-the-funnel conversion actions, such as purchase actions, but also upper- and mid-funnel actions. These upper-funnel actions are useful as secondary KPIs or leading indicators that people are responding positively to the ads, in cases when the bottom-funnel action doesn’t have enough data to measure statistically significant results.
Bidding strategies
We've found that campaigns that are utilizing conversion-based or conversion-value-based bidding strategies, such as Target cost per acquisition (CPA), Maximize Conversions, or Target return on ad spend (ROAS) have higher lift than others such as Manual cost per click (CPC) or Target cost per thousand impressions (CPM). We recommend trying conversion-based or conversion-value-based bidding strategies to verify whether this leads to stronger lift for your campaigns.
Attribution model
We've found that campaigns using data-driven attribution have been associated with higher lift. Our data-driven attribution model is calibrated based on incrementality signals. We recommend trying data-driven attribution when trying to optimize towards incremental conversions to verify whether this leads to stronger lift for your campaigns.
Experiment duration
It’s important to set your experiment duration to a length that appropriately captures the average conversion lag, which is the average time between impression and conversion. We allow for studies as short as 7 days but typically recommend more than 14 days, especially if the business has a relatively longer conversion lag or higher value conversion. For example, short purchase cycles such as ordering pizza or purchasing movie tickets may be able to run a short 7–14 day study, whereas a more expensive product such as a mattress or travel accommodations require a longer study that runs for more than 14 days. We’ve found as much as a -17% drop in Absolute Lift in studies with a longer conversion lag measuring lift for less than 14 days, so therefore we recommend to run a study of 14 days minimum.
Frequency with which to run Conversion Lift studies
Any type of incrementality experiment comes with an opportunity cost. This means that by running the experiment and holding back a portion of your audience from viewing your ads, you’re potentially missing out on driving incremental conversions. Because of this, advertisers should be thoughtful about when and how frequently they run incrementality experiments. Experiment frequency tends to align with budget cycles. For example, advertisers should test before making major budget decisions or to validate hypotheses from Media Mix Model results to ensure the most accurate allocation of funds. Additionally, like all experiments, lift results fall within a confidence interval, so it’s encouraged to implement a testing plan to measure incrementality regularly. Based on our historical data, we find most advertisers run about 1–2 studies per year. Speak with your account manager if incrementality optimization is your goal.
Creative matters
It’s important to remember that the creative is what the user views and what can compel them to convert. Stronger creative will lead to stronger incrementality. Ultimately, the ads should be aligned with the goal that you’re trying to measure. Great ads start with the Core ABCD Principles:
- Attention: Hook and sustain attention with an immersive story
- Branding: Brand early, often, and richly
- Connection: Help people think or feel something
- Direction: Ask them to take action
Instructions
Set up Conversion Lift based on users
- In your Google Ads account, click the Goalsicon
.
- Click the Measurementsdrop down in the section menu.
- Click Lift measurement .
- Click the plus
button.
- Select Conversion Lift.
- Choose which campaigns you’d like to opt into Conversion Lift.
- Note: Campaigns can only be active in one study at a time. If you’re unable to select a campaign, it most likely means that it’s already being used in another study.
- You can choose one or more conversions to focus on during your conversion lift study, or keep all primary conversions that are selected by default based on which campaigns you choose. Regardless of which conversions you choose, you can still filter for all conversions in the report at the end of your study. Choosing conversions that happen more often increases certainty.
- Note: You can choose one or more conversion goals to focus on. Your study will measure only the conversion actions you select within each conversion goal. When the study ends, you can view results by conversion goals only, not the conversion actions within them.
- Choose the type of metrics you want to focus on in the report for this study. The other type of metrics will still be available in the report.
- Conversionsfocuses on incremental conversions and incremental cost per action (iCPA)
- Conversion valuefocuses on the incremental conversion value and incremental return on ad spend you drive from your ads.
Your study power is calculated based on what you select. Learn more about your Conversion Lift based on users measurement data .
- Select the start and end dates for your study to update study duration.
- Update the size of the holdback by inputting a value from 1% to 50%.
- Note: If you’re running studies with Brand Lift and Search Lift you must leave the 30% holdback.
- Review the study power to get an estimate of the certainty of lift you can expect at the end of your study. For example, a study power of 70% to 80% means you can expect that if your study detects lift, the certainty in the detected lift will be between 70% and 80%.
- Note: A study power above 90% likely means you will see the most certain results at the end of your study. Changing the duration of your experiment and increasing your campaign budget will help you increase your study power. Additionally, you can increase your study power by increasing the study duration, your traffic split, and adding more conversion actions. Focusing on increased conversions and increased conversion value can affect your study power if the conversion values vary.
- Click Save.Your study will start on the date specified in the Lift study setting.
View your Conversion Lift measurement data
Conversion Lift measurement data is available for each lift study at the "Product" or "Brand" level in the "Lift Measurement" table. You can also click into a specific study to view more granular results.
Here’s how you view your Conversion Lift measurement data:
- Click the columns icon
.
- Click Modify columns.
- Select Conversion Lift, then click Apply.
Frequently asked questions
What is the best approach for implementing a Conversion Lift study? How can I proceed and which specific campaigns would be most suitable?
You can set up studies based on users by following the instructions above, or Set up Conversion Lift based on geography .
To decide which campaigns are more suitable, you will need to ask yourself what would you like to learn from this experiment, if you would like to know “how much additional value are these Demand Gen creating, given that all your other campaigns are still running”, then choosing all Demand Gen campaigns would be the most suitable for you.
We would recommend reaching out to your Google account representative to plan a learning agenda and decide which are the most relevant studies for you.
Can I add another lift measurement type (for example, Brand Lift) to an existing Conversion Lift study?
No, you can’t add another lift measurement to an ongoing study. You will need to end your existing experiment and create a new one to include Brand Lift or Search Lift.
Can I stop and restart my study?
Yes, you can pause by changing the end date. Once the study has ended you won’t be able to restart.
Can I add or remove a campaign from an existing Conversion Lift study?
Yes, you can add and remove campaigns from an existing Conversion Lift study, but you can't add the same campaign to multiple studies.
Can I make changes to a Conversion Lift study that’s already running?
Yes, you can make changes to the end date, lift measurement type, add or remove campaigns.
Can I make changes to a campaign that is included in a Conversion Lift study that’s already running?
Yes you can but you should proceed with caution. We only recommend making changes which are business as usual (ex. Change bids or budgets). Updating creatives and audiences during the experiment might impact the performance of the study, and it will be hard to get learnings of what is working for your strategy.
