Conversion modeling uses machine learning to quantify the impact of your marketing efforts when there aren’t clear links between ad interactions and conversions due to browser restrictions, regulatory updates, and customer privacy expectations. You can get the highest ROI for your ad spend by using automated bidding informed by conversion modeling.
Conversion modeling allows Google to assign attribution – links between ad interactions and conversions – that's otherwise unobservable and helps you respect your users' privacy needs.
| Observed conversions | Modeled conversions |
|---|---|
| Uses cookies and other identifiers to assign links between ad interactions and conversions. | Uses machine learning to assign links between ad interactions and conversions where the links aren’t observable. |
Note: Modeled data is included in your total reported conversions only when there’s a high confidence that your ad resulted in conversions. This rigor ensures that we avoid systematically over-reporting. And in cases where we don’t observe enough data to be able to confidently model, we don't provide conversion modeling. Learn more about Google's behavioral modeling approach.
In this article you can learn about:
How conversion modeling works
As the industry moves away from individual identifiers like third-party cookies on the web and device IDs in apps, privacy-safe modeling solutions solve for unknowns in the customer journey and can improve your marketing. Learn more about conversion modeling and measurement data. You can also watch this video about the Fundamentals of conversion modeling in Google Ads.
Conversion modeling uses the behavior of observed groups to predict the behavior of unobserved groups without identifying any one individual.
Conversion modeling, at a high-level, works like this:
1. Ad interactions are separated into two groups.
One group contains ad interactions that have a clear, observable link to a conversion. The other group contains ad interactions that don't have a clear, observable link to a conversion.
2. The observed group is divided into subgroups.
The group where links can be observed is divided into subgroups based on shared characteristics, and conversion rates are calculated for each. For example, a subgroup of observed conversions may all be in the same country and occurred in the early morning, whereas another subgroup might all be using a specific browser on their phones.
3. The unobserved group is sorted into those subgroups.
Using the same shared characteristics of each defined subgroup, unobserved ad interactions and conversions are sorted into those groups.
4. Unobserved ad interactions and conversions are linked.
Using the known conversion rates from the observed subgroups, machine learning links unobserved ad interactions and conversions, where appropriate. The observed and modeled conversions are then integrated into your ad performance reporting to help you make informed decisions.
Note: Holdback validation (a machine learning best practice) maintains the accuracy of Google’s models. Modeled conversions are compared to observed conversions that were held back, and the information is used to tune the models. Google will communicate changes that might have a large impact on your data.
Learn how conversion modeling improves your marketing with new sources of measurement data and watch this video about the Conversion modeling principles
Where is modeling happening in Google Ads?
Cross-device conversions
- Modeled to account for users who start their journey on one device with an ad interaction and complete the conversion on another. Learn more about Google Ads best practices
Conversion totals reporting, attribution paths, and conversion values
- For conversion values, we use the values from your observed conversions to predict the value of a modeled conversion. Learn more about attribution models and conversion values
When cookies are limited or not allowed
- Conversion modeling can be used under specific conditions in the following situations:
- If conversions can’t be measured because third party cookies aren’t allowed by browsers, then conversions are modeled based on your websites’ traffic. Learn how to improve modeling by upgrading to the global site tag
- Browsers that limit the time window for first-party cookies will have conversions (beyond the window) modeled using enhanced conversions.
Country specific consent requirements
- Some countries require consent to use cookies for advertising activities. When advertisers use consent mode, conversions are modeled for unconsented users.
iOS 14 impact
- Apple’s App Tracking Transparency (ATT) policy requires developers to obtain permission to use certain information from other apps and websites. Google won’t use information (such as IDFA) that falls under the ATT policy. Conversions whose ads originate on ATT impacted traffic are modeled.
Offline conversion imports and user accounts without enough weekly conversions might not include modeled conversions.
Benefits
Modeling can benefit your business including:
- Representative view of your performance: Gain a more accurate picture of your advertising outcomes (ROI), and a more accurate picture of the conversion path across devices and channels resulting from ad interactions.
- More efficient campaign optimization and bidding: Privacy regulations and technology limitations mean lost observation for certain cohorts of users (for example, unconsented users, or users using particular device types or browsers). This means our automated bidding algorithms will need to make optimization decisions based on incomplete data. Modeling solves for gaps in data to ensure automated bidding is working off of accurate information about your website or app activity.
- Accurate, privacy-centric measurement: Conversion modeling uses the behavior of observed groups to predict the behavior of other groups without identifying any one individual.
Improve your modeling
There are several approaches to enriching the data you provide to Google that can strengthen the first-party web data and modeling Google can provide to you. These solutions may lead to improvements in scale and performance.
Web solutions
Enhanced conversions for web |
Learn more About enhanced conversions for web
- What it does: Helps improve the accuracy of your account-wide conversion measurement by increasing observable data and improving overall quality of conversion modeling. The first-party data you already have is matched with signed-in Google accounts that engaged with your ads. When a match happens, a conversion is recorded.
- Who it’s for: Advertisers who want to measure events that happen on a website. Learn how to set up enhanced conversions for web manually with your global site tag or set up enhanced conversions for web manually with Google Tag Manager.
- Performance impact: Advertisers who implement enhanced conversions see a conversion uplift of 17% on Youtube.
Consent mode |
- What it does: Automatically adjusts how your Google tags behave based on the consent status of your users.
- Who’s it for: Advertisers who want to recover conversions from users in the European Economic Area (EEA) and UK who don't consent to tracking through the consent banner.
- Requirements:
- Correctly implemented consent mode or the IAB Transparency & Consent Framework (TCF v2.0).
- Have a daily ad click threshold of 100 clicks per day, per country and domain grouping.
- Performance impact: Results from Google Ads have shown that, on average, conversion modeling through consent mode recovers more than half of ad-click-to-conversion journeys lost. Those results may vary depending on consent rates and your consent mode setup.
iOS app conversion resources
App tracking transparency framework (ATT) |
Updates to iOS 14 measurement
Learn more about Updates to iOS 14 campaign measurement
- What it does: Allows access to an iOS device’s identifier, which can be used to measure advertising performance. Learn more about Apple’s implementation
- Who’s it for: iOS developers who want to run advertising campaigns for their apps and are OK implementing Apple’s ATT prompt.
- Performance impact: Increases observable data for app campaigns, leading to improvements in scale & performance.
SKAdNetwork |
Learn more about Best practices for iOS App campaigns
- What it does: Provides the only holistic view of how many installs your app campaigns have received. It powers reporting and campaign optimization. Learn more about Apple’s implementation.
- Who’s it for: iOS developers who want to run advertising campaigns for their apps.
On-device conversion measurement |
Learn more About on-device conversion measurement
- What it does: Helps recover observability for in-app conversions otherwise not counted as a result of decreasing The Identifier for Advertisers (IDFA) on iOS post-ATT. Uses email addresses provided by users through your app in a privacy-safe way, through applicable Google channels to match more conversions with user interactions. This matching and attribution is facilitated by the Google Analytics for Firebase SDK. Matching and attribution is performed in a process that ensures no user-data leaves the device, while providing advertising attribution for iOS campaigns.
- Who’s it for: App developers who want to run iOS app promotion campaigns and have 1P user-provided email addresses available.
- Performance impact: Increases observable data for app campaigns, leading to improvements in measurement and modeling and enabling better performance.
Privacy-centric approach
Google doesn’t allow fingerprint IDs or other attempts to identify individual users. Instead, Google aggregates data (such as historical conversion rates, device type, time of day, geo, and more) to predict the likelihood of conversions across the set of users who viewed or clicked on an ad.
