New Attribution in Google Analytics

Attribution has always been an integral part of both Google Analytics 360 and Google Analytics free solutions. A marketer would never be able to leverage analytics data in a meaningful way without a consolidated view of a customer journey. Even more, Google Analytics 360 adds a data-driven component to attribution modeling, enabling machine learning algorithms to determine how your vendors and efforts should be rewarded for generating conversions. It’s undoubtedly crucial to be able to look at your conversions at a different angle, stepping away from the one-size-fits-all last non-direct click perspective.

Google Analytics 360 introduced a new tool, called Attribution, accessible from the same report navigation pane. The first question we’d logically ask is what added value does this tool bring to attribution? Interestingly, Google released the new tool in conjunction with the existing Multi-Channel Funnels reports. At first glance, the tool appears more intuitive and flexible with a better Report Settings menu:

You can easily change the dimensions to select whether you want to see the conversion credit, change the model, and/or specify the path length.

Direct visits are treated differently here and are excluded from any attribution algorithms to give credit to other channels that assisted conversions based on available information. The direct channel should only get credit for a conversion if the entire path, for a specified lookback period, consisted of direct visits.

More visibility has been added into how attribution models work by showing a percentage of the conversion credit that has been given to a particular touchpoint in a customer’s journey:

Some limitations of the data-driven model have been removed in the new tool as compared to the Multi-Channel Funnels reports. Now, all touchpoints within a specified lookback period are used to train the model and to produce an output. This improves the prediction accuracy for those channels that assisted in long converting or non-converting paths.

Reporting time can now be customized to be better aligned with Google Analytics (conversion time) or Google Ads reporting (interaction time). The time range you select may indicate the time of a conversion or the time of an interaction. You have the ability to see how many conversions were generated by visitors that were touched by a particular channel within a specified period. This provides an understanding of a delayed response to your marketing for all campaigns you were running in a given time frame. Previously, it was pretty hard to report out of Google Analytics since it is entirely based on conversion time.

The Attribution tool can be set up to span across Analytics properties and thus, show totals for multiple Analytics profiles. However, it doesn’t impact how the credit is distributed and how the DDA (Data-Driven Attribution) model is trained – it generates data for each conversion individually and within one Analytics view.

The new solution also features a new set of channel definitions, known as Attribution Channel Groupings, that include a few improvements over Google Analytics’ Default Channel Grouping:

– Paid Search is split into Shopping, Paid Search Brand, and Paid Search Generic
– Display Remarketing was separated from Display General
– Video has been isolated into a new channel as well.

Each of these channels can perform differently in a path, requiring more or fewer investments, building awareness, consideration, and/or driving a purchase decision. They can have a different position within a journey and impact other channels. This all helps to better train data-driven models and to better distribute the credit under other models.

The concept of sampling attribution data has been eliminated and thus, doesn’t require an unsampled export to see an accurate dataset quickly. There are a few caveats, however. The Attribution tool will only work for the types of conversions you explicitly enable for running attribution against. This means that conversion attribution will only work going forward, no historical data will be available for those conversion types.

Impression data from Google Ads (GDN), Campaign Manager, and Display and Video 360 accounts that you can link to Google Analytics 360 is not available in Attribution projects nor custom-built attribution models.

The lookback window isn’t that flexible as in the Multi-Channel Funnels reports. You can set the lookback window to 30, 60, or 90 days, which is effective upfront but doesn’t allow you to adjust the lookback window for a period of fewer than 30 days.

From a technical standpoint, the new Attribution tool has been created to speed up attribution reporting by adding visibility into how attribution models work while getting rid of additional customizations such as custom models.

From a marketing perspective, it should help evaluate each channel’s contribution in a meaningful way and add data-driven context about the hidden or forgotten channels assisting conversions, especially for customers of Google Analytics free. You may want to use it in addition to Multi-Channel Funnels if the latter doesn’t provide the capability you need. The new solution is still in beta, so stay tuned for more updates once it’s launched.

Anton Dolgiy
anton.dolgiy@delvepartners.com


New Attribution in Google Analytics

Attribution has always been an integral part of both Google Analytics 360 and Google Analytics…

New Attribution in Google Analytics

Attribution has always been an integral part of both Google Analytics 360 and Google Analytics free solutions. A marketer would never be able to leverage analytics data in a meaningful way without a consolidated view of a customer journey. Even more, Google Analytics 360 adds a data-driven component to attribution modeling, enabling machine learning algorithms to determine how your vendors and efforts should be rewarded for generating conversions. It’s undoubtedly crucial to be able to look at your conversions at a different angle, stepping away from the one-size-fits-all last non-direct click perspective.

Google Analytics 360 introduced a new tool, called Attribution, accessible from the same report navigation pane. The first question we’d logically ask is what added value does this tool bring to attribution? Interestingly, Google released the new tool in conjunction with the existing Multi-Channel Funnels reports. At first glance, the tool appears more intuitive and flexible with a better Report Settings menu:

You can easily change the dimensions to select whether you want to see the conversion credit, change the model, and/or specify the path length.

Direct visits are treated differently here and are excluded from any attribution algorithms to give credit to other channels that assisted conversions based on available information. The direct channel should only get credit for a conversion if the entire path, for a specified lookback period, consisted of direct visits.

More visibility has been added into how attribution models work by showing a percentage of the conversion credit that has been given to a particular touchpoint in a customer’s journey:

Some limitations of the data-driven model have been removed in the new tool as compared to the Multi-Channel Funnels reports. Now, all touchpoints within a specified lookback period are used to train the model and to produce an output. This improves the prediction accuracy for those channels that assisted in long converting or non-converting paths.

Reporting time can now be customized to be better aligned with Google Analytics (conversion time) or Google Ads reporting (interaction time). The time range you select may indicate the time of a conversion or the time of an interaction. You have the ability to see how many conversions were generated by visitors that were touched by a particular channel within a specified period. This provides an understanding of a delayed response to your marketing for all campaigns you were running in a given time frame. Previously, it was pretty hard to report out of Google Analytics since it is entirely based on conversion time.

The Attribution tool can be set up to span across Analytics properties and thus, show totals for multiple Analytics profiles. However, it doesn’t impact how the credit is distributed and how the DDA (Data-Driven Attribution) model is trained – it generates data for each conversion individually and within one Analytics view.

The new solution also features a new set of channel definitions, known as Attribution Channel Groupings, that include a few improvements over Google Analytics’ Default Channel Grouping:

– Paid Search is split into Shopping, Paid Search Brand, and Paid Search Generic
– Display Remarketing was separated from Display General
– Video has been isolated into a new channel as well.

Each of these channels can perform differently in a path, requiring more or fewer investments, building awareness, consideration, and/or driving a purchase decision. They can have a different position within a journey and impact other channels. This all helps to better train data-driven models and to better distribute the credit under other models.

The concept of sampling attribution data has been eliminated and thus, doesn’t require an unsampled export to see an accurate dataset quickly. There are a few caveats, however. The Attribution tool will only work for the types of conversions you explicitly enable for running attribution against. This means that conversion attribution will only work going forward, no historical data will be available for those conversion types.

Impression data from Google Ads (GDN), Campaign Manager, and Display and Video 360 accounts that you can link to Google Analytics 360 is not available in Attribution projects nor custom-built attribution models.

The lookback window isn’t that flexible as in the Multi-Channel Funnels reports. You can set the lookback window to 30, 60, or 90 days, which is effective upfront but doesn’t allow you to adjust the lookback window for a period of fewer than 30 days.

From a technical standpoint, the new Attribution tool has been created to speed up attribution reporting by adding visibility into how attribution models work while getting rid of additional customizations such as custom models.

From a marketing perspective, it should help evaluate each channel’s contribution in a meaningful way and add data-driven context about the hidden or forgotten channels assisting conversions, especially for customers of Google Analytics free. You may want to use it in addition to Multi-Channel Funnels if the latter doesn’t provide the capability you need. The new solution is still in beta, so stay tuned for more updates once it’s launched.

Related Posts

How to Count Google Analytics Sessions Using Big Query

When you’re ready to start working with your data, it’s essential to use tools that…

Improve Your ROAS In One Day With Google Analytics 360

Improving return on ad spend is one of the most important goals for a marketing…

Join DELVE in Donating to UNICEF USA this Holiday Season

You’ve heard us say it here on the blog before: a data-driven approach to your…

New Attribution in Google Analytics

Attribution has always been an integral part of both Google Analytics 360 and Google Analytics…

Google Ads: Smart-Bidding Best Practices

Google’s machine learning models for bid optimization are collectively called “smart bidding” Most of these…

Optimizing App Predictions with AutoML Tables

Mikalai Tsytsarau, PhD,GCP Professional Data Engineer, DELVE AutoML Tables automatically build and deploy powerful machine…