All You Need to Know about Data-Driven Attribution
by
Brian Plant
·
Last updated:
Last updated:
Sep 6, 2024
Sep 6, 2024
If you use Google Analytics or Google Ads, you probably are using data-driven attribution (DDA) as it's the primary model those platforms use. However, you might not be fully aware of the details about it, including its strength, weaknesses, and possible better solutions. It is important to fully understand DDA, to ensure it's the correct solution for your business. So let's dive into one of the more advanced types of multi-touch attribution models, data-driven attribution (DDA).
What is Data-Driven Attribution and how does it work
Most multi-touch attribution models such as linear, time decay, and u-shape, have simple rules that determine how credit is divided up for purchases, when the user has multiple ad touches. For example, a linear model gives equal weight to touches, so if a user has any two touches, those two touches will always split credit 50%-50%.
Data-driven attribution distributes credit not based on simple rules, but based on a more complex algorithm that estimates the impact of each individual touchpoint based on historical purchase data for your business.
The machine learning algorithm evaluates all previous converting and non-converting user paths, identifying how within any user path each additional touchpoint impacted conversion. It then uses that probability to more accurately assign fraction credit amongst touchpoints.
A high-level illustration on how DDA compares user paths to determine credit. Every potential user path for converting and non-converting sessions is analyzed and informs the model.
A data-driven attribution algorithm takes factors like number of touchpoints, time in between touchpoints, device type, and types of touch, into account when deciding how credit should be assigned.
Because the model is trained on your business' data, each model is unique for each business, and requires a minimum amount of purchase data before it can be used. The model is also constantly evolving as it's given new data.
Benefits of Data-Driven Attribution
There are multiple benefits of data-driven attribution:
More accurate than simpler models
Compared to other rule based multi-touch models, data-driven is a more accurate attribution model that provides a nuanced understanding of how different channels contribute to conversions.
Better decisions
By showing closer to the true value of each touchpoint, it empowers markets to make smarter budget allocations.
Improved ROI
Because it is more accurate, using DDA over a rule-based attribution model helps improve the ROI of marketing campaigns.
Limitations of Data-Driven Attribution
Like any attribution methodology, data-driven attribution has its limitations. While data driven is more accurate than other multi-touch models, it shares the same flaws as them.
It doesn't directly measure causation
It identifies correlations between touchpoints and conversions, which are not necessarily causal relationships. So it cannot definitively prove that a specific touchpoint caused a conversion.
It is limited by data quality
Collecting every touchpoint at a user level across all channels is very difficult and for some channels, impossible. So channels where user touches are missing will be understated or completely not accounted for. For example, a click-based DDA would not consider the view-through impact of a channel like Snapchat, and linear TV advertising would not be considered at all.
How to overcome the limitations of Data-Driven Attribution
At WorkMagic, we have created a measurement solution specifically built to address the limitations of standard data-driven attribution. It's called Incrementality-based Attribution. It combines incrementality testing with data-driven attribution to give marketers the most accurate cross-channel incrementality based measurement at every level of granularity in real-time.
Incrementality-adjusted Attribution combines incrementality testing with data-driven attribution
While data-driven attribution doesn't measure causation, incrementality tests do.
To properly run ongoing incrementality tests, you need a robust team of data scientists and data engineers working together with your marketing team. While you can do this in-house, it's rarely cheaper or faster to do so. WorkMagic automates the whole process, empowering marketers to launch incrementality tests in minutes, removing the need for internal data science and data engineering resources.
The results of each incrementality test run using WorkMagic are used to calibrate the brand's unique attribution model, making it more accurate. Repeating this process of testing and calibrating increases the accuracy of a brand's custom attribution model each time and is a NorthStar guiding brands to get a better return from their marketing efforts.
To address the data limitations of standard DDA, WorkMagic uses additional datasets, alongside incrementality tests, that specifically help measure the full impact of all channels. This includes using WorkMagic's partner Fairing's post-purchase surveys, which are specifically designed to improve attribution of hard to measure channels.
When is Data-Driven Attribution the best solution
While data-driven attribution has limitations, for some advertisers it is still the best solution. Incrementality tests require a minimum amount of historical purchases, which makes them not a practical solution for smaller advertisers. For these advertisers, data-driven attribution works best.
Like incrementality-based attribution, WorkMagic's data-driven-attribution model was built to overcome the limitations of standard DDA models. It also uses post purchase surveys, powered by Fairing, to help measure non-click impact to properly credit sources like word of mouth, podcasts, or other traditionally hard to measure channels. Also, WorkMagic's aggregate learnings from incrementality tests run on our platform helps refine our model to better reflect your media's true impact.
Conclusion
Data-driven attribution is a popular model as it's used by Google Analytics and Google Ads. It is a superior multi-touch model compared to common rule based models, but is limited by the data available and doesn't directly measure casualty.
For smaller advertisers, data-driven attribution is the best solution. While larger advertisers should use incrementality-based attribution, which is a combination of data-driven attribution and incrementality tests.
If you use Google Analytics or Google Ads, you probably are using data-driven attribution (DDA) as it's the primary model those platforms use. However, you might not be fully aware of the details about it, including its strength, weaknesses, and possible better solutions. It is important to fully understand DDA, to ensure it's the correct solution for your business. So let's dive into one of the more advanced types of multi-touch attribution models, data-driven attribution (DDA).
What is Data-Driven Attribution and how does it work
Most multi-touch attribution models such as linear, time decay, and u-shape, have simple rules that determine how credit is divided up for purchases, when the user has multiple ad touches. For example, a linear model gives equal weight to touches, so if a user has any two touches, those two touches will always split credit 50%-50%.
Data-driven attribution distributes credit not based on simple rules, but based on a more complex algorithm that estimates the impact of each individual touchpoint based on historical purchase data for your business.
The machine learning algorithm evaluates all previous converting and non-converting user paths, identifying how within any user path each additional touchpoint impacted conversion. It then uses that probability to more accurately assign fraction credit amongst touchpoints.
A high-level illustration on how DDA compares user paths to determine credit. Every potential user path for converting and non-converting sessions is analyzed and informs the model.
A data-driven attribution algorithm takes factors like number of touchpoints, time in between touchpoints, device type, and types of touch, into account when deciding how credit should be assigned.
Because the model is trained on your business' data, each model is unique for each business, and requires a minimum amount of purchase data before it can be used. The model is also constantly evolving as it's given new data.
Benefits of Data-Driven Attribution
There are multiple benefits of data-driven attribution:
More accurate than simpler models
Compared to other rule based multi-touch models, data-driven is a more accurate attribution model that provides a nuanced understanding of how different channels contribute to conversions.
Better decisions
By showing closer to the true value of each touchpoint, it empowers markets to make smarter budget allocations.
Improved ROI
Because it is more accurate, using DDA over a rule-based attribution model helps improve the ROI of marketing campaigns.
Limitations of Data-Driven Attribution
Like any attribution methodology, data-driven attribution has its limitations. While data driven is more accurate than other multi-touch models, it shares the same flaws as them.
It doesn't directly measure causation
It identifies correlations between touchpoints and conversions, which are not necessarily causal relationships. So it cannot definitively prove that a specific touchpoint caused a conversion.
It is limited by data quality
Collecting every touchpoint at a user level across all channels is very difficult and for some channels, impossible. So channels where user touches are missing will be understated or completely not accounted for. For example, a click-based DDA would not consider the view-through impact of a channel like Snapchat, and linear TV advertising would not be considered at all.
How to overcome the limitations of Data-Driven Attribution
At WorkMagic, we have created a measurement solution specifically built to address the limitations of standard data-driven attribution. It's called Incrementality-based Attribution. It combines incrementality testing with data-driven attribution to give marketers the most accurate cross-channel incrementality based measurement at every level of granularity in real-time.
Incrementality-adjusted Attribution combines incrementality testing with data-driven attribution
While data-driven attribution doesn't measure causation, incrementality tests do.
To properly run ongoing incrementality tests, you need a robust team of data scientists and data engineers working together with your marketing team. While you can do this in-house, it's rarely cheaper or faster to do so. WorkMagic automates the whole process, empowering marketers to launch incrementality tests in minutes, removing the need for internal data science and data engineering resources.
The results of each incrementality test run using WorkMagic are used to calibrate the brand's unique attribution model, making it more accurate. Repeating this process of testing and calibrating increases the accuracy of a brand's custom attribution model each time and is a NorthStar guiding brands to get a better return from their marketing efforts.
To address the data limitations of standard DDA, WorkMagic uses additional datasets, alongside incrementality tests, that specifically help measure the full impact of all channels. This includes using WorkMagic's partner Fairing's post-purchase surveys, which are specifically designed to improve attribution of hard to measure channels.
When is Data-Driven Attribution the best solution
While data-driven attribution has limitations, for some advertisers it is still the best solution. Incrementality tests require a minimum amount of historical purchases, which makes them not a practical solution for smaller advertisers. For these advertisers, data-driven attribution works best.
Like incrementality-based attribution, WorkMagic's data-driven-attribution model was built to overcome the limitations of standard DDA models. It also uses post purchase surveys, powered by Fairing, to help measure non-click impact to properly credit sources like word of mouth, podcasts, or other traditionally hard to measure channels. Also, WorkMagic's aggregate learnings from incrementality tests run on our platform helps refine our model to better reflect your media's true impact.
Conclusion
Data-driven attribution is a popular model as it's used by Google Analytics and Google Ads. It is a superior multi-touch model compared to common rule based models, but is limited by the data available and doesn't directly measure casualty.
For smaller advertisers, data-driven attribution is the best solution. While larger advertisers should use incrementality-based attribution, which is a combination of data-driven attribution and incrementality tests.
If you use Google Analytics or Google Ads, you probably are using data-driven attribution (DDA) as it's the primary model those platforms use. However, you might not be fully aware of the details about it, including its strength, weaknesses, and possible better solutions. It is important to fully understand DDA, to ensure it's the correct solution for your business. So let's dive into one of the more advanced types of multi-touch attribution models, data-driven attribution (DDA).
What is Data-Driven Attribution and how does it work
Most multi-touch attribution models such as linear, time decay, and u-shape, have simple rules that determine how credit is divided up for purchases, when the user has multiple ad touches. For example, a linear model gives equal weight to touches, so if a user has any two touches, those two touches will always split credit 50%-50%.
Data-driven attribution distributes credit not based on simple rules, but based on a more complex algorithm that estimates the impact of each individual touchpoint based on historical purchase data for your business.
The machine learning algorithm evaluates all previous converting and non-converting user paths, identifying how within any user path each additional touchpoint impacted conversion. It then uses that probability to more accurately assign fraction credit amongst touchpoints.
A high-level illustration on how DDA compares user paths to determine credit. Every potential user path for converting and non-converting sessions is analyzed and informs the model.
A data-driven attribution algorithm takes factors like number of touchpoints, time in between touchpoints, device type, and types of touch, into account when deciding how credit should be assigned.
Because the model is trained on your business' data, each model is unique for each business, and requires a minimum amount of purchase data before it can be used. The model is also constantly evolving as it's given new data.
Benefits of Data-Driven Attribution
There are multiple benefits of data-driven attribution:
More accurate than simpler models
Compared to other rule based multi-touch models, data-driven is a more accurate attribution model that provides a nuanced understanding of how different channels contribute to conversions.
Better decisions
By showing closer to the true value of each touchpoint, it empowers markets to make smarter budget allocations.
Improved ROI
Because it is more accurate, using DDA over a rule-based attribution model helps improve the ROI of marketing campaigns.
Limitations of Data-Driven Attribution
Like any attribution methodology, data-driven attribution has its limitations. While data driven is more accurate than other multi-touch models, it shares the same flaws as them.
It doesn't directly measure causation
It identifies correlations between touchpoints and conversions, which are not necessarily causal relationships. So it cannot definitively prove that a specific touchpoint caused a conversion.
It is limited by data quality
Collecting every touchpoint at a user level across all channels is very difficult and for some channels, impossible. So channels where user touches are missing will be understated or completely not accounted for. For example, a click-based DDA would not consider the view-through impact of a channel like Snapchat, and linear TV advertising would not be considered at all.
How to overcome the limitations of Data-Driven Attribution
At WorkMagic, we have created a measurement solution specifically built to address the limitations of standard data-driven attribution. It's called Incrementality-based Attribution. It combines incrementality testing with data-driven attribution to give marketers the most accurate cross-channel incrementality based measurement at every level of granularity in real-time.
Incrementality-adjusted Attribution combines incrementality testing with data-driven attribution
While data-driven attribution doesn't measure causation, incrementality tests do.
To properly run ongoing incrementality tests, you need a robust team of data scientists and data engineers working together with your marketing team. While you can do this in-house, it's rarely cheaper or faster to do so. WorkMagic automates the whole process, empowering marketers to launch incrementality tests in minutes, removing the need for internal data science and data engineering resources.
The results of each incrementality test run using WorkMagic are used to calibrate the brand's unique attribution model, making it more accurate. Repeating this process of testing and calibrating increases the accuracy of a brand's custom attribution model each time and is a NorthStar guiding brands to get a better return from their marketing efforts.
To address the data limitations of standard DDA, WorkMagic uses additional datasets, alongside incrementality tests, that specifically help measure the full impact of all channels. This includes using WorkMagic's partner Fairing's post-purchase surveys, which are specifically designed to improve attribution of hard to measure channels.
When is Data-Driven Attribution the best solution
While data-driven attribution has limitations, for some advertisers it is still the best solution. Incrementality tests require a minimum amount of historical purchases, which makes them not a practical solution for smaller advertisers. For these advertisers, data-driven attribution works best.
Like incrementality-based attribution, WorkMagic's data-driven-attribution model was built to overcome the limitations of standard DDA models. It also uses post purchase surveys, powered by Fairing, to help measure non-click impact to properly credit sources like word of mouth, podcasts, or other traditionally hard to measure channels. Also, WorkMagic's aggregate learnings from incrementality tests run on our platform helps refine our model to better reflect your media's true impact.
Conclusion
Data-driven attribution is a popular model as it's used by Google Analytics and Google Ads. It is a superior multi-touch model compared to common rule based models, but is limited by the data available and doesn't directly measure casualty.
For smaller advertisers, data-driven attribution is the best solution. While larger advertisers should use incrementality-based attribution, which is a combination of data-driven attribution and incrementality tests.