Share declines over time
Posted: Wed Dec 11, 2024 6:01 am
The time decay model gives more credit to touchpoints that are closer to the time of conversion. For example, credit can be given based on a 7-day half-life, meaning that a click that occurred 8 days before conversion gets half as much credit for the conversion as a click that occurred 1 day before conversion. The advantage of this model is that it takes into account the remaining touchpoints and that it values the last channel—the one from which the conversion occurred. The disadvantage is that it marginalizes the points that initiate the funnel.
Here’s what attribution with time decay might look like for our example:
Time distribution model
Taking into account the position
Another example of multi-channel attribution. This time, the credits are distributed as follows: the first and last channels are assigned 40% of the conversion credit, while the remaining 60% is distributed to the remaining channels. The advantage of this model is that it allows you to optimize your budget for activities that effectively open conversion paths, as well as those that close the sale.
In our example, the attribution will look like this:
Google Ads - 40%,
SEO (second contact) - 6.66%,
Newsletter - 6.66%,
Social Media - 6.66%,
SEO (conversion) - 40%.
Position-aware model
Data-driven attribution
This is omnichannel attribution that relies heavily on data analysis. Credit for conversions is assigned using algorithmic machine learning models. Advanced models take into account both conversion and non-conversion paths. They include factors such as time since conversion, device type, number of ad interactions, order of exposure to ads, and type of creative assets. The model compares what happened with what could have happened to determine which touchpoints are most likely to lead to conversions. In other words, a data-driven attribution model assigns credit to touchpoints based on a calculated probability.
The big advantage of this model is that it provides more precise data than other models. It takes into account more factors than just the position on the conversion path or the time it takes to convert. The disadvantages, however, include the fact that we do not know exactly how these shares are calculated.
And this is what attribution could look like for our example if we choose a data-driven model:
Data-driven model
Attribution in Google Analytics
The popular analytics tool Google Analytics offers the ability to model conversions. Unfortunately, only 3 models are available (last click, data-driven attribution, and last click in Google Ads). So let's take a look at what Google Analytics offers us when it comes to conversion attribution modeling.
Attention!
Google Analytics 4 is under constant development, so the naming and location of reports may change over time.
Attribution reports can be found in the “Ads” section of the left-hand menu. There are currently two reports available: “Model Comparison” and “Attribution Funnels.”
Attribution Model Comparison
After selecting this report, you will see a table that will present the number of key events (as conversions are currently called in Google Analytics) and revenue for the two selected attribution models along with the percentage differences between them.
GA4 Attribution Model Comparison
By default, the report shows data for all key events facebook database set in GA4. To analyze just one selected conversion, expand the list in the upper corner above the title and deselect the other events.
Using the drop-down box in the first column, we can choose which dimension we will analyze attribution by. We can choose from:
default channel group,
source/medium,
source,
medium,
campaign.
If we have set up our own custom channel group, it will also appear in the list.
When comparing attribution models, it is worth remembering that each of them excludes direct visits from the conversion. The exception are conversion paths consisting exclusively of direct visits.
Conversion Paths (Key Events)
The second useful report is Key Event Paths. It shows you what paths customers take to get to conversions and how different attribution models assign credit for conversions on those paths.
At the top of the report we have 3 charts showing which channels/sources/media/campaigns are involved in initiating contact with the customer, which are in the middle of their path, and which ones finalize the conversion.
Customer touchpoints at different stages of the customer journey
The graph on the left shows the first 25% of touchpoints in a path, rounded to the nearest whole number. This segment is empty if the path has only one touchpoint.
The second graph shows the middle 50% of the touchpoints in the path. If the path has less than 3 touchpoints, this segment is empty.
The right graph shows the last 25% of touchpoints in the path, rounded to the nearest integer. If the path has only 1 touchpoint, then it will be assigned to this segment.
The second section of the page contains a table that will help us analyze the effectiveness of the selected paths. Each conversion path contains information about:
the total number of conversions that the selected path led to,
total purchase revenue (website or app purchases, minus cashback),
the number of days that passed from the first interaction to conversion,
the number of touchpoints with the page needed to convert.
Customer journeys in Google Analytics 4
Just like in the model comparison report, here we can select all key events (conversions) for analysis or only those that interest us the most.
Summary
Conversion attribution is an invaluable tool in a marketer’s arsenal. It allows you to assess the effectiveness of different marketing channels using attribution models. And while attribution modeling presents marketers with various challenges—such as multi-channel, difficulties integrating online and offline data, the variety of devices used by a single person, and privacy protections—it ultimately helps us better understand user behavior. Tools like Google Analytics help us discover which actions are most effective, which gives us the ability to optimize our marketing strategy and manage our advertising budget more effectively.
Here’s what attribution with time decay might look like for our example:
Time distribution model
Taking into account the position
Another example of multi-channel attribution. This time, the credits are distributed as follows: the first and last channels are assigned 40% of the conversion credit, while the remaining 60% is distributed to the remaining channels. The advantage of this model is that it allows you to optimize your budget for activities that effectively open conversion paths, as well as those that close the sale.
In our example, the attribution will look like this:
Google Ads - 40%,
SEO (second contact) - 6.66%,
Newsletter - 6.66%,
Social Media - 6.66%,
SEO (conversion) - 40%.
Position-aware model
Data-driven attribution
This is omnichannel attribution that relies heavily on data analysis. Credit for conversions is assigned using algorithmic machine learning models. Advanced models take into account both conversion and non-conversion paths. They include factors such as time since conversion, device type, number of ad interactions, order of exposure to ads, and type of creative assets. The model compares what happened with what could have happened to determine which touchpoints are most likely to lead to conversions. In other words, a data-driven attribution model assigns credit to touchpoints based on a calculated probability.
The big advantage of this model is that it provides more precise data than other models. It takes into account more factors than just the position on the conversion path or the time it takes to convert. The disadvantages, however, include the fact that we do not know exactly how these shares are calculated.
And this is what attribution could look like for our example if we choose a data-driven model:
Data-driven model
Attribution in Google Analytics
The popular analytics tool Google Analytics offers the ability to model conversions. Unfortunately, only 3 models are available (last click, data-driven attribution, and last click in Google Ads). So let's take a look at what Google Analytics offers us when it comes to conversion attribution modeling.
Attention!
Google Analytics 4 is under constant development, so the naming and location of reports may change over time.
Attribution reports can be found in the “Ads” section of the left-hand menu. There are currently two reports available: “Model Comparison” and “Attribution Funnels.”
Attribution Model Comparison
After selecting this report, you will see a table that will present the number of key events (as conversions are currently called in Google Analytics) and revenue for the two selected attribution models along with the percentage differences between them.
GA4 Attribution Model Comparison
By default, the report shows data for all key events facebook database set in GA4. To analyze just one selected conversion, expand the list in the upper corner above the title and deselect the other events.
Using the drop-down box in the first column, we can choose which dimension we will analyze attribution by. We can choose from:
default channel group,
source/medium,
source,
medium,
campaign.
If we have set up our own custom channel group, it will also appear in the list.
When comparing attribution models, it is worth remembering that each of them excludes direct visits from the conversion. The exception are conversion paths consisting exclusively of direct visits.
Conversion Paths (Key Events)
The second useful report is Key Event Paths. It shows you what paths customers take to get to conversions and how different attribution models assign credit for conversions on those paths.
At the top of the report we have 3 charts showing which channels/sources/media/campaigns are involved in initiating contact with the customer, which are in the middle of their path, and which ones finalize the conversion.
Customer touchpoints at different stages of the customer journey
The graph on the left shows the first 25% of touchpoints in a path, rounded to the nearest whole number. This segment is empty if the path has only one touchpoint.
The second graph shows the middle 50% of the touchpoints in the path. If the path has less than 3 touchpoints, this segment is empty.
The right graph shows the last 25% of touchpoints in the path, rounded to the nearest integer. If the path has only 1 touchpoint, then it will be assigned to this segment.
The second section of the page contains a table that will help us analyze the effectiveness of the selected paths. Each conversion path contains information about:
the total number of conversions that the selected path led to,
total purchase revenue (website or app purchases, minus cashback),
the number of days that passed from the first interaction to conversion,
the number of touchpoints with the page needed to convert.
Customer journeys in Google Analytics 4
Just like in the model comparison report, here we can select all key events (conversions) for analysis or only those that interest us the most.
Summary
Conversion attribution is an invaluable tool in a marketer’s arsenal. It allows you to assess the effectiveness of different marketing channels using attribution models. And while attribution modeling presents marketers with various challenges—such as multi-channel, difficulties integrating online and offline data, the variety of devices used by a single person, and privacy protections—it ultimately helps us better understand user behavior. Tools like Google Analytics help us discover which actions are most effective, which gives us the ability to optimize our marketing strategy and manage our advertising budget more effectively.