From the left navigation bar, select [Altitude].
From the left navigation menu, select Attribution → Models.
In the top-right corner, select Create Model.
Refer to the sections below for more information on each part of creating a custom model.
This section covers the basic parts of a custom model, as well as how often the model produces data.
Enter the Model Name.
This is the name you will see in reporting when you compare data from each model.
Enter the model's Description.
You should provide a detailed description of the model's logic here.
Set the model's Scheduling. You have 3 options:
For models where you want complete control over when they generate data. To run the model, you'll use the Actions drop-down menu on the My Models screen and select Run Model.
For models that you only want to run for a short amount of time. These models generate data every night until the chosen expiration date.
If you select this option, you will need to set an Expiry Date that is no more than 30 days out from the current date. You can always extend a test model by updating the Expiry Date before it passes.
For models you intend to run for a much longer period of time. These models generate data every night and will never expire on their own.
This section covers the selections you will make that changes how a conversion path is constructed.
Set the Lookback Window. Maximum 90 days.
Conversion paths include much more than simply when a conversion occurred. Consumer touchpoints, like clicks, will likely happen before the final conversion happens. The Lookback Window determines which touchpoints are included in a conversion path by setting how far in advance a touchpoint can occur and still be included in the final conversion path.
To view these settings, select Show advanced settings in the Path Model section of the Model Builder screen.
Set how many days after the Lookback Window ends that the conversion path can be updated.
As impact.com learns about more events in the conversion path, like learning about additional devices via identity matching, the path might change. The maximum re-processing period is 15 days.
Toggle whether you want to trim consumer touchpoints if the consumer that generated them has not interacted with you for a period of time. This setting will trim touchpoints prior to the inactive period. The minimum period for a consumer to be considered as inactive is 1 day and the maximum period is 30 days.
For example, if a customer generates a touchpoint on January 1st and completes the conversion on January 16th, but you set the Customer Inactivity window for 15 days, that initial touchpoint will be scrubbed from the conversion path.
Identity Matching Level
Select whether you want to utilize identity matching at just your impact.com account-level, or if you want the help of the impact.com consortium to assist with identity matching efforts.
impact.com highly recommends using our consortium, as identity matching efforts can utilize data provided by other brands opted into the network.
In this section, you will select your base attribution model as well as your modifications to it. Next to Base Model, you'll need to select your base attribution model. See the table below for information on each choice.
Exponential Time Decay
Credit is allocated to the partners driving path touchpoints according to a decay curve win an exponential (curved) shape. Touchpoints decay at a half-life chosen by you, and you also choose whether the First or Last touchpoint has the highest value.
Blended Exponential Time Decay
Credit is allocated to the partners driving path touchpoint according to a decay curve with an exponential (curved) slope. Touchpoints decay at a half-life chosen by you, and you also choose whether the Introducing partner or the Closing partner would receive the most credit.
Linear Time Decay
Credit is allocated to the partners driving path touchpoints according to a decay curve with a linear (straight) slope. Touchpoints decay at a half-life chosen by you, and you also choose whether the First or Last touchpoint has the highest value.
Blended Linear Time Decay
Credit is allocated to the partners driving path touchpoints according to a decay curve with a linear (straight) slope. Touchpoints decay at a half-life chosen by your, and you also choose whether the Introducing partner or Closing partner would receive the most credit.
100% of the action's credit is allocated to the partner that drove the first touchpoint.
100% of the action's credit is allocated to the partner that drove the last touchpoint.
Credit is divided equally among all touchpoints involved in the conversion path.
You choose what percent of the credit will go to the first touchpoint in the conversion path and what percent will go towards the last. All remaining credit (100% minus the sum of the first and last touchpoint credit percentages) will be evenly distributed among all other touchpoints in the path.
The settings above, like the Lookback Window and Consumer Inactivity, will influence which interactions are determined to be part of the path and subsequently how credit is allocated along with your selection of a Base Model. For example, if you configure your lookback window to be 10 days and the raw path (consumer journey) extends to 11 days, touchpoints beyond day 10 will not be included in your path. Consequently, model output for all models save for Last Interaction will be affected as the models will not allocate credit to touchpoints outside the path.
To view these settings, select Show advanced settings in the Attribution Model section of the Model Builder screen.
Toggle whether you want to group impressions (a type of touchpoint) together if they happen within a chosen time window. If you enable this setting, set the window's hours, minutes, and seconds.
Toggle whether you want to create a product-level attribution model with the same settings you have chosen for the model you are currently developing. This will attribute the value of a given product to the various channels, partners, etc. on top of also attributing the conversion.
You can view Product Attribution data by selecting Attribution → Product Attribution in the left navigation menu.
Be sure to select Save when you are satisfied with your custom model!