There may be situations where you have additional important information regarding user registrations and marketing spend that does not cleanly map to the Cohort Long-form (CLF) data specification.
Two common examples would be:
1. Above the line Spend - Non digital marketing efforts (eg. TV)
2. iOS SkadNetwork attribution
In order to consolidate total marketing spend and user registrations within the Kohort platform, this data can be passed as additional segments, which will roll up in reports at global or platform level. The mapping is intended to allow the inclusion of unattributed spend by registration date, for as much categorical information as is available. Additional information such as user registrations or IAP purchase value can be included as optional measures for now, but are not yet automatically included in reporting.
Input Format
Populate all unknown categories with:
SKAN - SKAdNetwork
ATL - Above the Line spend
Populate all known categories as normal.
Only populate row entries where registration_date == activity_date.
Set the active_users to 0.
Roll up campaigns with the same level of detail on the same registration date into a single daily entry.
The following are some examples of how to account for cases like these in the Cohort Long-Form format.
Categorical Mapping
Examples below assume that varying degrees of categorical information can be extracted based on the marketing campaign.
Example input data - with varying category information:
registration_date | registrations | marketing_spend | campaign | purchase |
2022-10-04 | 1077 | 430 | Test | 1 |
2022-10-04 | 200 | 80 | US-FB | 10 |
2022-10-04 | 278 | 31 | US | 90 |
2022-10-04 | 123 | 87 | FB | 150 |
2022-10-04 | 330 | 79 | TV Ad Campaign |
|
Mapping to Cohort Long Form - with varying category information:
registration_date | activity_date | registrations | active_users | marketing_spend | category_country | category_network | category_platform | category_channel | purchase |
2022-10-04 | 2022-10-04 | 1077 | 0 | 430 | SKAN | SKAN | IOS | PAID | 1 |
2022-10-04 | 2022-10-04 | 200 | 0 | 80 | US | FB | IOS | PAID | 10 |
2022-10-04 | 2022-10-04 | 278 | 0 | 31 | US | SKAN | IOS | PAID | 90 |
2022-10-04 | 2022-10-04 | 123 | 0 | 87 | SKAN | FB | IOS | PAID | 150 |
2022-10-04 | 2022-10-04 | 330 | 0 | 79 | ATL | ATL | ATL | PAID | Null |
When performing this mapping operation pay close attention to the following:
active_users has been set to as 0 to avoid double counting.
registration_date == activity_date for all row entries
The category_platform IOS and category_channel PAID are implied in the case of SKAN data, but should still be supplied.
The category has been explicitly assigned as SKAN, where it was not known e.g.
In Example 1, the campaign was not geo- or network-specific. Hence both country and network were mapped to SKAN.
In Example 2, the campaign was targeted to US and Facebook users, hence these values could be specified for country and network categories.
For Above The Line spend, similarly all known categories should be supplied, all implied categories made specific, and all unknown categories set to ATL.
Non-conflicting optional fields like registrations, purchase_value are included if available.
Aggregation
Where multiple campaigns have the same level of category detail for a day, roll these up into one entry:
Example input data - two campaigns without category information:
registration_date | registrations | marketing_spend | campaign | purchase |
2022-10-04 | 1077 | 430 | Charlie creative v1 | 1 |
2022-10-04 | 200 | 80 | Hitch trailer v4 | 10 |
Mapping to Cohort Long Form - single input row:
registration_date | activity_date | registrations | active_users | marketing_spend | category_country | category_network | category_platform | category_channel | purchase |
2022-10-04 | 2022-10-04 | 1077+200 | 0 | 430+80 | SKAN | SKAN | IOS | PAID | 11 |