We explain what a taxonomy is in the context of using second-party and third-party data.
Taxonomy
Taxonomy in the context of using second-party and third-party cohorts in Permutive is the list of all possible cohort ids and their descriptions. Thanks to the taxonomy we can map raw audience data from the data provider to human-readable names and use them to build Permutive cohorts.
Example
Below you can find a taxonomy that describes four different cohorts. Note that the ID has to be unique for the given Data Provider.
|
ID |
Name |
Description |
CPM |
Lifetime (days) |
|
4412 |
Demographic - Inferred Gender - Female |
Users whose gender has been inferred as female. |
1.00 |
30 |
|
4981 |
Demographic - Inferred Gender - Male |
Users whose gender has been inferred as male. |
1.00 |
30 |
|
4011 |
Demographic - Declared Gender - Female |
Users who have specified their gender as female. |
1.50 |
30 |
|
4099 |
Demographic - Declared Gender - Male |
Users who have specified their gender as male. |
1.50 |
30 |
The taxonomy for a given Data Provider has to be exhaustive and cover every possible cohort id that would be present in the audience data (that dataset contains user ids and the cohorts they belong to).
It should not change often and any modifications shouldn't change the meaning of a given cohort. Otherwise, the data would lose consistency as the same cohort would mean different things depending on time. It is ok to make cosmetical changes to names, descriptions or displayed CPM, etc.
This is in opposition to the audience data that is more transient and is updated as needed to reflect users' membership in cohorts.