Few catches for Models in Audience Manager

Last update: 2022-11-08

Description

Few catches around Audience Manager Models.

Resolution

  • How does lookalike audience work and its benefits?

https://experienceleague.adobe.com/docs/audience-manager/user-guide/features/algorithmic-models/look-alike-modeling/understanding-models.html?lang=en

  • How to determine the apt parameters/inputs to create it.

A few thousand users should be enough to run the model on, given that there is significant trait overlap between the baseline population and the population in the selected data sources. Look-Alike Modeling produces more accurate results, the larger the baseline is.

  • What is trait weight and how does it work?

Trait weight scale is a percentage that runs from 0% to 100%. Traits ranked closer to 100% means they’re more like the audience in your baseline population. TraitWeight ranks newly discovered traits in order of influence or desirability.

  • What is the minimum threshold population of the base trait/segment that can be taken into consideration?

As stated above, a few thousand users should be enough to run the model on, given that there is significant trait overlap between the baseline population and the population in the selected data sources.

  • Why and how to select data sources to increase and decrease relevance to base trait?

Use data sources which have at least some overlap with your baseline trait/segment, but at the same time bring in additional users. You should also consider the cost associated with each data feed. Cost and pricing models vary among data providers in Audience Marketplace.

  • Which traits to exclude and on what basis?

https://experienceleague.adobe.com/docs/audience-manager/user-guide/features/algorithmic-models/look-alike-modeling/trait-exclusion-algo-models.html?lang=en

  • Once the model run is completed, what are the next steps?

Once the model run has completed you can start creating your traits and segments.

  • Clarity on creating other traits and segments from the data after the model run and its working and usage.

https://experienceleague.adobe.com/docs/audience-manager/user-guide/features/traits/trait-builder/create-algorithmic-traits.html?lang=en

  • Range vs accuracy tradeoff.

https://experienceleague.adobe.com/docs/audience-manager/user-guide/features/traits/trait-accuracy-reach.html?lang=en

  • How to achieve optimal accuracy with increased reach and range?

https://experienceleague.adobe.com/docs/audience-manager/user-guide/features/traits/trait-accuracy-reach.html?lang=en#accuracy-and-reach-affect-audience-size

  • Factors dependent on the successful creation/implementation of lookalike and its scope

Please go through the https://experienceleague.adobe.com/docs/audience-manager/user-guide/features/algorithmic-models/look-alike-modeling/understanding-models.html?lang=en link for point 10 & 11 and let us know if you have any specific questions/issues which we may assist you with in any of the models.

  • How many days/timeframes of model run and the optimal time to create segments from the same?

Currently, you can create up to 20 algorithmic models and 50 algorithmic traits. The model retrains once every 8 days, along with refreshing the algorithmic trait population.

On this page