Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.permutive.com/llms.txt

Use this file to discover all available pages before exploring further.

Include and Exclude segments allow you to control which users and behavioral signals are used during the lookalike modeling process.

​
Include Segments

By default, Permutive considers your entire audience when looking for lookalikes. Using Include Segments restricts the model to only find similar users within specific cohorts.

​
Use Cases

  • Geographic Restriction: Only find lookalikes among users in a specific country.
  • Device Restriction: Only find lookalikes among mobile app users.
Important: You must include a reasonable number of cohorts (typically 10 or more) to give the model enough behavioral signals to learn from.Adding only a single or a handful of cohorts is a common cause of model failure. Without a diverse set of features, the machine learning algorithm cannot identify the unique patterns that define your seed segment.

​
Exclude Segments

Exclude Segments tells the model to ignore specific users or signals during training.

​
Use Cases

  • Exclude Seed Users: If you want to find only new users who look like your seed, but aren’t already in it, you can exclude the seed segment itself.
  • Remove Noise: Exclude cohorts that are too similar to the seed or contain irrelevant behavioral data.
1
Access Model Settings
2
During the Create Model workflow, look for the Include or Exclude segment fields.
3
Add Cohorts
4
Search for and select the cohorts you want to use as filters.
5
Build Model
6
The model will now only learn from and score users based on these restrictions.
Be careful not to over-restrict the model. If you exclude too many segments, the model may not have enough data to identify meaningful patterns, leading to poor performance or failed builds.