In this article
An overview of Contextual Cohorts and Cohort Affinity
Contextual Cohorts and Cohort Affinity allow you to target ads purely based on-page data, i.e. without processing or storing any user data. This, therefore, allows you to scale into environments where you otherwise might not have the appropriate consent to target users with data.
Our goal is to enable you to work with all available contextual signals in a single place. Our solution includes NLP classifications but also allows you to leverage data from their own tagging, third-party solutions as well as insights from your consented users.
Contextual Cohorts integrate with IBM Watson and allow you to leverage automated content classification when building Contextual Cohorts. Below is an example of a contextual cohort that targets positive soccer content:
Automated Classifications ensure consistent classifications across content, which is particularly valuable if the editorial tagging is inconsistent. It also unlocks dimensions that might not be available from the CMS like entities, concepts, or sentiment. Permutive’s integration with IBM Watson comes without any technical lift to the publisher.
Note: This is only available if you have Watson as part of your contract.
The second data type is Page Properties. This includes any custom meta data you have for your content. This could be editorial tagging, in-house classifications, third-party classifications or metadata like word counts. The available fields mirror the custom properties (or a subset of those) that you collect as part of the Pageview event for consented users.
This allows you to manage all campaign targeting (user-based and contextual) in one place. It also helps you to manage rules in a single place, rather than on a line-item level. This is particularly relevant for more complex definitions, e.g. when targeting a number of keywords for brand safety purposes.
Cohort Affinity allows you to leverage the insights from your consented users to make contextual targeting decisions. This bridges the gap between Custom Cohorts and your contextual signals, by taking learning from consented users to targeted content where a given Custom Cohort shows more engagement than the site average. When building a cohort using Cohort Affinity you can define a minimum affinity score.
The below example will target content that has seen above-average engagement from the “Sports Enthusiasts” cohort:
Cohort Affinity allows you to build a unique targeting offering, which is based on the unique understanding of their audience and which is in line with your audience strategy.
You can also combine any of the above predicate types in your Contextual Cohorts. Below is an example that uses Watson for semantic targeting and Page Properties for category targeting: