Reports

Reports are the raw intelligence layer in Consumr.AI. Everything else, AI twins, respondents, surveys, is built on top of reports.

The three report types

Behavior / Persona report Built from a lookalike or cohort audience. This report constructs the behavioral and demographic profile of the segment: who these people are, what interests they hold, how they spend their time, what media they consume. The persona report is the identity layer of an AI twin; it answers “who is this person?”

Intent report Built from keywords. This report captures what the cohort is actively searching for, researching, or considering purchasing. It answers “what are they trying to do right now?” Intent is one of the twin’s two memory sources.

Mentions report Also built from keywords. This report captures what the cohort is talking about, sharing, and reacting to in conversations and social content. It answers “what are they saying about this category?” Mentions is the second memory source.

A twin is a compilation of three reports

This is the foundational technical fact: an AI twin is built from exactly three reports: one Behavior report and the associated Intent and Mentions reports. The Behavior report supplies the persona; the Intent and Mentions reports supply the memories.

• If the reports are narrow and well-targeted, the twin will be specific and insightful.
• If the inputs were too broad (blue “Broad” indicator), the twin’s responses will be correspondingly generic.
• Two segments that share the same underlying reports will produce twins that answer almost identically, regardless of what you named the segments.

Reports persist after twins and respondents are deleted

Reports are durable. When a twin is deleted, or when respondents created for a survey run are cleaned up afterward, the underlying reports remain. The reverse-segmentation method relies on this: you can use reports to create respondents independently of any existing twin, run a survey to identify emerging segment patterns, and then build new twins from those findings, all without having kept the intermediate twins or respondents around. See the How-To section for the full walkthrough of that method.