Core concepts

A few terms come up everywhere in Neuralift. Skim these once and the rest of the guide will read easily. The full list lives in the Glossary.

Organization

Your organization is the workspace where your team collaborates. It holds your data connections, your use cases, your members, and your delivery settings. You reach it at a URL that includes your workspace and organization names.

Use case

A use case is a single segmentation project: a business question you want to answer with segments, such as “Which customers are about to churn?” You define a goal and business context, point it at a prepared dataset, and Neuralift runs it. (In some places the running form of a use case is called a data run.)

Prepared dataset

A prepared dataset is the cleaned, analysis-ready table that a use case runs on. Neuralift builds it from the source tables you sync from your warehouse, so segmentation always works from consistent, well-structured input. See how data preparation works.

Data dictionary

When Neuralift prepares data for a use case, it profiles the prepared dataset into a data dictionary: a catalog of every column, classified as a KPI, continuous, categorical, or ID field, with distribution stats. Neuralift can annotate tables and columns with business context that sharpens how segments are named and explained.

Segment

A segment is a distinct group of customers that Neuralift discovers during a run. Each segment comes with its size, KPI profile, auto-generated insights, a persona, and the features that make it distinct.

Action

An action is a strategic recommendation generated for the run: a ranked play you can run across your customer base, broken down into action plans. Each plan owns one slice of the action’s audience, described by a one-sentence angle, and is backed by per-segment segment tactics — the evidence behind the play.

Publishing

A run’s results start private to Neuralift, who reviews them with your team. Publishing a run makes its segments, actions, and exports visible to everyone in the organization. Neuralift can unpublish to hide them again.

Export & delivery

• An export is a downloadable CSV of segment-labeled data.
• A delivery pushes segment-labeled results back into your data platform via Delta Share, cloud storage, or sFTP.

MCP

MCP (Model Context Protocol) is an open standard that lets AI assistants such as Claude, ChatGPT, and Copilot read your published Neuralift results, so you can ask questions about segments, actions, and insights in natural language. It is currently in Preview; see the MCP overview and release stages.