Getting Started

AudienceGPT is an AI-powered audience taxonomy classification platform. You describe audience segments in natural language, and the AI classifies them into a structured 7-layer taxonomy used across the digital advertising ecosystem. This guide walks you through account setup, navigation, and your first classification.

Account Setup

Your organization administrator will invite you to AudienceGPT via email. Click the invitation link to create your account through Clerk, the authentication provider.

1. Accept your invite -- Click the link in the invitation email.
2. Create your account -- Set up your credentials through the Clerk sign-in flow. You can use email/password or a supported SSO provider.
3. Choose your organization -- If you belong to multiple organizations, select the one you want to work with. You can switch organizations at any time from the sidebar.

tip

If your organization uses SSO (Single Sign-On), your administrator will provide instructions specific to your identity provider.

Dashboard Overview

After signing in, you land on the Classify page (/classify), the primary interface for creating audience topics. The application is organized into several sections accessible from the left sidebar:

Sidebar Item	Path	Purpose
Classify	/classify	Chat-based topic classification, campaign brief analysis, matrix generation
Library	/library	Browse, filter, and manage your classified topics
Import	/import	Bulk CSV import with column mapping and classification
Destinations	/destinations	Monitor activated segments and delivery health
Hygiene	/hygiene	Data health dashboard for outdated topics and stale activations

info

The sidebar collapses on smaller screens. Use the hamburger menu icon to toggle it open.

Sidebar Navigation

The left sidebar provides quick access to all major features. At the bottom, you will find:

• Organization selector -- Switch between organizations if you belong to more than one.
• Your profile -- Access account settings and sign-out options.

The classify page is the hub for most day-to-day work. From here you can classify individual topics, upload campaign briefs, generate matrix taxonomies, and browse the global catalog -- all through a conversational chat interface.

Your First Classification

Let's walk through classifying your first audience topic step by step.

Step 1: Enter a Topic

Navigate to the Classify page. In the chat input at the bottom of the screen, type the name of an audience topic you want to classify. This should be a behavioral intent topic -- something people are actively researching or doing online.

Examples of good topics:

• Toyota RAV4 -- Brand/product interest
• Kitchen Remodeling -- Service/function interest
• Cybersecurity Solutions -- Solution interest
• Ozempic -- Product/health interest

Type your topic and press Enter or click the send button.

Step 2: Provide Context (Gathering Phase)

The AI assistant will ask follow-up questions to gather context about your topic. This is the gathering phase of the conversation. You may be asked about:

• Keywords -- Additional terms related to the topic (e.g., "hybrid SUV, compact crossover")
• Category hints -- Which industry or vertical the topic belongs to
• Segment type -- Whether this is a B2B, B2C, or other segment

Answer the questions naturally. The AI uses your responses to improve classification accuracy.

tip

The more context you provide, the more accurate the classification will be. If you are classifying a well-known brand or product, the AI may use web search to verify details automatically.

Step 3: Confirm Your Input

Once the AI has enough context, it presents a summary of the topic information for your confirmation. Review the details:

• Topic name -- The primary name for this audience segment
• Category -- The detected parent category (e.g., "Auto", "Business Technology")
• Keywords -- Any additional keyword signals

If everything looks correct, confirm to proceed. If something needs adjustment, tell the AI what to change.

Step 4: Classification

After confirmation, the AI runs the topic through the 7-layer classification engine. This takes a few seconds. During this time, the system:

1. Determines the intent type (brand, product, service, etc.)
2. Assesses intensity level (how active the interest signal is)
3. Maps to an awareness stage (Schwartz model)
4. Identifies the segment type (B2B, B2C, etc.)
5. Flags sensitivity if applicable (regulated categories)
6. Evaluates buyer journey position
7. Calculates a composite score (0--100)

The AI may also perform a web search to verify brand identity, product details, or company information, which prevents misclassification (e.g., an activewear brand being classified as B2B SaaS).

Step 5: Review Your Results

The classification results are displayed in a structured card showing all 7 layers. Here is what each layer means at a glance:

Layer	What It Tells You
Intent Type	What kind of interest this represents (brand awareness, product research, etc.)
Intensity	How strong the behavioral signal is (dormant through critical)
Awareness	Where the audience is in the awareness journey (unaware through retention)
Segment	The business model context (B2B, B2C, B2B2C, B2E, B2G)
Sensitivity	Whether the topic falls under regulated categories
Buyer Journey	The purchase readiness stage
Composite Score	A single 0--100 score summarizing overall intent strength

You will also see DSP segment names generated for major platforms (Trade Desk, LiveRamp, and Internal naming formats). These are the names used when activating segments through advertising platforms.

info

For a deep dive into each classification layer, see the Classification guide.

Understanding the 7-Layer Results

Here is a quick reference for interpreting your first classification:

Composite Score Ranges

Score Range	Label	Meaning
80--100	Hot Lead	High purchase intent, ready-to-buy signals
60--79	Warm Prospect	Active evaluation, comparing options
40--59	Active Researcher	Engaged research phase, gathering information
20--39	Early Explorer	Initial awareness, browsing content
0--19	Cold Audience	Passive or dormant interest

Intensity Levels

Level	Score	Description
Dormant	0	No active signals detected
Passive	15	Minimal, background-level interest
Curious	30	Light research, browsing related content
Active	50	Regular engagement with the topic
Engaged	70	Sustained, repeated interaction
Urgent	85	Time-sensitive need or strong buying signals
Critical	95	Immediate action required, highest priority

Adding Topics to Your Library

After reviewing the classification results, you have the option to add the topic to your library. Click the "Add to Library" button in the results card. The topic is saved to your organization's library where you can:

• View it alongside all your other classified topics
• Filter and search across your taxonomy
• Export it for use in DSP platforms
• Track changes through the audit history
• Reclassify it when the engine is updated

warning

Topics are saved per organization. If you switch organizations, you will see a different library. Topics can be shared across organizations by an administrator through the global catalog.

What's Next

Now that you have classified your first topic, explore these features to get more from AudienceGPT:

• Classification Deep Dive -- Learn about AI vs. rule-based modes, all 7 layers in detail, and DSP naming formats
• Library Management -- Browse, filter, and take bulk actions on your topics
• Campaign Brief Analysis -- Upload a campaign brief and get AI-recommended audience topics
• Matrix Generation -- Create combinatorial taxonomies (e.g., all US states crossed with age ranges)
• CSV Import -- Bulk import topics from spreadsheets with automated classification