Fundamentals

Core Concepts

Understand the key concepts in AI Data Stream: properties, connections, conversations, annotations, reports, and teams.

This guide explains the core concepts you’ll work with in AI Data Stream.

Properties

A property represents a website or app you want to analyze. It’s the container for all your data connections, conversations, and annotations.

What Properties Do

  • Group related data sources together (e.g., your website’s GA4 + Search Console)
  • Organize conversations and reports by project
  • Control team access at the property level
  • Store annotations specific to that website/app

Creating Properties

  1. Go to your dashboard
  2. Click Create Property (Create First Property if you don’t have one yet)
  3. Enter a Property name (e.g., “Main Website”, “Mobile App”, “Client: Acme Corp”)
  4. Optionally add a Description and choose a Property Color
  5. Click Create Property

Switching Between Properties

Click the property name at the top of the sidebar to open the property list. The current property is ticked.

  • Properties are listed alphabetically.
  • With six or more properties, a Search properties… box appears at the top. Type part of a name or description to filter; press Enter to open the only remaining match.
  • A Recent group above the full list shows the last three properties you worked in, so you can switch back in one click.
  • Manage properties and Create property stay at the bottom of the list.

Switching keeps you in the same section (dashboard, chat, annotations) for the new property.

Connections

A connection links a data source to your property. Each property can have multiple connections to pull data from different platforms.

Available Connections

  • Google Analytics 4 - Website traffic, behavior, conversions
  • Google Search Console - Organic search performance, keywords
  • Google Ads - Ad campaigns, spend, conversions
  • Google Tag Manager - Tags, triggers, container configuration
  • LinkedIn Ads - B2B campaign performance and professional audience demographics
  • Meta Ads - Facebook and Instagram campaign performance, audiences, and creatives
  • Cloudflare - AI crawler activity, AI referrals, and whole-site edge traffic
  • Sitemap - Which URLs appeared in and disappeared from your sitemap, and when
  • PageSpeed Insights - Core Web Vitals, performance scores
  • DataForSEO - Backlinks (referring domains, anchor text, spam scores) AI search visibility (where AI answers cite you) and brand mentions across the web
  • WordPress - Content and posts via REST API
  • GitHub - Repos, commits, deployments, issues, PRs
  • HubSpot - CRM deals, contacts, companies
  • Pipedrive - Sales pipeline, activities

Some connections are only available on certain plans. See Plans & Billing for details.

Connection Features

  • Read-only access: We only read your data. The one exception is optional: annotations you choose to sync to GA4
  • Real-time queries: Data is fetched live, not stored
  • OAuth authentication: Secure, revocable access
  • Active/inactive toggle: Pause connections without deleting

See Connecting Data Sources for setup details.

Conversations

A conversation is an AI chat session tied to a property. You ask questions in natural language, and the AI analyzes your connected data sources to provide insights.

Starting a Conversation

  1. Open Chat in the sidebar and click New Chat
  2. Select which data sources to include
  3. Choose your AI model
  4. Type your question and press Send

Conversation Features

  • Context memory: The AI remembers previous messages in the conversation
  • Multi-source queries: Ask about GA4 and Search Console data together
  • Annotation awareness: The AI checks annotations when analyzing trends
  • Export options: Copy or download conversation history

Tips for Better Conversations

  • Be specific about dates: “last 7 days” vs “November 2024”
  • Mention the metrics you care about: “traffic”, “conversions”, “bounce rate”
  • Ask follow-up questions to dig deeper
  • Use prompt presets for consistent analysis styles

See Using the AI Chat for more tips.

Annotations

Annotations are notes attached to dates in your property’s timeline. They help the AI understand why metrics changed.

Why Annotations Matter

When you ask “Why did traffic drop last week?”, the AI checks your annotations. If you noted “Site was down for maintenance on March 5”, the AI can factor that into its analysis instead of speculating.

Annotation Types

Manual annotations you create:

  • Marketing campaigns launched
  • Website redesigns
  • Product launches
  • Known issues or outages

Global annotations from feeds:

  • Google algorithm updates (Core Updates, Spam Updates, etc.)
  • Google Search status incidents
  • Admin-created global events

Creating Annotations

  1. Go to your property’s Annotations tab
  2. Click Add Annotation
  3. Choose a Category (required)
  4. Choose Single Date or Range and set the date
  5. Enter a Title and, optionally, a Description
  6. Click Create

Annotation Categories

Every team starts with five categories: Marketing Campaign, Site Change, Content Update, Product Launch and Bug Fix. Only the team owner can add, edit or delete categories.

Global annotations use their own built-in categories:

  • Algorithm Update - Search algorithm changes
  • System Outage - Service disruptions

See Annotations for the full guide.

Reports

Skills are reusable analysis recipes with expert methodology built in. Running a skill produces a report - consistent, formatted output you can share or schedule.

The Skills Catalog

AI Data Stream includes more than 30 ready-made skills organized by category:

  • Audits - in-depth teardowns like the Google Ads Account Audit, SEO Audit, and Core Web Vitals Audit
  • Monitors - rhythm checks like the Weekly Traffic Monitor and Quick Health Check, designed to run on a schedule
  • Analysis - focused deep dives like Custom Date Comparison and Traffic Source Deep Dive
  • CRM - pipeline and customer analyses for HubSpot or Pipedrive
  • Ecommerce - revenue, funnel, and promotion analyses for sites with ecommerce tracking

Running a Report

  1. Click Skills in the sidebar
  2. Pick a skill (or browse the skills catalog to add more)
  3. Fill in any required fields
  4. Click Run Report

Creating Custom Skills

You can create your own skills:

  1. On the Skills page, click Add skill → Blank Skill (then write it yourself or use Generate with AI)
  2. Define input fields (optional)
  3. Write the prompt
  4. Save it to your property

See Reports for detailed information.

Teams

A team is a group of users who share access to properties and data. Every account starts with a personal team.

Team Roles

  • Owner: The account creator, with full control of the team, billing and all properties
  • Member: Can access the properties they’re assigned, at the access level set for each property

Team size depends on your plan. See Plans & Billing for details.

Property Permissions

On higher-tier plans, you can set per-property permissions:

  • Read: View data, conversations, annotations
  • Write: Create annotations, run reports, start conversations
  • Admin: Manage connections, edit property settings

See Team Management for setup instructions.

How Concepts Connect

Team (your account)
└── AI Providers (your API keys)
└── Properties (websites/apps)
    ├── Connections (GA4, Search Console, etc.)
    ├── Conversations (AI chat sessions)
    ├── Annotations (timeline notes)
    └── Skills (saved analyses)

Example Setup

A marketing agency might organize like this:

Team: Marketing Agency Pro
├── AI Providers: Anthropic, Google
├── Property: Client A - Main Site
│   ├── GA4 + Search Console + Google Ads
│   └── Annotations: Campaign launches, site updates
├── Property: Client A - Blog
│   ├── GA4 + Search Console
│   └── Annotations: Content calendar
└── Property: Client B - Ecommerce
    ├── GA4 + Search Console + Google Ads
    └── Reports: Weekly Ecommerce Review

Next Steps