Overview

Checkpoint provides comprehensive analytics for understanding the AI agent and bot traffic hitting your applications. View real-time detections, analyze trends, and export data for reporting.

Activity Feed (Legacy Tab: Monitor)

The real-time, per-request feed is the Activity screen in the current dashboard (left rail, under See — see Dashboard Overview). Monitor is the legacy name for this feed, from when it was a tab nested inside a single Analytics page rather than its own rail item. The fields and filters below apply either way.

Each entry shows:

Field	Description
Timestamp	When the detection occurred
Detection Class	`human`, `ai_agent`, `bot`, or `incomplete_data`
Confidence	Score from 0–100
Agent Name	Identified agent (e.g., ChatGPT, Googlebot)
IP Address	Source IP of the request
User Agent	The request's user agent string
Path	URL path that was accessed
Session	Consolidated session identifier

Filtering

Filter the Activity feed by:

Detection class (AI agent, bot, human)
Confidence range
Time period
Agent name
IP address

Session Grouping

Detections are grouped into sessions. Checkpoint consolidates requests from the same visitor into a single session, even if the client-side session ID resets (which happens every 30 minutes for privacy). This gives you an accurate count of unique visitors rather than inflated session numbers.

Session consolidation uses multiple signals (fingerprint, IP address, user agent, agent type) to group related requests within 30-minute windows.

Analytics (Legacy Tab: Analyze)

The Analytics screen (legacy tab name: Analyze) provides aggregated analytics across your detection data.

Key Metrics

Metric	Description
Total Sessions	Unique visitor sessions in the selected period
Detection Rate	Percentage of sessions classified as agents or bots
AI Agent Sessions	Sessions classified as AI agents
Bot Sessions	Sessions classified as traditional bots
Human Sessions	Sessions classified as human visitors

Agent Type Breakdown

A visual breakdown of traffic by detection class:

AI Agents — ChatGPT, Claude, Perplexity, Gemini, and other AI assistants
Bots — Googlebot, Bingbot, scrapers, and automated tools
Humans — Regular browser traffic
Incomplete Data — Requests with insufficient signals

Traffic Trends

Historical charts showing detection patterns over time:

Detection volume by day/week/month
Classification distribution over time
Confidence score distribution
New vs returning agents

Top Agents

Ranked list of the most frequently detected agents, showing:

Agent name
Detection count
Average confidence score
First and last seen dates

Date Range Selection

All analytics views support custom date ranges:

Last 24 hours — Recent activity
Last 7 days — Weekly overview
Last 30 days — Monthly trends
Custom range — Pick specific start and end dates

Cross-Project Analytics

Organization-level analytics (available at the organization Analytics page, legacy name: Analyze) aggregate data across all projects, giving you a bird's-eye view of agent activity across your properties.

Understanding Detection Data

Detection Rate

The detection rate is the percentage of sessions classified as non-human:

Detection Rate = (AI Agent Sessions + Bot Sessions) / Total Sessions × 100

A typical detection rate varies by industry:

Industry	Typical Detection Rate
E-commerce	30–50%
Content/Media	40–60%
SaaS	20–40%
API	50–70%

Confidence Distribution

The Confidence Distribution chart groups sessions into five display buckets — 0–20%, 20–40%, 40–60%, 60–80%, and 80–100% — shaded by three severity tiers (low, medium, high). This is a UI grouping for the histogram only; it does not match the canonical confidence bands. For the four-band model used everywhere else (91–100 very high, 71–90 high, 31–70 medium, 0–30 low, and what a score of exactly 0 means), see Confidence Scores.

Use this distribution to choose your enforcement threshold. Start by logging everything, then set your block threshold once you can see where real traffic falls.

Policy Tuning Workflow

Start in detect mode (no enforcement)
Review Activity for false positives
Analyze confidence score distribution
Set enforcement threshold (typically 71+, the high band)
Enable enforcement with tuned policy
Monitor continuously for changes

Next Steps

Dashboard Overview — Navigate the full dashboard
Managing Projects — Project setup and configuration
Policies — Configure enforcement based on analytics

Analytics & Reporting