What is headless analytics?

Your coding agent already has access to your database, your Stripe account, your GitHub repo, and your codebase. Headless analytics adds your web traffic to that picture. Instead of checking a dashboard in one tab while your agent works in another, the agent queries your analytics directly — and can cross-reference it with everything else it already knows.

"Show me the pages with the highest bounce rate, cross-reference with this week's deploys, and check if any of them correlate with a Stripe churn spike." That query spans analytics, git history, and payment data. No dashboard can do that. An agent with access to all three does it in one turn.

Headless analytics is web analytics without a dashboard. The tracking works the same — a script on your site collects pageviews, sessions, referrers, and performance data. The difference is what happens next. Instead of rendering charts in a browser, the data is served through an API that AI agents query directly alongside every other data source they're connected to. A single analytics snapshot returns as roughly 55 tokens of structured JSON — compared to thousands of tokens for a dashboard screenshot — leaving the agent's context window free for the data that matters.

How is this different from regular analytics?

Both track the same metrics. The difference is where the data goes and who reads it.

Why should an agent query analytics?

Your agent can correlate data across sources. A dashboard shows you analytics in isolation. An agent connected to your analytics, your database, and your payment provider can answer questions that span all three: "Which traffic sources bring users who actually convert to paid?" or "Did last Tuesday's deploy affect bounce rates on pages that drive revenue?" These aren't analytics questions or database questions — they're business questions, and an agent with access to both can answer them without you manually joining data across tabs.

Browser and server-side tracking in one place. Most products have two kinds of user activity: what happens in the browser (pageviews, clicks, conversions) and what happens on the server (API calls, webhook deliveries, background jobs). Headless analytics tracks both through the same system. Your agent queries browser traffic and server events from one tool, not two dashboards.

The agent can act on what it finds. A coding agent that sees a 73% bounce rate on /checkout can inspect the page, identify the problem, and push a fix in the same session. There's no handoff between the person reading data and the person writing code — they're the same agent.

Agents don't forget to check. A scheduled agent can pull a daily snapshot at 07:00, compare it to yesterday, flag anything unusual, and have a summary ready before you start work. Traditional analytics requires someone to remember to look at a dashboard. Most people don't.

How do you set it up?

With Lodd, a coding agent handles the full setup from a single prompt. Tell Claude Code, Cursor, or Codex to "add lodd.dev analytics to this project" and it will:

1. Write the MCP server config
2. Authenticate via email and a 6-digit code
3. Create a site and get the tracking script
4. Embed the script in your HTML

After that, the full analytics toolkit is available. Ask your agent about visitors, top pages, traffic sources, conversion funnels, load times, custom events, or anything else. It queries the data directly and responds. The full tool reference covers every available query.

For conversation agents like Claude Desktop or claude.ai, setup is even simpler: add the MCP connector URL and OAuth handles the rest.

Who is this for?

Developers who work through AI coding agents. If Claude Code, Cursor, or Codex is your primary interface, headless analytics keeps your data where you already work instead of behind another tab.

Solo founders running lean. You don't have time to check a dashboard every day, and you probably don't. An agent that monitors your analytics and tells you when something changes is more useful than a dashboard you'll forget to open.

Privacy-conscious sites. Lodd uses no cookies, stores country-level geo only, and hashes all IPs with a daily-rotating salt deleted after 24 hours. GDPR-compliant without consent banners. More detail in the privacy policy.

How does it handle privacy?

Traditional analytics tools often need cookies for session tracking, which means consent banners under GDPR. Headless analytics avoids this by design.

Lodd's approach: no cookies, no fingerprinting, country-only geolocation. IPs are hashed with a daily-rotating salt that is deleted after 24 hours. Bot traffic is tagged and excluded automatically.

For per-user analytics, Lodd uses actor hashes. You hash a user identifier before sending it. Lodd never sees the original and cannot bridge actors to CRM records or external databases. This is a deliberate privacy design choice, not a gap.

What tools are available?

Lodd provides MCP tools across several categories:

• Analytics: snapshots, time series, funnels, realtime visitor counts
• Breakdowns: pages, traffic sources, countries, browsers, devices, entry/exit pages, bot reports
• Performance: page load times (avg, median, p95) by page, device, country, or browser
• Conversion attribution: which pages and traffic sources drive specific events
• Custom events: counts, individual records, time series per event
• Actor analytics: active users, per-user timelines, cohort retention
• Trackable links: short URLs with source attribution and click tracking
• Annotations: deploy notes and change records, embedded automatically in timeseries data

All tools accept filters (country, browser, OS, device type, UTM source, referrer) and period ranges from "today" to custom spans up to 365 days. Free up to 2,500 events/month.