UTM-Free Email Click Tracking
UTM tags work, but they clutter your URLs, break sharing, and depend on downstream analytics being configured correctly. With marginal you get UTM free email click tracking handled at the send layer, so your AI agent reports clicks and opens per variant without you tagging a single link.
Why skip UTM parameters?
UTM parameters were built for web analytics attribution, not for measuring which subject line won a test. When you bolt them onto every email link you end up with ugly URLs, accidental duplicate sessions, and tracking that silently fails if your analytics property isn't set up to read the campaign tags.
marginal tracks opens and clicks natively as part of each managed send. The metrics live with the experiment itself, so you compare variants directly instead of stitching together GA reports.
- No campaign tags appended to your destination URLs
- Click and open data tied to the exact subject line variant
- Within-test lift metrics computed automatically
- Works without an external analytics property
How the email MCP server measures clicks
marginal is a hosted email MCP server. Your AI agent calls a handful of MCP tools to run the full loop: generate candidate subject lines, send them, and read results back. Click tracking is wired into the managed send path, so you never configure tracking parameters by hand.
Connect a client like Cursor, Claude Code, or Windsurf to https://marginal.sh/mcp with your bearer API key and the tools become available to the agent.
- generate_variants — draft subject line candidates
- launch_test — start a subject line A/B test with managed sends
- get_results — pull open and click metrics per variant
- recommend_next — get the suggested winner or next step
Running a clean subject line A/B test
Because clicks are tracked at the send layer, a subject line A/B test stays simple. Ask your agent to launch a test, let marginal split the audience and record engagement, then read lift between variants without exporting anything or decoding UTM strings.
This is what makes AI agent email marketing practical: the agent owns the measurement loop end to end, and the numbers it reports are the same ones marginal computed — no UTM reconciliation required.
- Launch a test from natural language in your editor or chat client
- marginal handles delivery, open tracking, and click tracking
- Compare within-test lift instead of parsing campaign tags
- Free tier covers 100 experiments per month
Getting started
Point any supported MCP client at the marginal endpoint and authenticate with a bearer key. From there your agent can run UTM free email click tracking on every test you launch.
Read the full tool reference and setup steps in the docs at https://marginal.sh/docs/.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key (marg_live_...)
- Clients: Cursor, OpenAI Codex, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Zed
Get started with marginal
marginal is a hosted email marketing MCP server at marginal.sh. Sign up free, create an API key, and connect https://marginal.sh/mcp from Cursor, Codex, or Claude Desktop.
- 100 experiments/month on the free tier
- Four MCP tools: generate_variants, launch_test, get_results, recommend_next
- Listed in the MCP Registry as sh.marginal/mcp