Email A/B Testing Without Customer.io
You don't need to wire up the Customer.io API to start experimenting on subject lines and send copy. marginal is a hosted email MCP server that handles variant generation, sends, and lift metrics — so your AI agent can run tests without touching your ESP setup.
Why skip the Customer.io integration for experiments
Customer.io is a strong lifecycle messaging platform, but standing up its API, segments, and broadcast templates just to test a few subject lines is a lot of overhead. If your goal is fast email A/B testing without Customer.io in the loop, marginal lets you launch experiments directly from your editor or agent.
This is useful when you want a clean separation between your production messaging stack and your experimentation layer. You keep Customer.io for what it does well, and run lightweight ESP-free email experiments on the side to learn what copy actually moves opens and clicks.
- No Customer.io workspace, API key, or broadcast config required to test
- Managed sends with built-in open and click tracking
- Within-test lift metrics so you can compare variants directly
- A free tier of 100 experiments per month to get started
How marginal works as a Customer.io alternative for testing
marginal is not a full ESP replacement — it's a focused experimentation server. As a Customer.io alternative for the specific job of subject line A/B testing, it exposes four MCP tools your agent calls in sequence.
Because it's hosted at https://marginal.sh/mcp, there's nothing to deploy or self-host. You authenticate with a Bearer API key and start launching tests.
- generate_variants — draft multiple subject line or copy options
- launch_test — send the experiment and split traffic across variants
- get_results — pull open/click data and lift per variant
- recommend_next — get a suggested follow-up based on results
Running a subject line A/B test from your AI agent
AI agent email marketing works best when the agent can act end to end. With marginal connected as an email MCP server, an agent in Cursor, Claude Code, Windsurf, Cline, Continue, Zed, OpenAI Codex, or Claude Desktop can generate candidate subject lines, launch a subject line A/B test, and then read back the results to decide what wins.
The flow is conversational: ask your agent to test three subject lines for a re-engagement email, let it call generate_variants and launch_test, then have it summarize get_results once data comes in. No Customer.io broadcast or template editing is needed at any step.
When to keep Customer.io alongside marginal
Running experiments without Customer.io doesn't mean abandoning it. Many teams use marginal to discover winning subject lines and copy, then promote those winners into their Customer.io lifecycle campaigns.
Treat marginal as your rapid testing surface and Customer.io as your orchestration layer. You get the speed of ESP-free email experiments plus the deliverability and automation of your existing stack — without merging the two integrations together.
- Use marginal to validate subject lines before they ship
- Move proven winners into Customer.io broadcasts and journeys
- Read the docs at https://marginal.sh/docs/ for tool details
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