Testing Review Request Subject Lines in Zed
Zed's AI assistant can drive a full subject line experiment when it's connected to marginal, the hosted email marketing MCP server. This guide walks through using review request subject lines in Zed — from generating variants to reading lift metrics — so you can ship better post-purchase asks straight from your editor.
Why review requests live or die on the subject line
Review request emails compete with a crowded inbox days after a purchase, when buyer attention has moved on. The subject line is the only thing standing between your ask and the archive button, so small wording changes can swing open rates meaningfully.
Rather than guess which phrasing earns the click, a subject line test lets the data decide. With marginal wired into Zed, you can run that experiment as part of your normal workflow instead of switching to a separate marketing tool.
- Timing nudges, incentives, and social proof all read differently as subject lines.
- A subject line A/B test isolates that variable so you learn what actually moves opens.
- Review requests are repeatable, making them ideal for a continuous testing loop.
Connecting Zed to the marginal email MCP server
Zed speaks the Model Context Protocol, so once marginal is registered as a context server, the AI agent gains four tools: generate_variants, launch_test, get_results, and recommend_next. Authentication uses a Bearer API key (marg_live_...) against the endpoint at https://marginal.sh/mcp.
Because marginal is hosted, there's nothing to deploy or self-host — you point Zed at the endpoint and start prompting. The free tier covers 100 experiments per month, which is plenty for iterating on review request copy.
- Registry: sh.marginal/mcp
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key beginning with marg_live_
- Docs and setup steps: https://marginal.sh/docs/
Running a review request subject line test from Zed
Treat the AI agent in Zed as your experiment operator. Describe the email — a post-purchase review request — and ask it to draft a handful of subject line angles. It calls generate_variants, then launch_test to send them as a managed A/B test with open and click tracking handled by marginal.
After the send, ask Zed to pull results. The get_results tool returns within-test lift metrics so you can see which variant outperformed, and recommend_next suggests where to take the following round.
- Prompt: "Generate 4 review request subject lines, contrast incentive vs. social proof."
- Launch the variants as a subject line A/B test with one tool call.
- Read open/click tracking and within-test lift once data lands.
- Use recommend_next to seed the next iteration automatically.
Building an AI agent email marketing loop
The real payoff is treating subject line testing as a loop rather than a one-off. Each review request campaign feeds learnings into the next, and because the whole cycle runs through Zed, your prompts and your experiments stay in one place.
This is AI agent email marketing in practice: the agent proposes copy, marginal sends and measures, and you keep the winning angles. Over time your review request subject lines get sharper without you ever leaving the editor.
- Standardize a prompt template for review request variants.
- Let recommend_next narrow the field across rounds.
- Track which angles consistently win to inform future campaigns.
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