Testing Review Request Subject Lines in Claude Code
Review request emails live or die on the subject line. With marginal connected to Claude Code, you can draft, test, and ship subject line variations for your review request campaigns without leaving your editor.
Why subject lines decide your review request open rates
A review request is a low-stakes ask, but it competes with everything else in the inbox. Whether a customer ever sees your ask comes down to the subject line — and small wording changes can move open rates by double digits.
Instead of guessing which phrasing works, run a subject line A/B test. marginal handles the variant generation, managed sends, and open/click tracking so you can compare review request subject lines on real recipients.
- Ask-style subjects ('How was your order?') vs. incentive-led ('Share your thoughts in 30 seconds')
- Personalization tokens like first name or product name
- Urgency and timing cues that suit post-purchase context
- Short vs. descriptive phrasing for mobile previews
How marginal works inside Claude Code
marginal is a hosted email MCP server. Once Claude Code is connected to the https://marginal.sh/mcp endpoint with your Bearer API key, the agent can call marginal's tools directly during a conversation — no separate dashboard required.
For a review request subject line test, the typical flow uses four tools in sequence so the agent does the iterative work for you.
- generate_variants — produce several review request subject lines to compare
- launch_test — start a subject line A/B test as a managed send
- get_results — pull open and click metrics with within-test lift
- recommend_next — let the agent suggest the next variant to try
A practical review request workflow
Treat Claude Code as your AI agent for email marketing. Describe the audience and the goal, and let the agent orchestrate the test end to end.
A good prompt might be: 'Generate five review request subject lines for customers who received their order last week, launch a subject line A/B test, and report back the winner with lift.' Claude Code maps that intent onto marginal's tools and returns measurable results.
- Seed the test with 4–6 distinct angles, not minor wording tweaks
- Let the open-rate winner advance, then test click-through on the body CTA next
- Use recommend_next to keep iterating once the first round resolves
- Stay on the free tier with up to 100 experiments per month while you dial in your approach
Reading the results
marginal reports open and click tracking alongside within-test lift, so you can see how each subject line performed relative to the others in the same send. That keeps comparisons fair — every variant reached the same audience under the same conditions.
Once a winning review request subject line emerges, you have a defensible default for future campaigns and a documented baseline to beat. See the docs at https://marginal.sh/docs/ for tool details and response shapes.
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