Running a Sunset Flow Subject Line Test
A sunset flow is your last, best shot at re-engaging subscribers before you suppress them. The subject line decides whether that email gets opened at all — which makes a sunset flow subject line test one of the highest-leverage experiments you can run. With marginal, your AI agent handles the whole loop end to end.
Why test subject lines in a sunset flow?
Lapsing subscribers have already tuned you out, so the usual open rates don't apply. A subject line that works in a welcome series may fall flat with someone who hasn't clicked in 90 days. Testing tells you which angle — urgency, curiosity, a discount, or a plain "we miss you" — actually pulls them back.
Because re-engagement audiences are smaller and more decayed, a disciplined subject line A/B test protects you from suppressing people you could have saved with better copy.
- Re-engaged subscribers cost far less than acquiring new ones
- Subject line is the single biggest lever on sunset open rates
- Within-test lift metrics show which variant moves the needle
- Losing the test still teaches you what your lapsed audience ignores
How marginal runs the test
marginal is a hosted email MCP server, so your AI agent connects to https://marginal.sh/mcp with a Bearer API key and drives the experiment through four tools. There's nothing to self-host and no separate sending stack to wire up.
A typical sunset flow subject line test in marginal flows through these steps:
- generate_variants — draft several sunset subject lines from one prompt
- launch_test — split your lapsed segment and send the managed test
- get_results — pull open and click tracking with within-test lift
- recommend_next — let marginal suggest the winner or a follow-up variant
Subject line angles worth testing
Ask your agent to generate variants across distinct emotional angles rather than minor wording tweaks. Bigger contrasts produce clearer signal, which matters when your re-engagement segment is small.
- Direct miss: "We haven't seen you in a while"
- Last-chance urgency: "This is the last email you'll get from us"
- Incentive: "Here's 20% to come back"
- Curiosity: "Did we do something wrong?"
- Value reminder: "You're about to lose your saved settings"
Fitting it into AI agent email marketing
Because marginal exposes everything through MCP, AI agent email marketing workflows in Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Zed, or OpenAI Codex can launch and read tests without you leaving the editor. Your agent proposes variants, ships the test, and reports lift in the same conversation.
The free tier covers 100 experiments per month, which is plenty to iterate on a sunset flow until the open rate stops climbing. See the docs at https://marginal.sh/docs/ to connect a client and run your first test.
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