Re-engagement subject line tests in Claude Desktop
Winning back dormant subscribers starts with the subject line. With marginal connected to Claude Desktop, you can generate, launch, and evaluate re-engagement subject lines without leaving the conversation — letting the AI agent handle the email marketing busywork.
Why re-engagement subject lines need their own test
Inactive subscribers behave differently from your active list. They've stopped opening, so the subject line is doing nearly all the work of restarting the relationship. A line that performs well for engaged readers can fall flat with someone who hasn't opened in 90 days.
That's why a dedicated re-engagement subject line test matters. Instead of guessing whether 'We miss you' beats a discount-led hook, you let real opens decide — and marginal gives you the within-test lift numbers to make that call confidently.
- Dormant readers respond to curiosity, value, and loss-aversion framing differently
- Open rates are the clearest early signal of a successful win-back
- Small subject line changes can swing reactivation meaningfully
- A controlled subject line A/B test removes the guesswork
Connecting marginal to Claude Desktop
marginal is a hosted email marketing MCP server, so there's nothing to deploy. You point Claude Desktop at the marginal endpoint, supply your Bearer API key, and the four tools — generate_variants, launch_test, get_results, and recommend_next — become available in chat.
Once connected, you describe your re-engagement goal in plain language and Claude Desktop drives the workflow on your behalf.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key in the marg_live_... format
- Registry identifier: sh.marginal/mcp
- Full setup steps live at https://marginal.sh/docs/
Running the re-engagement test from a single chat
A typical session reads like a conversation. You tell Claude Desktop you want to win back subscribers who haven't opened in three months, and it calls generate_variants to draft several distinct subject line angles. After you pick the candidates, launch_test ships the managed send and splits traffic across variants.
From there, get_results returns open and click tracking with within-test lift, and recommend_next suggests whether to send the leader to the remaining audience. The AI agent email marketing loop stays inside Claude Desktop end to end.
- generate_variants — produce re-engagement subject line options
- launch_test — start the managed subject line A/B test
- get_results — read opens, clicks, and lift metrics
- recommend_next — get a data-backed next step
Getting useful re-engagement results
Re-engagement subject lines in Claude Desktop work best when you give the agent context about why subscribers went quiet and what you're offering to bring them back. The more specific your prompt, the sharper the variants marginal generates.
The free tier covers 100 experiments per month, which is plenty to test several win-back angles before committing budget to a full reactivation campaign.
- State the inactivity window (e.g. no opens in 90 days) so framing fits
- Test contrasting angles — incentive, curiosity, and 'last chance' — not minor word swaps
- Let results accumulate before reading lift to avoid early noise
- Use recommend_next to decide when to roll out the winning subject line
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